2020-11-02 20:50:16 +08:00
|
|
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
|
|
|
; RUN: llc -march=amdgcn -mcpu=bonaire -verify-machineinstrs -mattr=+load-store-opt < %s | FileCheck -enable-var-scope -check-prefixes=GCN,CI %s
|
|
|
|
|
; RUN: llc -march=amdgcn -mcpu=gfx900 -verify-machineinstrs -mattr=+load-store-opt,-unaligned-access-mode < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GFX9,GFX9-ALIGNED %s
|
|
|
|
|
; RUN: llc -march=amdgcn -mcpu=gfx900 -verify-machineinstrs -mattr=+load-store-opt,+unaligned-access-mode < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GFX9,GFX9-UNALIGNED %s
|
2014-10-11 06:01:59 +08:00
|
|
|
|
|
|
|
|
; FIXME: We don't get cases where the address was an SGPR because we
|
|
|
|
|
; get a copy to the address register for each one.
|
|
|
|
|
|
2014-11-14 03:56:13 +08:00
|
|
|
@lds = addrspace(3) global [512 x float] undef, align 4
|
2015-07-15 01:57:36 +08:00
|
|
|
@lds.f64 = addrspace(3) global [512 x double] undef, align 8
|
2014-10-11 06:01:59 +08:00
|
|
|
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @simple_read2_f32(float addrspace(1)* %out) #0 {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: simple_read2_f32:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v1, vcc, lds@abs32@lo, v0
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[1:2], v1 offset1:8
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_mov_b32 s3, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s2, 0
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v2, v1, v2
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v1, 0
|
|
|
|
|
; CI-NEXT: buffer_store_dword v2, v[0:1], s[0:3], 0 addr64
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: simple_read2_f32:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v2, 2, v0
|
|
|
|
|
; GFX9-NEXT: v_add_u32_e32 v0, lds@abs32@lo, v2
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[0:1], v0 offset1:8
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v0, v0, v1
|
|
|
|
|
; GFX9-NEXT: global_store_dword v2, v0, s[0:1]
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2016-02-11 14:02:01 +08:00
|
|
|
%x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx0 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %x.i
|
2015-02-28 05:17:42 +08:00
|
|
|
%val0 = load float, float addrspace(3)* %arrayidx0, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%add.x = add nsw i32 %x.i, 8
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx1 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %add.x
|
2015-02-28 05:17:42 +08:00
|
|
|
%val1 = load float, float addrspace(3)* %arrayidx1, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%sum = fadd float %val0, %val1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%out.gep = getelementptr inbounds float, float addrspace(1)* %out, i32 %x.i
|
2014-10-11 06:01:59 +08:00
|
|
|
store float %sum, float addrspace(1)* %out.gep, align 4
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @simple_read2_f32_max_offset(float addrspace(1)* %out) #0 {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: simple_read2_f32_max_offset:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v1, vcc, lds@abs32@lo, v0
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[1:2], v1 offset1:255
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_mov_b32 s3, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s2, 0
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v2, v1, v2
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v1, 0
|
|
|
|
|
; CI-NEXT: buffer_store_dword v2, v[0:1], s[0:3], 0 addr64
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: simple_read2_f32_max_offset:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v2, 2, v0
|
|
|
|
|
; GFX9-NEXT: v_add_u32_e32 v0, lds@abs32@lo, v2
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[0:1], v0 offset1:255
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v0, v0, v1
|
|
|
|
|
; GFX9-NEXT: global_store_dword v2, v0, s[0:1]
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2016-02-11 14:02:01 +08:00
|
|
|
%x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx0 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %x.i
|
2015-02-28 05:17:42 +08:00
|
|
|
%val0 = load float, float addrspace(3)* %arrayidx0, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%add.x = add nsw i32 %x.i, 255
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx1 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %add.x
|
2015-02-28 05:17:42 +08:00
|
|
|
%val1 = load float, float addrspace(3)* %arrayidx1, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%sum = fadd float %val0, %val1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%out.gep = getelementptr inbounds float, float addrspace(1)* %out, i32 %x.i
|
2014-10-11 06:01:59 +08:00
|
|
|
store float %sum, float addrspace(1)* %out.gep, align 4
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @simple_read2_f32_too_far(float addrspace(1)* %out) #0 {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: simple_read2_f32_too_far:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v1, vcc, lds@abs32@lo, v0
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: ds_read_b32 v2, v1
|
|
|
|
|
; CI-NEXT: ds_read_b32 v1, v1 offset:1028
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_mov_b32 s3, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s2, 0
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v2, v2, v1
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v1, 0
|
|
|
|
|
; CI-NEXT: buffer_store_dword v2, v[0:1], s[0:3], 0 addr64
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: simple_read2_f32_too_far:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; GFX9-NEXT: v_add_u32_e32 v1, lds@abs32@lo, v0
|
|
|
|
|
; GFX9-NEXT: ds_read_b32 v2, v1
|
|
|
|
|
; GFX9-NEXT: ds_read_b32 v1, v1 offset:1028
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v1, v2, v1
|
|
|
|
|
; GFX9-NEXT: global_store_dword v0, v1, s[0:1]
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2016-02-11 14:02:01 +08:00
|
|
|
%x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx0 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %x.i
|
2015-02-28 05:17:42 +08:00
|
|
|
%val0 = load float, float addrspace(3)* %arrayidx0, align 4
|
2014-10-11 06:12:32 +08:00
|
|
|
%add.x = add nsw i32 %x.i, 257
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx1 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %add.x
|
2015-02-28 05:17:42 +08:00
|
|
|
%val1 = load float, float addrspace(3)* %arrayidx1, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%sum = fadd float %val0, %val1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%out.gep = getelementptr inbounds float, float addrspace(1)* %out, i32 %x.i
|
2014-10-11 06:01:59 +08:00
|
|
|
store float %sum, float addrspace(1)* %out.gep, align 4
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @simple_read2_f32_x2(float addrspace(1)* %out) #0 {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: simple_read2_f32_x2:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v3, vcc, lds@abs32@lo, v0
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[1:2], v3 offset1:8
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_mov_b32 s3, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s2, 0
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v4, v1, v2
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[1:2], v3 offset0:11 offset1:27
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v1, v1, v2
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v2, v4, v1
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v1, 0
|
|
|
|
|
; CI-NEXT: buffer_store_dword v2, v[0:1], s[0:3], 0 addr64
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: simple_read2_f32_x2:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v4, 2, v0
|
|
|
|
|
; GFX9-NEXT: v_add_u32_e32 v2, lds@abs32@lo, v4
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[0:1], v2 offset1:8
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[2:3], v2 offset0:11 offset1:27
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v0, v0, v1
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v1, v2, v3
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v0, v0, v1
|
|
|
|
|
; GFX9-NEXT: global_store_dword v4, v0, s[0:1]
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2016-02-11 14:02:01 +08:00
|
|
|
%tid.x = tail call i32 @llvm.amdgcn.workitem.id.x() #1
|
2014-10-11 06:01:59 +08:00
|
|
|
%idx.0 = add nsw i32 %tid.x, 0
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx0 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %idx.0
|
2015-02-28 05:17:42 +08:00
|
|
|
%val0 = load float, float addrspace(3)* %arrayidx0, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
|
|
|
|
|
%idx.1 = add nsw i32 %tid.x, 8
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx1 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %idx.1
|
2015-02-28 05:17:42 +08:00
|
|
|
%val1 = load float, float addrspace(3)* %arrayidx1, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%sum.0 = fadd float %val0, %val1
|
|
|
|
|
|
|
|
|
|
%idx.2 = add nsw i32 %tid.x, 11
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx2 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %idx.2
|
2015-02-28 05:17:42 +08:00
|
|
|
%val2 = load float, float addrspace(3)* %arrayidx2, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
|
|
|
|
|
%idx.3 = add nsw i32 %tid.x, 27
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx3 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %idx.3
|
2015-02-28 05:17:42 +08:00
|
|
|
%val3 = load float, float addrspace(3)* %arrayidx3, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%sum.1 = fadd float %val2, %val3
|
|
|
|
|
|
|
|
|
|
%sum = fadd float %sum.0, %sum.1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%out.gep = getelementptr inbounds float, float addrspace(1)* %out, i32 %idx.0
|
2014-10-11 06:01:59 +08:00
|
|
|
store float %sum, float addrspace(1)* %out.gep, align 4
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Make sure there is an instruction between the two sets of reads.
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @simple_read2_f32_x2_barrier(float addrspace(1)* %out) #0 {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: simple_read2_f32_x2_barrier:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v3, vcc, lds@abs32@lo, v0
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[1:2], v3 offset1:8
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: s_barrier
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_mov_b32 s3, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s2, 0
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v4, v1, v2
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[1:2], v3 offset0:11 offset1:27
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v1, v1, v2
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v2, v4, v1
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v1, 0
|
|
|
|
|
; CI-NEXT: buffer_store_dword v2, v[0:1], s[0:3], 0 addr64
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: simple_read2_f32_x2_barrier:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v4, 2, v0
|
|
|
|
|
; GFX9-NEXT: v_add_u32_e32 v2, lds@abs32@lo, v4
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[0:1], v2 offset1:8
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: s_barrier
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[2:3], v2 offset0:11 offset1:27
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v0, v0, v1
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v1, v2, v3
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v0, v0, v1
|
|
|
|
|
; GFX9-NEXT: global_store_dword v4, v0, s[0:1]
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2016-02-11 14:02:01 +08:00
|
|
|
%tid.x = tail call i32 @llvm.amdgcn.workitem.id.x() #1
|
2014-10-11 06:01:59 +08:00
|
|
|
%idx.0 = add nsw i32 %tid.x, 0
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx0 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %idx.0
|
2015-02-28 05:17:42 +08:00
|
|
|
%val0 = load float, float addrspace(3)* %arrayidx0, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
|
|
|
|
|
%idx.1 = add nsw i32 %tid.x, 8
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx1 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %idx.1
|
2015-02-28 05:17:42 +08:00
|
|
|
%val1 = load float, float addrspace(3)* %arrayidx1, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%sum.0 = fadd float %val0, %val1
|
|
|
|
|
|
2016-02-11 14:02:01 +08:00
|
|
|
call void @llvm.amdgcn.s.barrier() #2
|
2014-10-11 06:01:59 +08:00
|
|
|
|
|
|
|
|
%idx.2 = add nsw i32 %tid.x, 11
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx2 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %idx.2
|
2015-02-28 05:17:42 +08:00
|
|
|
%val2 = load float, float addrspace(3)* %arrayidx2, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
|
|
|
|
|
%idx.3 = add nsw i32 %tid.x, 27
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx3 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %idx.3
|
2015-02-28 05:17:42 +08:00
|
|
|
%val3 = load float, float addrspace(3)* %arrayidx3, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%sum.1 = fadd float %val2, %val3
|
|
|
|
|
|
|
|
|
|
%sum = fadd float %sum.0, %sum.1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%out.gep = getelementptr inbounds float, float addrspace(1)* %out, i32 %idx.0
|
2014-10-11 06:01:59 +08:00
|
|
|
store float %sum, float addrspace(1)* %out.gep, align 4
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; For some reason adding something to the base address for the first
|
|
|
|
|
; element results in only folding the inner pair.
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @simple_read2_f32_x2_nonzero_base(float addrspace(1)* %out) #0 {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: simple_read2_f32_x2_nonzero_base:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v3, vcc, lds@abs32@lo, v0
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[1:2], v3 offset0:2 offset1:8
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_mov_b32 s3, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s2, 0
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v4, v1, v2
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[1:2], v3 offset0:11 offset1:27
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v1, v1, v2
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v2, v4, v1
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v1, 0
|
|
|
|
|
; CI-NEXT: buffer_store_dword v2, v[0:1], s[0:3], 0 addr64 offset:8
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: simple_read2_f32_x2_nonzero_base:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v4, 2, v0
|
|
|
|
|
; GFX9-NEXT: v_add_u32_e32 v2, lds@abs32@lo, v4
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[0:1], v2 offset0:2 offset1:8
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[2:3], v2 offset0:11 offset1:27
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v0, v0, v1
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v1, v2, v3
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v0, v0, v1
|
|
|
|
|
; GFX9-NEXT: global_store_dword v4, v0, s[0:1] offset:8
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2016-02-11 14:02:01 +08:00
|
|
|
%tid.x = tail call i32 @llvm.amdgcn.workitem.id.x() #1
|
2014-10-11 06:01:59 +08:00
|
|
|
%idx.0 = add nsw i32 %tid.x, 2
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx0 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %idx.0
|
2015-02-28 05:17:42 +08:00
|
|
|
%val0 = load float, float addrspace(3)* %arrayidx0, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
|
|
|
|
|
%idx.1 = add nsw i32 %tid.x, 8
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx1 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %idx.1
|
2015-02-28 05:17:42 +08:00
|
|
|
%val1 = load float, float addrspace(3)* %arrayidx1, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%sum.0 = fadd float %val0, %val1
|
|
|
|
|
|
|
|
|
|
%idx.2 = add nsw i32 %tid.x, 11
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx2 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %idx.2
|
2015-02-28 05:17:42 +08:00
|
|
|
%val2 = load float, float addrspace(3)* %arrayidx2, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
|
|
|
|
|
%idx.3 = add nsw i32 %tid.x, 27
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx3 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %idx.3
|
2015-02-28 05:17:42 +08:00
|
|
|
%val3 = load float, float addrspace(3)* %arrayidx3, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%sum.1 = fadd float %val2, %val3
|
|
|
|
|
|
|
|
|
|
%sum = fadd float %sum.0, %sum.1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%out.gep = getelementptr inbounds float, float addrspace(1)* %out, i32 %idx.0
|
2014-10-11 06:01:59 +08:00
|
|
|
store float %sum, float addrspace(1)* %out.gep, align 4
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Be careful of vectors of pointers. We don't know if the 2 pointers
|
|
|
|
|
; in the vectors are really the same base, so this is not safe to
|
|
|
|
|
; merge.
|
|
|
|
|
; Base pointers come from different subregister of same super
|
|
|
|
|
; register. We can't safely merge this.
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @read2_ptr_is_subreg_arg_f32(float addrspace(1)* %out, <2 x float addrspace(3)*> %lds.ptr) #0 {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: read2_ptr_is_subreg_arg_f32:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0xb
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: s_mov_b32 s7, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s6, 0
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v1, s0
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v2, s1
|
|
|
|
|
; CI-NEXT: ds_read_b32 v1, v1 offset:32
|
|
|
|
|
; CI-NEXT: ds_read_b32 v2, v2
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v2, v1, v2
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v1, 0
|
|
|
|
|
; CI-NEXT: buffer_store_dword v2, v[0:1], s[4:7], 0 addr64
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: read2_ptr_is_subreg_arg_f32:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x2c
|
|
|
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v1, s0
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v2, s1
|
|
|
|
|
; GFX9-NEXT: ds_read_b32 v1, v1 offset:32
|
|
|
|
|
; GFX9-NEXT: ds_read_b32 v2, v2
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v1, v1, v2
|
|
|
|
|
; GFX9-NEXT: global_store_dword v0, v1, s[2:3]
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2016-02-11 14:02:01 +08:00
|
|
|
%x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
|
2014-10-11 06:01:59 +08:00
|
|
|
%index.0 = insertelement <2 x i32> undef, i32 %x.i, i32 0
|
|
|
|
|
%index.1 = insertelement <2 x i32> %index.0, i32 8, i32 0
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%gep = getelementptr inbounds float, <2 x float addrspace(3)*> %lds.ptr, <2 x i32> %index.1
|
2014-10-11 06:01:59 +08:00
|
|
|
%gep.0 = extractelement <2 x float addrspace(3)*> %gep, i32 0
|
|
|
|
|
%gep.1 = extractelement <2 x float addrspace(3)*> %gep, i32 1
|
2015-02-28 05:17:42 +08:00
|
|
|
%val0 = load float, float addrspace(3)* %gep.0, align 4
|
|
|
|
|
%val1 = load float, float addrspace(3)* %gep.1, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%add.x = add nsw i32 %x.i, 8
|
|
|
|
|
%sum = fadd float %val0, %val1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%out.gep = getelementptr inbounds float, float addrspace(1)* %out, i32 %x.i
|
2014-10-11 06:01:59 +08:00
|
|
|
store float %sum, float addrspace(1)* %out.gep, align 4
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Apply a constant scalar offset after the pointer vector extract. We
|
|
|
|
|
; are rejecting merges that have the same, constant 0 offset, so make
|
|
|
|
|
; sure we are really rejecting it because of the different
|
|
|
|
|
; subregisters.
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @read2_ptr_is_subreg_arg_offset_f32(float addrspace(1)* %out, <2 x float addrspace(3)*> %lds.ptr) #0 {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: read2_ptr_is_subreg_arg_offset_f32:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0xb
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: s_mov_b32 s7, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s6, 0
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v1, s0
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v2, s1
|
|
|
|
|
; CI-NEXT: ds_read_b32 v1, v1 offset:32
|
|
|
|
|
; CI-NEXT: ds_read_b32 v2, v2 offset:32
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v2, v1, v2
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v1, 0
|
|
|
|
|
; CI-NEXT: buffer_store_dword v2, v[0:1], s[4:7], 0 addr64
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: read2_ptr_is_subreg_arg_offset_f32:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x2c
|
|
|
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v1, s0
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v2, s1
|
|
|
|
|
; GFX9-NEXT: ds_read_b32 v1, v1 offset:32
|
|
|
|
|
; GFX9-NEXT: ds_read_b32 v2, v2 offset:32
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v1, v1, v2
|
|
|
|
|
; GFX9-NEXT: global_store_dword v0, v1, s[2:3]
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2016-02-11 14:02:01 +08:00
|
|
|
%x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
|
2014-10-11 06:01:59 +08:00
|
|
|
%index.0 = insertelement <2 x i32> undef, i32 %x.i, i32 0
|
|
|
|
|
%index.1 = insertelement <2 x i32> %index.0, i32 8, i32 0
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%gep = getelementptr inbounds float, <2 x float addrspace(3)*> %lds.ptr, <2 x i32> %index.1
|
2014-10-11 06:01:59 +08:00
|
|
|
%gep.0 = extractelement <2 x float addrspace(3)*> %gep, i32 0
|
|
|
|
|
%gep.1 = extractelement <2 x float addrspace(3)*> %gep, i32 1
|
|
|
|
|
|
|
|
|
|
; Apply an additional offset after the vector that will be more obviously folded.
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%gep.1.offset = getelementptr float, float addrspace(3)* %gep.1, i32 8
|
2014-10-11 06:01:59 +08:00
|
|
|
|
2015-02-28 05:17:42 +08:00
|
|
|
%val0 = load float, float addrspace(3)* %gep.0, align 4
|
|
|
|
|
%val1 = load float, float addrspace(3)* %gep.1.offset, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%add.x = add nsw i32 %x.i, 8
|
|
|
|
|
%sum = fadd float %val0, %val1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%out.gep = getelementptr inbounds float, float addrspace(1)* %out, i32 %x.i
|
2014-10-11 06:01:59 +08:00
|
|
|
store float %sum, float addrspace(1)* %out.gep, align 4
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @read2_ptr_is_subreg_f32(float addrspace(1)* %out) #0 {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: read2_ptr_is_subreg_f32:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v1, vcc, lds@abs32@lo, v0
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[1:2], v1 offset1:8
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_mov_b32 s3, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s2, 0
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v2, v1, v2
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v1, 0
|
|
|
|
|
; CI-NEXT: buffer_store_dword v2, v[0:1], s[0:3], 0 addr64
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: read2_ptr_is_subreg_f32:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v2, 2, v0
|
|
|
|
|
; GFX9-NEXT: v_add_u32_e32 v0, lds@abs32@lo, v2
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[0:1], v0 offset1:8
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v0, v0, v1
|
|
|
|
|
; GFX9-NEXT: global_store_dword v2, v0, s[0:1]
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2016-02-11 14:02:01 +08:00
|
|
|
%x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
|
2014-10-11 06:01:59 +08:00
|
|
|
%ptr.0 = insertelement <2 x [512 x float] addrspace(3)*> undef, [512 x float] addrspace(3)* @lds, i32 0
|
|
|
|
|
%ptr.1 = insertelement <2 x [512 x float] addrspace(3)*> %ptr.0, [512 x float] addrspace(3)* @lds, i32 1
|
|
|
|
|
%x.i.v.0 = insertelement <2 x i32> undef, i32 %x.i, i32 0
|
|
|
|
|
%x.i.v.1 = insertelement <2 x i32> %x.i.v.0, i32 %x.i, i32 1
|
|
|
|
|
%idx = add <2 x i32> %x.i.v.1, <i32 0, i32 8>
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%gep = getelementptr inbounds [512 x float], <2 x [512 x float] addrspace(3)*> %ptr.1, <2 x i32> <i32 0, i32 0>, <2 x i32> %idx
|
2014-10-11 06:01:59 +08:00
|
|
|
%gep.0 = extractelement <2 x float addrspace(3)*> %gep, i32 0
|
|
|
|
|
%gep.1 = extractelement <2 x float addrspace(3)*> %gep, i32 1
|
2015-02-28 05:17:42 +08:00
|
|
|
%val0 = load float, float addrspace(3)* %gep.0, align 4
|
|
|
|
|
%val1 = load float, float addrspace(3)* %gep.1, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%add.x = add nsw i32 %x.i, 8
|
|
|
|
|
%sum = fadd float %val0, %val1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%out.gep = getelementptr inbounds float, float addrspace(1)* %out, i32 %x.i
|
2014-10-11 06:01:59 +08:00
|
|
|
store float %sum, float addrspace(1)* %out.gep, align 4
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @simple_read2_f32_volatile_0(float addrspace(1)* %out) #0 {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: simple_read2_f32_volatile_0:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v1, vcc, lds@abs32@lo, v0
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: ds_read_b32 v2, v1
|
|
|
|
|
; CI-NEXT: ds_read_b32 v1, v1 offset:32
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_mov_b32 s3, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s2, 0
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v2, v2, v1
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v1, 0
|
|
|
|
|
; CI-NEXT: buffer_store_dword v2, v[0:1], s[0:3], 0 addr64
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: simple_read2_f32_volatile_0:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; GFX9-NEXT: v_add_u32_e32 v1, lds@abs32@lo, v0
|
|
|
|
|
; GFX9-NEXT: ds_read_b32 v2, v1
|
|
|
|
|
; GFX9-NEXT: ds_read_b32 v1, v1 offset:32
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v1, v2, v1
|
|
|
|
|
; GFX9-NEXT: global_store_dword v0, v1, s[0:1]
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2016-02-11 14:02:01 +08:00
|
|
|
%x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx0 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %x.i
|
2015-02-28 05:17:42 +08:00
|
|
|
%val0 = load volatile float, float addrspace(3)* %arrayidx0, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%add.x = add nsw i32 %x.i, 8
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx1 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %add.x
|
2015-02-28 05:17:42 +08:00
|
|
|
%val1 = load float, float addrspace(3)* %arrayidx1, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%sum = fadd float %val0, %val1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%out.gep = getelementptr inbounds float, float addrspace(1)* %out, i32 %x.i
|
2014-10-11 06:01:59 +08:00
|
|
|
store float %sum, float addrspace(1)* %out.gep, align 4
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @simple_read2_f32_volatile_1(float addrspace(1)* %out) #0 {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: simple_read2_f32_volatile_1:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v1, vcc, lds@abs32@lo, v0
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: ds_read_b32 v2, v1
|
|
|
|
|
; CI-NEXT: ds_read_b32 v1, v1 offset:32
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_mov_b32 s3, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s2, 0
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v2, v2, v1
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v1, 0
|
|
|
|
|
; CI-NEXT: buffer_store_dword v2, v[0:1], s[0:3], 0 addr64
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: simple_read2_f32_volatile_1:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; GFX9-NEXT: v_add_u32_e32 v1, lds@abs32@lo, v0
|
|
|
|
|
; GFX9-NEXT: ds_read_b32 v2, v1
|
|
|
|
|
; GFX9-NEXT: ds_read_b32 v1, v1 offset:32
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v1, v2, v1
|
|
|
|
|
; GFX9-NEXT: global_store_dword v0, v1, s[0:1]
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2016-02-11 14:02:01 +08:00
|
|
|
%x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx0 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %x.i
|
2015-02-28 05:17:42 +08:00
|
|
|
%val0 = load float, float addrspace(3)* %arrayidx0, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%add.x = add nsw i32 %x.i, 8
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx1 = getelementptr inbounds [512 x float], [512 x float] addrspace(3)* @lds, i32 0, i32 %add.x
|
2015-02-28 05:17:42 +08:00
|
|
|
%val1 = load volatile float, float addrspace(3)* %arrayidx1, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%sum = fadd float %val0, %val1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%out.gep = getelementptr inbounds float, float addrspace(1)* %out, i32 %x.i
|
2014-10-11 06:01:59 +08:00
|
|
|
store float %sum, float addrspace(1)* %out.gep, align 4
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Can't fold since not correctly aligned.
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @unaligned_read2_f32(float addrspace(1)* %out, float addrspace(3)* %lds) #0 {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: unaligned_read2_f32:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_load_dword s0, s[0:1], 0xb
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: s_mov_b32 s7, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s6, 0
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v1, vcc, s0, v0
|
|
|
|
|
; CI-NEXT: ds_read_u8 v2, v1 offset:35
|
|
|
|
|
; CI-NEXT: ds_read_u8 v3, v1 offset:34
|
|
|
|
|
; CI-NEXT: ds_read_u8 v4, v1 offset:33
|
|
|
|
|
; CI-NEXT: ds_read_u8 v5, v1 offset:32
|
|
|
|
|
; CI-NEXT: ds_read_u8 v6, v1 offset:3
|
|
|
|
|
; CI-NEXT: ds_read_u8 v7, v1 offset:2
|
|
|
|
|
; CI-NEXT: ds_read_u8 v8, v1 offset:1
|
|
|
|
|
; CI-NEXT: ds_read_u8 v1, v1
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(7)
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v2, 8, v2
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(3)
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v6, 8, v6
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(2)
|
|
|
|
|
; CI-NEXT: v_or_b32_e32 v6, v6, v7
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v4, 8, v4
|
|
|
|
|
; CI-NEXT: v_or_b32_e32 v2, v2, v3
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(1)
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v8, 8, v8
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_or_b32_e32 v1, v8, v1
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v6, 16, v6
|
|
|
|
|
; CI-NEXT: v_or_b32_e32 v4, v4, v5
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v2, 16, v2
|
|
|
|
|
; CI-NEXT: v_or_b32_e32 v1, v6, v1
|
|
|
|
|
; CI-NEXT: v_or_b32_e32 v2, v2, v4
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v2, v1, v2
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v1, 0
|
|
|
|
|
; CI-NEXT: buffer_store_dword v2, v[0:1], s[4:7], 0 addr64
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-ALIGNED-LABEL: unaligned_read2_f32:
|
|
|
|
|
; GFX9-ALIGNED: ; %bb.0:
|
|
|
|
|
; GFX9-ALIGNED-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x24
|
|
|
|
|
; GFX9-ALIGNED-NEXT: s_load_dword s0, s[0:1], 0x2c
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; GFX9-ALIGNED-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_add_u32_e32 v1, s0, v0
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read_u8 v2, v1
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read_u8 v3, v1 offset:1
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read_u8 v4, v1 offset:2
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read_u8 v5, v1 offset:3
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read_u8 v6, v1 offset:32
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read_u8 v7, v1 offset:33
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read_u8 v8, v1 offset:34
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read_u8 v1, v1 offset:35
|
|
|
|
|
; GFX9-ALIGNED-NEXT: s_waitcnt lgkmcnt(6)
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_lshl_or_b32 v2, v3, 8, v2
|
|
|
|
|
; GFX9-ALIGNED-NEXT: s_waitcnt lgkmcnt(4)
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_lshl_or_b32 v3, v5, 8, v4
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_lshl_or_b32 v2, v3, 16, v2
|
|
|
|
|
; GFX9-ALIGNED-NEXT: s_waitcnt lgkmcnt(2)
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_lshl_or_b32 v3, v7, 8, v6
|
|
|
|
|
; GFX9-ALIGNED-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_lshl_or_b32 v1, v1, 8, v8
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_lshl_or_b32 v1, v1, 16, v3
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_add_f32_e32 v1, v2, v1
|
|
|
|
|
; GFX9-ALIGNED-NEXT: global_store_dword v0, v1, s[2:3]
|
|
|
|
|
; GFX9-ALIGNED-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-UNALIGNED-LABEL: unaligned_read2_f32:
|
|
|
|
|
; GFX9-UNALIGNED: ; %bb.0:
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x24
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: s_load_dword s0, s[0:1], 0x2c
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: v_lshlrev_b32_e32 v2, 2, v0
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: v_add_u32_e32 v0, s0, v2
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: ds_read2_b32 v[0:1], v0 offset1:8
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: v_add_f32_e32 v0, v0, v1
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: global_store_dword v2, v0, s[2:3]
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: s_endpgm
|
2016-02-11 14:02:01 +08:00
|
|
|
%x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx0 = getelementptr inbounds float, float addrspace(3)* %lds, i32 %x.i
|
2015-02-28 05:17:42 +08:00
|
|
|
%val0 = load float, float addrspace(3)* %arrayidx0, align 1
|
2014-10-11 06:01:59 +08:00
|
|
|
%add.x = add nsw i32 %x.i, 8
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx1 = getelementptr inbounds float, float addrspace(3)* %lds, i32 %add.x
|
2015-02-28 05:17:42 +08:00
|
|
|
%val1 = load float, float addrspace(3)* %arrayidx1, align 1
|
2014-10-11 06:01:59 +08:00
|
|
|
%sum = fadd float %val0, %val1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%out.gep = getelementptr inbounds float, float addrspace(1)* %out, i32 %x.i
|
2014-10-11 06:01:59 +08:00
|
|
|
store float %sum, float addrspace(1)* %out.gep, align 4
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @misaligned_2_simple_read2_f32(float addrspace(1)* %out, float addrspace(3)* %lds) #0 {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: misaligned_2_simple_read2_f32:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_load_dword s0, s[0:1], 0xb
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: s_mov_b32 s7, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s6, 0
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v1, vcc, s0, v0
|
|
|
|
|
; CI-NEXT: ds_read_u16 v2, v1 offset:34
|
|
|
|
|
; CI-NEXT: ds_read_u16 v3, v1 offset:32
|
|
|
|
|
; CI-NEXT: ds_read_u16 v4, v1 offset:2
|
|
|
|
|
; CI-NEXT: ds_read_u16 v1, v1
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(3)
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v2, 16, v2
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(2)
|
|
|
|
|
; CI-NEXT: v_or_b32_e32 v2, v2, v3
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(1)
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v4, 16, v4
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_or_b32_e32 v1, v4, v1
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v2, v1, v2
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v1, 0
|
|
|
|
|
; CI-NEXT: buffer_store_dword v2, v[0:1], s[4:7], 0 addr64
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-ALIGNED-LABEL: misaligned_2_simple_read2_f32:
|
|
|
|
|
; GFX9-ALIGNED: ; %bb.0:
|
|
|
|
|
; GFX9-ALIGNED-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x24
|
|
|
|
|
; GFX9-ALIGNED-NEXT: s_load_dword s0, s[0:1], 0x2c
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; GFX9-ALIGNED-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_add_u32_e32 v1, s0, v0
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read_u16 v2, v1
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read_u16 v3, v1 offset:2
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read_u16 v4, v1 offset:32
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read_u16 v1, v1 offset:34
|
|
|
|
|
; GFX9-ALIGNED-NEXT: s_waitcnt lgkmcnt(2)
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_lshl_or_b32 v2, v3, 16, v2
|
|
|
|
|
; GFX9-ALIGNED-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_lshl_or_b32 v1, v1, 16, v4
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_add_f32_e32 v1, v2, v1
|
|
|
|
|
; GFX9-ALIGNED-NEXT: global_store_dword v0, v1, s[2:3]
|
|
|
|
|
; GFX9-ALIGNED-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-UNALIGNED-LABEL: misaligned_2_simple_read2_f32:
|
|
|
|
|
; GFX9-UNALIGNED: ; %bb.0:
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x24
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: s_load_dword s0, s[0:1], 0x2c
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: v_lshlrev_b32_e32 v2, 2, v0
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: v_add_u32_e32 v0, s0, v2
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: ds_read2_b32 v[0:1], v0 offset1:8
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: v_add_f32_e32 v0, v0, v1
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: global_store_dword v2, v0, s[2:3]
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: s_endpgm
|
2016-02-11 14:02:01 +08:00
|
|
|
%x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx0 = getelementptr inbounds float, float addrspace(3)* %lds, i32 %x.i
|
2015-02-28 05:17:42 +08:00
|
|
|
%val0 = load float, float addrspace(3)* %arrayidx0, align 2
|
2014-10-11 06:01:59 +08:00
|
|
|
%add.x = add nsw i32 %x.i, 8
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx1 = getelementptr inbounds float, float addrspace(3)* %lds, i32 %add.x
|
2015-02-28 05:17:42 +08:00
|
|
|
%val1 = load float, float addrspace(3)* %arrayidx1, align 2
|
2014-10-11 06:01:59 +08:00
|
|
|
%sum = fadd float %val0, %val1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%out.gep = getelementptr inbounds float, float addrspace(1)* %out, i32 %x.i
|
2014-10-11 06:01:59 +08:00
|
|
|
store float %sum, float addrspace(1)* %out.gep, align 4
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @simple_read2_f64(double addrspace(1)* %out) #0 {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: simple_read2_f64:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v4, 3, v0
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v0, vcc, lds.f64@abs32@lo, v4
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: ds_read2_b64 v[0:3], v0 offset1:8
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_mov_b32 s3, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s2, 0
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v5, 0
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_f64 v[0:1], v[0:1], v[2:3]
|
|
|
|
|
; CI-NEXT: buffer_store_dwordx2 v[0:1], v[4:5], s[0:3], 0 addr64
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: simple_read2_f64:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v4, 3, v0
|
|
|
|
|
; GFX9-NEXT: v_add_u32_e32 v0, lds.f64@abs32@lo, v4
|
|
|
|
|
; GFX9-NEXT: ds_read2_b64 v[0:3], v0 offset1:8
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_f64 v[0:1], v[0:1], v[2:3]
|
|
|
|
|
; GFX9-NEXT: global_store_dwordx2 v4, v[0:1], s[0:1]
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2016-02-11 14:02:01 +08:00
|
|
|
%x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx0 = getelementptr inbounds [512 x double], [512 x double] addrspace(3)* @lds.f64, i32 0, i32 %x.i
|
2015-02-28 05:17:42 +08:00
|
|
|
%val0 = load double, double addrspace(3)* %arrayidx0, align 8
|
2014-10-11 06:01:59 +08:00
|
|
|
%add.x = add nsw i32 %x.i, 8
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx1 = getelementptr inbounds [512 x double], [512 x double] addrspace(3)* @lds.f64, i32 0, i32 %add.x
|
2015-02-28 05:17:42 +08:00
|
|
|
%val1 = load double, double addrspace(3)* %arrayidx1, align 8
|
2014-10-11 06:01:59 +08:00
|
|
|
%sum = fadd double %val0, %val1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%out.gep = getelementptr inbounds double, double addrspace(1)* %out, i32 %x.i
|
2014-10-11 06:01:59 +08:00
|
|
|
store double %sum, double addrspace(1)* %out.gep, align 8
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @simple_read2_f64_max_offset(double addrspace(1)* %out) #0 {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: simple_read2_f64_max_offset:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v4, 3, v0
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v0, vcc, lds.f64@abs32@lo, v4
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: ds_read2_b64 v[0:3], v0 offset1:255
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_mov_b32 s3, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s2, 0
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v5, 0
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_f64 v[0:1], v[0:1], v[2:3]
|
|
|
|
|
; CI-NEXT: buffer_store_dwordx2 v[0:1], v[4:5], s[0:3], 0 addr64
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: simple_read2_f64_max_offset:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v4, 3, v0
|
|
|
|
|
; GFX9-NEXT: v_add_u32_e32 v0, lds.f64@abs32@lo, v4
|
|
|
|
|
; GFX9-NEXT: ds_read2_b64 v[0:3], v0 offset1:255
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_f64 v[0:1], v[0:1], v[2:3]
|
|
|
|
|
; GFX9-NEXT: global_store_dwordx2 v4, v[0:1], s[0:1]
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2016-02-11 14:02:01 +08:00
|
|
|
%x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx0 = getelementptr inbounds [512 x double], [512 x double] addrspace(3)* @lds.f64, i32 0, i32 %x.i
|
2015-02-28 05:17:42 +08:00
|
|
|
%val0 = load double, double addrspace(3)* %arrayidx0, align 8
|
2014-10-11 06:01:59 +08:00
|
|
|
%add.x = add nsw i32 %x.i, 255
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx1 = getelementptr inbounds [512 x double], [512 x double] addrspace(3)* @lds.f64, i32 0, i32 %add.x
|
2015-02-28 05:17:42 +08:00
|
|
|
%val1 = load double, double addrspace(3)* %arrayidx1, align 8
|
2014-10-11 06:01:59 +08:00
|
|
|
%sum = fadd double %val0, %val1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%out.gep = getelementptr inbounds double, double addrspace(1)* %out, i32 %x.i
|
2014-10-11 06:01:59 +08:00
|
|
|
store double %sum, double addrspace(1)* %out.gep, align 8
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @simple_read2_f64_too_far(double addrspace(1)* %out) #0 {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: simple_read2_f64_too_far:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v0, 3, v0
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v3, vcc, lds.f64@abs32@lo, v0
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: ds_read_b64 v[1:2], v3
|
|
|
|
|
; CI-NEXT: ds_read_b64 v[3:4], v3 offset:2056
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_mov_b32 s3, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s2, 0
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_f64 v[2:3], v[1:2], v[3:4]
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v1, 0
|
|
|
|
|
; CI-NEXT: buffer_store_dwordx2 v[2:3], v[0:1], s[0:3], 0 addr64
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: simple_read2_f64_too_far:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v4, 3, v0
|
|
|
|
|
; GFX9-NEXT: v_add_u32_e32 v2, lds.f64@abs32@lo, v4
|
|
|
|
|
; GFX9-NEXT: ds_read_b64 v[0:1], v2
|
|
|
|
|
; GFX9-NEXT: ds_read_b64 v[2:3], v2 offset:2056
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_f64 v[0:1], v[0:1], v[2:3]
|
|
|
|
|
; GFX9-NEXT: global_store_dwordx2 v4, v[0:1], s[0:1]
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2016-02-11 14:02:01 +08:00
|
|
|
%x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx0 = getelementptr inbounds [512 x double], [512 x double] addrspace(3)* @lds.f64, i32 0, i32 %x.i
|
2015-02-28 05:17:42 +08:00
|
|
|
%val0 = load double, double addrspace(3)* %arrayidx0, align 8
|
2014-10-11 06:12:32 +08:00
|
|
|
%add.x = add nsw i32 %x.i, 257
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx1 = getelementptr inbounds [512 x double], [512 x double] addrspace(3)* @lds.f64, i32 0, i32 %add.x
|
2015-02-28 05:17:42 +08:00
|
|
|
%val1 = load double, double addrspace(3)* %arrayidx1, align 8
|
2014-10-11 06:01:59 +08:00
|
|
|
%sum = fadd double %val0, %val1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%out.gep = getelementptr inbounds double, double addrspace(1)* %out, i32 %x.i
|
2014-10-11 06:01:59 +08:00
|
|
|
store double %sum, double addrspace(1)* %out.gep, align 8
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
; Alignment only 4
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @misaligned_read2_f64(double addrspace(1)* %out, double addrspace(3)* %lds) #0 {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: misaligned_read2_f64:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: s_load_dword s2, s[0:1], 0xb
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v0, 3, v0
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_mov_b32 s3, 0xf000
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v3, vcc, s2, v0
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[1:2], v3 offset1:1
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[3:4], v3 offset0:14 offset1:15
|
|
|
|
|
; CI-NEXT: s_mov_b32 s2, 0
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_f64 v[2:3], v[1:2], v[3:4]
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v1, 0
|
|
|
|
|
; CI-NEXT: buffer_store_dwordx2 v[2:3], v[0:1], s[0:3], 0 addr64
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: misaligned_read2_f64:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: s_load_dword s2, s[0:1], 0x2c
|
|
|
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v4, 3, v0
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_u32_e32 v2, s2, v4
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[0:1], v2 offset1:1
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[2:3], v2 offset0:14 offset1:15
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_f64 v[0:1], v[0:1], v[2:3]
|
|
|
|
|
; GFX9-NEXT: global_store_dwordx2 v4, v[0:1], s[0:1]
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2016-02-11 14:02:01 +08:00
|
|
|
%x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx0 = getelementptr inbounds double, double addrspace(3)* %lds, i32 %x.i
|
2015-02-28 05:17:42 +08:00
|
|
|
%val0 = load double, double addrspace(3)* %arrayidx0, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%add.x = add nsw i32 %x.i, 7
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx1 = getelementptr inbounds double, double addrspace(3)* %lds, i32 %add.x
|
2015-02-28 05:17:42 +08:00
|
|
|
%val1 = load double, double addrspace(3)* %arrayidx1, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%sum = fadd double %val0, %val1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%out.gep = getelementptr inbounds double, double addrspace(1)* %out, i32 %x.i
|
2014-10-11 06:01:59 +08:00
|
|
|
store double %sum, double addrspace(1)* %out.gep, align 4
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
2014-11-14 03:56:13 +08:00
|
|
|
@foo = addrspace(3) global [4 x i32] undef, align 4
|
2014-10-15 01:21:19 +08:00
|
|
|
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @load_constant_adjacent_offsets(i32 addrspace(1)* %out) {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: load_constant_adjacent_offsets:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v0, foo@abs32@lo
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[0:1], v0 offset1:1
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_mov_b32 s3, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s2, -1
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v0, vcc, v1, v0
|
|
|
|
|
; CI-NEXT: buffer_store_dword v0, off, s[0:3], 0
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: load_constant_adjacent_offsets:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v0, foo@abs32@lo
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[0:1], v0 offset1:1
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_u32_e32 v2, v0, v1
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v0, s0
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v1, s1
|
|
|
|
|
; GFX9-NEXT: global_store_dword v[0:1], v2, off
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2015-03-14 02:20:45 +08:00
|
|
|
%val0 = load i32, i32 addrspace(3)* getelementptr inbounds ([4 x i32], [4 x i32] addrspace(3)* @foo, i32 0, i32 0), align 4
|
|
|
|
|
%val1 = load i32, i32 addrspace(3)* getelementptr inbounds ([4 x i32], [4 x i32] addrspace(3)* @foo, i32 0, i32 1), align 4
|
2014-10-15 01:21:19 +08:00
|
|
|
%sum = add i32 %val0, %val1
|
|
|
|
|
store i32 %sum, i32 addrspace(1)* %out, align 4
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @load_constant_disjoint_offsets(i32 addrspace(1)* %out) {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: load_constant_disjoint_offsets:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v0, foo@abs32@lo
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[0:1], v0 offset1:2
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_mov_b32 s3, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s2, -1
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v0, vcc, v1, v0
|
|
|
|
|
; CI-NEXT: buffer_store_dword v0, off, s[0:3], 0
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: load_constant_disjoint_offsets:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v0, foo@abs32@lo
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[0:1], v0 offset1:2
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_u32_e32 v2, v0, v1
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v0, s0
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v1, s1
|
|
|
|
|
; GFX9-NEXT: global_store_dword v[0:1], v2, off
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2015-03-14 02:20:45 +08:00
|
|
|
%val0 = load i32, i32 addrspace(3)* getelementptr inbounds ([4 x i32], [4 x i32] addrspace(3)* @foo, i32 0, i32 0), align 4
|
|
|
|
|
%val1 = load i32, i32 addrspace(3)* getelementptr inbounds ([4 x i32], [4 x i32] addrspace(3)* @foo, i32 0, i32 2), align 4
|
2014-10-15 01:21:19 +08:00
|
|
|
%sum = add i32 %val0, %val1
|
|
|
|
|
store i32 %sum, i32 addrspace(1)* %out, align 4
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
2014-11-14 03:56:13 +08:00
|
|
|
@bar = addrspace(3) global [4 x i64] undef, align 4
|
2014-10-16 02:06:43 +08:00
|
|
|
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @load_misaligned64_constant_offsets(i64 addrspace(1)* %out) {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: load_misaligned64_constant_offsets:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v2, bar@abs32@lo
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[0:1], v2 offset1:1
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[2:3], v2 offset0:2 offset1:3
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_mov_b32 s3, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s2, -1
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v0, vcc, v0, v2
|
|
|
|
|
; CI-NEXT: v_addc_u32_e32 v1, vcc, v1, v3, vcc
|
|
|
|
|
; CI-NEXT: buffer_store_dwordx2 v[0:1], off, s[0:3], 0
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-ALIGNED-LABEL: load_misaligned64_constant_offsets:
|
|
|
|
|
; GFX9-ALIGNED: ; %bb.0:
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_mov_b32_e32 v2, bar@abs32@lo
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read2_b32 v[0:1], v2 offset1:1
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read2_b32 v[2:3], v2 offset0:2 offset1:3
|
|
|
|
|
; GFX9-ALIGNED-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24
|
|
|
|
|
; GFX9-ALIGNED-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_add_co_u32_e32 v0, vcc, v0, v2
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_addc_co_u32_e32 v1, vcc, v1, v3, vcc
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_mov_b32_e32 v3, s1
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_mov_b32_e32 v2, s0
|
|
|
|
|
; GFX9-ALIGNED-NEXT: global_store_dwordx2 v[2:3], v[0:1], off
|
|
|
|
|
; GFX9-ALIGNED-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-UNALIGNED-LABEL: load_misaligned64_constant_offsets:
|
|
|
|
|
; GFX9-UNALIGNED: ; %bb.0:
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: v_mov_b32_e32 v0, bar@abs32@lo
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: ds_read_b128 v[0:3], v0
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: v_add_co_u32_e32 v0, vcc, v0, v2
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: v_addc_co_u32_e32 v1, vcc, v1, v3, vcc
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: v_mov_b32_e32 v3, s1
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: v_mov_b32_e32 v2, s0
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: global_store_dwordx2 v[2:3], v[0:1], off
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: s_endpgm
|
2015-03-14 02:20:45 +08:00
|
|
|
%val0 = load i64, i64 addrspace(3)* getelementptr inbounds ([4 x i64], [4 x i64] addrspace(3)* @bar, i32 0, i32 0), align 4
|
|
|
|
|
%val1 = load i64, i64 addrspace(3)* getelementptr inbounds ([4 x i64], [4 x i64] addrspace(3)* @bar, i32 0, i32 1), align 4
|
2014-10-16 02:06:43 +08:00
|
|
|
%sum = add i64 %val0, %val1
|
|
|
|
|
store i64 %sum, i64 addrspace(1)* %out, align 8
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
2014-11-14 03:56:13 +08:00
|
|
|
@bar.large = addrspace(3) global [4096 x i64] undef, align 4
|
2014-10-16 02:06:43 +08:00
|
|
|
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @load_misaligned64_constant_large_offsets(i64 addrspace(1)* %out) {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: load_misaligned64_constant_large_offsets:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: s_mov_b32 s4, bar.large@abs32@lo
|
|
|
|
|
; CI-NEXT: s_add_i32 s5, s4, 0x4000
|
|
|
|
|
; CI-NEXT: s_addk_i32 s4, 0x7ff8
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v0, s5
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v2, s4
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[0:1], v0 offset1:1
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[2:3], v2 offset1:1
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_mov_b32 s3, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s2, -1
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v0, vcc, v0, v2
|
|
|
|
|
; CI-NEXT: v_addc_u32_e32 v1, vcc, v1, v3, vcc
|
|
|
|
|
; CI-NEXT: buffer_store_dwordx2 v[0:1], off, s[0:3], 0
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: load_misaligned64_constant_large_offsets:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: s_mov_b32 s2, bar.large@abs32@lo
|
|
|
|
|
; GFX9-NEXT: s_add_i32 s3, s2, 0x4000
|
|
|
|
|
; GFX9-NEXT: s_addk_i32 s2, 0x7ff8
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v0, s3
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v2, s2
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[0:1], v0 offset1:1
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[2:3], v2 offset1:1
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_co_u32_e32 v0, vcc, v0, v2
|
|
|
|
|
; GFX9-NEXT: v_addc_co_u32_e32 v1, vcc, v1, v3, vcc
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v3, s1
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v2, s0
|
|
|
|
|
; GFX9-NEXT: global_store_dwordx2 v[2:3], v[0:1], off
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2015-03-14 02:20:45 +08:00
|
|
|
%val0 = load i64, i64 addrspace(3)* getelementptr inbounds ([4096 x i64], [4096 x i64] addrspace(3)* @bar.large, i32 0, i32 2048), align 4
|
|
|
|
|
%val1 = load i64, i64 addrspace(3)* getelementptr inbounds ([4096 x i64], [4096 x i64] addrspace(3)* @bar.large, i32 0, i32 4095), align 4
|
2014-10-16 02:06:43 +08:00
|
|
|
%sum = add i64 %val0, %val1
|
|
|
|
|
store i64 %sum, i64 addrspace(1)* %out, align 8
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
2014-11-14 03:56:13 +08:00
|
|
|
@sgemm.lA = internal unnamed_addr addrspace(3) global [264 x float] undef, align 4
|
|
|
|
|
@sgemm.lB = internal unnamed_addr addrspace(3) global [776 x float] undef, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @sgemm_inner_loop_read2_sequence(float addrspace(1)* %C, i32 %lda, i32 %ldb) #0 {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: sgemm_inner_loop_read2_sequence:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_lshl_b32 s0, s2, 2
|
|
|
|
|
; CI-NEXT: s_add_i32 s1, s0, 0xc20
|
|
|
|
|
; CI-NEXT: s_addk_i32 s0, 0xc60
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v0, s1
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v4, s0
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[2:3], v0 offset1:1
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[4:5], v4 offset1:1
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v8, 2, v1
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[0:1], v8 offset1:1
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[6:7], v8 offset0:32 offset1:33
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[8:9], v8 offset0:64 offset1:65
|
|
|
|
|
; CI-NEXT: s_mov_b32 s7, 0xf000
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v2, v2, v3
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v2, v2, v4
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v2, v2, v5
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v0, v2, v0
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v0, v0, v1
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v0, v0, v6
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v0, v0, v7
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v0, v0, v8
|
|
|
|
|
; CI-NEXT: s_mov_b32 s6, -1
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v0, v0, v9
|
|
|
|
|
; CI-NEXT: buffer_store_dword v0, off, s[4:7], 0
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: sgemm_inner_loop_read2_sequence:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: s_lshl_b32 s2, s2, 2
|
|
|
|
|
; GFX9-NEXT: s_add_i32 s3, s2, 0xc20
|
|
|
|
|
; GFX9-NEXT: s_addk_i32 s2, 0xc60
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v0, s3
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v4, s2
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[2:3], v0 offset1:1
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[4:5], v4 offset1:1
|
|
|
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v8, 2, v1
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[0:1], v8 offset1:1
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[6:7], v8 offset0:32 offset1:33
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[8:9], v8 offset0:64 offset1:65
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v2, v2, v3
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v2, v2, v4
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v2, v2, v5
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v0, v2, v0
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v0, v0, v1
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v0, v0, v6
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v0, v0, v7
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v0, v0, v8
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v2, v0, v9
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v0, s0
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v1, s1
|
|
|
|
|
; GFX9-NEXT: global_store_dword v[0:1], v2, off
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2016-02-11 14:02:01 +08:00
|
|
|
%x.i = tail call i32 @llvm.amdgcn.workgroup.id.x() #1
|
|
|
|
|
%y.i = tail call i32 @llvm.amdgcn.workitem.id.y() #1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx44 = getelementptr inbounds [264 x float], [264 x float] addrspace(3)* @sgemm.lA, i32 0, i32 %x.i
|
2015-02-28 05:17:42 +08:00
|
|
|
%tmp16 = load float, float addrspace(3)* %arrayidx44, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%add47 = add nsw i32 %x.i, 1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx48 = getelementptr inbounds [264 x float], [264 x float] addrspace(3)* @sgemm.lA, i32 0, i32 %add47
|
2015-02-28 05:17:42 +08:00
|
|
|
%tmp17 = load float, float addrspace(3)* %arrayidx48, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%add51 = add nsw i32 %x.i, 16
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx52 = getelementptr inbounds [264 x float], [264 x float] addrspace(3)* @sgemm.lA, i32 0, i32 %add51
|
2015-02-28 05:17:42 +08:00
|
|
|
%tmp18 = load float, float addrspace(3)* %arrayidx52, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%add55 = add nsw i32 %x.i, 17
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx56 = getelementptr inbounds [264 x float], [264 x float] addrspace(3)* @sgemm.lA, i32 0, i32 %add55
|
2015-02-28 05:17:42 +08:00
|
|
|
%tmp19 = load float, float addrspace(3)* %arrayidx56, align 4
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx60 = getelementptr inbounds [776 x float], [776 x float] addrspace(3)* @sgemm.lB, i32 0, i32 %y.i
|
2015-02-28 05:17:42 +08:00
|
|
|
%tmp20 = load float, float addrspace(3)* %arrayidx60, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%add63 = add nsw i32 %y.i, 1
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx64 = getelementptr inbounds [776 x float], [776 x float] addrspace(3)* @sgemm.lB, i32 0, i32 %add63
|
2015-02-28 05:17:42 +08:00
|
|
|
%tmp21 = load float, float addrspace(3)* %arrayidx64, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%add67 = add nsw i32 %y.i, 32
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx68 = getelementptr inbounds [776 x float], [776 x float] addrspace(3)* @sgemm.lB, i32 0, i32 %add67
|
2015-02-28 05:17:42 +08:00
|
|
|
%tmp22 = load float, float addrspace(3)* %arrayidx68, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%add71 = add nsw i32 %y.i, 33
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx72 = getelementptr inbounds [776 x float], [776 x float] addrspace(3)* @sgemm.lB, i32 0, i32 %add71
|
2015-02-28 05:17:42 +08:00
|
|
|
%tmp23 = load float, float addrspace(3)* %arrayidx72, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%add75 = add nsw i32 %y.i, 64
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx76 = getelementptr inbounds [776 x float], [776 x float] addrspace(3)* @sgemm.lB, i32 0, i32 %add75
|
2015-02-28 05:17:42 +08:00
|
|
|
%tmp24 = load float, float addrspace(3)* %arrayidx76, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%add79 = add nsw i32 %y.i, 65
|
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction
One of several parallel first steps to remove the target type of pointers,
replacing them with a single opaque pointer type.
This adds an explicit type parameter to the gep instruction so that when the
first parameter becomes an opaque pointer type, the type to gep through is
still available to the instructions.
* This doesn't modify gep operators, only instructions (operators will be
handled separately)
* Textual IR changes only. Bitcode (including upgrade) and changing the
in-memory representation will be in separate changes.
* geps of vectors are transformed as:
getelementptr <4 x float*> %x, ...
->getelementptr float, <4 x float*> %x, ...
Then, once the opaque pointer type is introduced, this will ultimately look
like:
getelementptr float, <4 x ptr> %x
with the unambiguous interpretation that it is a vector of pointers to float.
* address spaces remain on the pointer, not the type:
getelementptr float addrspace(1)* %x
->getelementptr float, float addrspace(1)* %x
Then, eventually:
getelementptr float, ptr addrspace(1) %x
Importantly, the massive amount of test case churn has been automated by
same crappy python code. I had to manually update a few test cases that
wouldn't fit the script's model (r228970,r229196,r229197,r229198). The
python script just massages stdin and writes the result to stdout, I
then wrapped that in a shell script to handle replacing files, then
using the usual find+xargs to migrate all the files.
update.py:
import fileinput
import sys
import re
ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))")
def conv(match, line):
if not match:
return line
line = match.groups()[0]
if len(match.groups()[5]) == 0:
line += match.groups()[2]
line += match.groups()[3]
line += ", "
line += match.groups()[1]
line += "\n"
return line
for line in sys.stdin:
if line.find("getelementptr ") == line.find("getelementptr inbounds"):
if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("):
line = conv(re.match(ibrep, line), line)
elif line.find("getelementptr ") != line.find("getelementptr ("):
line = conv(re.match(normrep, line), line)
sys.stdout.write(line)
apply.sh:
for name in "$@"
do
python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name"
rm -f "$name.tmp"
done
The actual commands:
From llvm/src:
find test/ -name *.ll | xargs ./apply.sh
From llvm/src/tools/clang:
find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}"
From llvm/src/tools/polly:
find test/ -name *.ll | xargs ./apply.sh
After that, check-all (with llvm, clang, clang-tools-extra, lld,
compiler-rt, and polly all checked out).
The extra 'rm' in the apply.sh script is due to a few files in clang's test
suite using interesting unicode stuff that my python script was throwing
exceptions on. None of those files needed to be migrated, so it seemed
sufficient to ignore those cases.
Reviewers: rafael, dexonsmith, grosser
Differential Revision: http://reviews.llvm.org/D7636
llvm-svn: 230786
2015-02-28 03:29:02 +08:00
|
|
|
%arrayidx80 = getelementptr inbounds [776 x float], [776 x float] addrspace(3)* @sgemm.lB, i32 0, i32 %add79
|
2015-02-28 05:17:42 +08:00
|
|
|
%tmp25 = load float, float addrspace(3)* %arrayidx80, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
%sum.0 = fadd float %tmp16, %tmp17
|
|
|
|
|
%sum.1 = fadd float %sum.0, %tmp18
|
|
|
|
|
%sum.2 = fadd float %sum.1, %tmp19
|
|
|
|
|
%sum.3 = fadd float %sum.2, %tmp20
|
|
|
|
|
%sum.4 = fadd float %sum.3, %tmp21
|
|
|
|
|
%sum.5 = fadd float %sum.4, %tmp22
|
|
|
|
|
%sum.6 = fadd float %sum.5, %tmp23
|
|
|
|
|
%sum.7 = fadd float %sum.6, %tmp24
|
|
|
|
|
%sum.8 = fadd float %sum.7, %tmp25
|
|
|
|
|
store float %sum.8, float addrspace(1)* %C, align 4
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @misaligned_read2_v2i32(<2 x i32> addrspace(1)* %out, <2 x i32> addrspace(3)* %in) #0 {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: misaligned_read2_v2i32:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_load_dword s0, s[0:1], 0xb
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: s_mov_b32 s7, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s6, -1
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v0, s0
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[0:1], v0 offset1:1
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: buffer_store_dwordx2 v[0:1], off, s[4:7], 0
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: misaligned_read2_v2i32:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_load_dword s0, s[0:1], 0x2c
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v2, s2
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v0, s0
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[0:1], v0 offset1:1
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v3, s3
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: global_store_dwordx2 v[2:3], v[0:1], off
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2015-02-28 05:17:42 +08:00
|
|
|
%load = load <2 x i32>, <2 x i32> addrspace(3)* %in, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
store <2 x i32> %load, <2 x i32> addrspace(1)* %out, align 8
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
2017-03-22 05:39:51 +08:00
|
|
|
define amdgpu_kernel void @misaligned_read2_i64(i64 addrspace(1)* %out, i64 addrspace(3)* %in) #0 {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: misaligned_read2_i64:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_load_dword s0, s[0:1], 0xb
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: s_mov_b32 s7, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s6, -1
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v0, s0
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[0:1], v0 offset1:1
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: buffer_store_dwordx2 v[0:1], off, s[4:7], 0
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: misaligned_read2_i64:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_load_dword s0, s[0:1], 0x2c
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v2, s2
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v0, s0
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[0:1], v0 offset1:1
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v3, s3
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: global_store_dwordx2 v[2:3], v[0:1], off
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2015-02-28 05:17:42 +08:00
|
|
|
%load = load i64, i64 addrspace(3)* %in, align 4
|
2014-10-11 06:01:59 +08:00
|
|
|
store i64 %load, i64 addrspace(1)* %out, align 8
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
2016-11-03 22:37:13 +08:00
|
|
|
define amdgpu_kernel void @ds_read_diff_base_interleaving(
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: ds_read_diff_base_interleaving:
|
|
|
|
|
; CI: ; %bb.0: ; %bb
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0xb
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v1, 4, v1
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: s_mov_b32 s7, 0xf000
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v2, vcc, s0, v1
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v4, vcc, s1, v0
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v3, vcc, s2, v1
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v6, vcc, s3, v0
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[0:1], v2 offset1:1
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[2:3], v3 offset1:1
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[4:5], v4 offset1:4
|
|
|
|
|
; CI-NEXT: s_mov_b32 s6, -1
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_mul_f32_e32 v0, v0, v4
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v4, 2.0, v0
|
|
|
|
|
; CI-NEXT: v_mul_f32_e32 v5, v1, v5
|
|
|
|
|
; CI-NEXT: ds_read2_b32 v[0:1], v6 offset1:4
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_mul_f32_e32 v0, v2, v0
|
|
|
|
|
; CI-NEXT: v_sub_f32_e32 v0, v4, v0
|
|
|
|
|
; CI-NEXT: v_sub_f32_e32 v0, v0, v5
|
|
|
|
|
; CI-NEXT: v_mul_f32_e32 v1, v3, v1
|
|
|
|
|
; CI-NEXT: v_sub_f32_e32 v0, v0, v1
|
|
|
|
|
; CI-NEXT: buffer_store_dword v0, off, s[4:7], 0 offset:40
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: ds_read_diff_base_interleaving:
|
|
|
|
|
; GFX9: ; %bb.0: ; %bb
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x2c
|
|
|
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v1, 4, v1
|
|
|
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_u32_e32 v2, s0, v1
|
|
|
|
|
; GFX9-NEXT: v_add_u32_e32 v3, s1, v0
|
|
|
|
|
; GFX9-NEXT: v_add_u32_e32 v4, s2, v1
|
|
|
|
|
; GFX9-NEXT: v_add_u32_e32 v6, s3, v0
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[0:1], v2 offset1:1
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[2:3], v3 offset1:4
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[4:5], v4 offset1:1
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[6:7], v6 offset1:4
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(2)
|
|
|
|
|
; GFX9-NEXT: v_mul_f32_e32 v0, v0, v2
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v0, 2.0, v0
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_mul_f32_e32 v2, v4, v6
|
|
|
|
|
; GFX9-NEXT: v_sub_f32_e32 v0, v0, v2
|
|
|
|
|
; GFX9-NEXT: v_mul_f32_e32 v1, v1, v3
|
|
|
|
|
; GFX9-NEXT: v_sub_f32_e32 v0, v0, v1
|
|
|
|
|
; GFX9-NEXT: v_mul_f32_e32 v1, v5, v7
|
|
|
|
|
; GFX9-NEXT: v_sub_f32_e32 v2, v0, v1
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v0, s4
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v1, s5
|
|
|
|
|
; GFX9-NEXT: global_store_dword v[0:1], v2, off offset:40
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
2016-11-03 22:37:13 +08:00
|
|
|
float addrspace(1)* nocapture %arg,
|
|
|
|
|
[4 x [4 x float]] addrspace(3)* %arg1,
|
|
|
|
|
[4 x [4 x float]] addrspace(3)* %arg2,
|
|
|
|
|
[4 x [4 x float]] addrspace(3)* %arg3,
|
|
|
|
|
[4 x [4 x float]] addrspace(3)* %arg4) #1 {
|
|
|
|
|
bb:
|
|
|
|
|
%tmp = getelementptr float, float addrspace(1)* %arg, i64 10
|
|
|
|
|
%tmp5 = tail call i32 @llvm.amdgcn.workitem.id.x() #2
|
|
|
|
|
%tmp6 = tail call i32 @llvm.amdgcn.workitem.id.y() #2
|
|
|
|
|
%tmp7 = getelementptr [4 x [4 x float]], [4 x [4 x float]] addrspace(3)* %arg1, i32 0, i32 %tmp6, i32 0
|
|
|
|
|
%tmp8 = getelementptr [4 x [4 x float]], [4 x [4 x float]] addrspace(3)* %arg2, i32 0, i32 0, i32 %tmp5
|
|
|
|
|
%tmp9 = getelementptr [4 x [4 x float]], [4 x [4 x float]] addrspace(3)* %arg3, i32 0, i32 %tmp6, i32 0
|
|
|
|
|
%tmp10 = getelementptr [4 x [4 x float]], [4 x [4 x float]] addrspace(3)* %arg4, i32 0, i32 0, i32 %tmp5
|
|
|
|
|
%tmp11 = getelementptr [4 x [4 x float]], [4 x [4 x float]] addrspace(3)* %arg1, i32 0, i32 %tmp6, i32 1
|
|
|
|
|
%tmp12 = getelementptr [4 x [4 x float]], [4 x [4 x float]] addrspace(3)* %arg2, i32 0, i32 1, i32 %tmp5
|
|
|
|
|
%tmp13 = getelementptr [4 x [4 x float]], [4 x [4 x float]] addrspace(3)* %arg3, i32 0, i32 %tmp6, i32 1
|
|
|
|
|
%tmp14 = getelementptr [4 x [4 x float]], [4 x [4 x float]] addrspace(3)* %arg4, i32 0, i32 1, i32 %tmp5
|
|
|
|
|
%tmp15 = load float, float addrspace(3)* %tmp7
|
|
|
|
|
%tmp16 = load float, float addrspace(3)* %tmp8
|
|
|
|
|
%tmp17 = fmul float %tmp15, %tmp16
|
|
|
|
|
%tmp18 = fadd float 2.000000e+00, %tmp17
|
|
|
|
|
%tmp19 = load float, float addrspace(3)* %tmp9
|
|
|
|
|
%tmp20 = load float, float addrspace(3)* %tmp10
|
|
|
|
|
%tmp21 = fmul float %tmp19, %tmp20
|
|
|
|
|
%tmp22 = fsub float %tmp18, %tmp21
|
|
|
|
|
%tmp23 = load float, float addrspace(3)* %tmp11
|
|
|
|
|
%tmp24 = load float, float addrspace(3)* %tmp12
|
|
|
|
|
%tmp25 = fmul float %tmp23, %tmp24
|
|
|
|
|
%tmp26 = fsub float %tmp22, %tmp25
|
|
|
|
|
%tmp27 = load float, float addrspace(3)* %tmp13
|
|
|
|
|
%tmp28 = load float, float addrspace(3)* %tmp14
|
|
|
|
|
%tmp29 = fmul float %tmp27, %tmp28
|
|
|
|
|
%tmp30 = fsub float %tmp26, %tmp29
|
|
|
|
|
store float %tmp30, float addrspace(1)* %tmp
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
AMDGPU: Do not combine loads/store across physreg defs
Summary:
Since this pass operates on machine SSA form, this should only really
affect M0 in practice.
Fixes various piglit variable-indexing/vs-varying-array-mat4-index-*
Change-Id: Ib2a1dc3a8d7b08225a8da49a86f533faa0986aa8
Fixes: r317751 ("AMDGPU: Merge S_BUFFER_LOAD_DWORD_IMM into x2, x4")
Reviewers: arsenm, mareko, rampitec
Subscribers: kzhuravl, wdng, yaxunl, dstuttard, tpr, t-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D40343
llvm-svn: 325677
2018-02-21 21:31:35 +08:00
|
|
|
define amdgpu_kernel void @ds_read_call_read(i32 addrspace(1)* %out, i32 addrspace(3)* %arg) {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: ds_read_call_read:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: s_mov_b32 s40, SCRATCH_RSRC_DWORD0
|
|
|
|
|
; CI-NEXT: s_mov_b32 s41, SCRATCH_RSRC_DWORD1
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[36:37], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_load_dword s0, s[0:1], 0xb
|
|
|
|
|
; CI-NEXT: s_mov_b32 s42, -1
|
|
|
|
|
; CI-NEXT: s_mov_b32 s43, 0xe8f000
|
|
|
|
|
; CI-NEXT: s_add_u32 s40, s40, s3
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
|
|
|
; CI-NEXT: s_addc_u32 s41, s41, 0
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v40, vcc, s0, v0
|
|
|
|
|
; CI-NEXT: s_getpc_b64 s[0:1]
|
|
|
|
|
; CI-NEXT: s_add_u32 s0, s0, void_func_void@gotpcrel32@lo+4
|
|
|
|
|
; CI-NEXT: s_addc_u32 s1, s1, void_func_void@gotpcrel32@hi+12
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x0
|
|
|
|
|
; CI-NEXT: ds_read_b32 v41, v40
|
|
|
|
|
; CI-NEXT: s_mov_b64 s[0:1], s[40:41]
|
|
|
|
|
; CI-NEXT: s_mov_b64 s[2:3], s[42:43]
|
|
|
|
|
; CI-NEXT: s_mov_b32 s32, 0
|
|
|
|
|
; CI-NEXT: s_mov_b32 s39, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s38, -1
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: s_swappc_b64 s[30:31], s[4:5]
|
|
|
|
|
; CI-NEXT: ds_read_b32 v0, v40 offset:4
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_i32_e32 v0, vcc, v41, v0
|
|
|
|
|
; CI-NEXT: buffer_store_dword v0, off, s[36:39], 0
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: ds_read_call_read:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: s_mov_b32 s36, SCRATCH_RSRC_DWORD0
|
|
|
|
|
; GFX9-NEXT: s_mov_b32 s37, SCRATCH_RSRC_DWORD1
|
|
|
|
|
; GFX9-NEXT: s_mov_b32 s38, -1
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[34:35], s[0:1], 0x24
|
|
|
|
|
; GFX9-NEXT: s_load_dword s2, s[0:1], 0x2c
|
|
|
|
|
; GFX9-NEXT: s_mov_b32 s39, 0xe00000
|
|
|
|
|
; GFX9-NEXT: s_add_u32 s36, s36, s3
|
|
|
|
|
; GFX9-NEXT: s_addc_u32 s37, s37, 0
|
|
|
|
|
; GFX9-NEXT: s_getpc_b64 s[0:1]
|
|
|
|
|
; GFX9-NEXT: s_add_u32 s0, s0, void_func_void@gotpcrel32@lo+4
|
|
|
|
|
; GFX9-NEXT: s_addc_u32 s1, s1, void_func_void@gotpcrel32@hi+12
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_lshl_add_u32 v40, v0, 2, s2
|
|
|
|
|
; GFX9-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x0
|
|
|
|
|
; GFX9-NEXT: ds_read_b32 v41, v40
|
|
|
|
|
; GFX9-NEXT: s_mov_b64 s[0:1], s[36:37]
|
|
|
|
|
; GFX9-NEXT: s_mov_b64 s[2:3], s[38:39]
|
|
|
|
|
; GFX9-NEXT: s_mov_b32 s32, 0
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: s_swappc_b64 s[30:31], s[4:5]
|
|
|
|
|
; GFX9-NEXT: ds_read_b32 v0, v40 offset:4
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_u32_e32 v2, v41, v0
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v0, s34
|
|
|
|
|
; GFX9-NEXT: v_mov_b32_e32 v1, s35
|
|
|
|
|
; GFX9-NEXT: global_store_dword v[0:1], v2, off
|
|
|
|
|
; GFX9-NEXT: s_endpgm
|
AMDGPU: Do not combine loads/store across physreg defs
Summary:
Since this pass operates on machine SSA form, this should only really
affect M0 in practice.
Fixes various piglit variable-indexing/vs-varying-array-mat4-index-*
Change-Id: Ib2a1dc3a8d7b08225a8da49a86f533faa0986aa8
Fixes: r317751 ("AMDGPU: Merge S_BUFFER_LOAD_DWORD_IMM into x2, x4")
Reviewers: arsenm, mareko, rampitec
Subscribers: kzhuravl, wdng, yaxunl, dstuttard, tpr, t-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D40343
llvm-svn: 325677
2018-02-21 21:31:35 +08:00
|
|
|
%x = call i32 @llvm.amdgcn.workitem.id.x()
|
|
|
|
|
%arrayidx0 = getelementptr i32, i32 addrspace(3)* %arg, i32 %x
|
|
|
|
|
%arrayidx1 = getelementptr i32, i32 addrspace(3)* %arrayidx0, i32 1
|
|
|
|
|
%v0 = load i32, i32 addrspace(3)* %arrayidx0, align 4
|
|
|
|
|
call void @void_func_void()
|
|
|
|
|
%v1 = load i32, i32 addrspace(3)* %arrayidx1, align 4
|
|
|
|
|
%r = add i32 %v0, %v1
|
|
|
|
|
store i32 %r, i32 addrspace(1)* %out, align 4
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
2018-02-23 18:45:56 +08:00
|
|
|
define amdgpu_ps <2 x float> @ds_read_interp_read(i32 inreg %prims, float addrspace(3)* %inptr) {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: ds_read_interp_read:
|
|
|
|
|
; CI: ; %bb.0:
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: ds_read_b32 v2, v0
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, s0
|
|
|
|
|
; CI-NEXT: v_interp_mov_f32 v1, p10, attr0.x
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: ds_read_b32 v0, v0 offset:16
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_add_f32_e32 v1, v0, v1
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v0, v2
|
|
|
|
|
; CI-NEXT: ; return to shader part epilog
|
|
|
|
|
;
|
|
|
|
|
; GFX9-LABEL: ds_read_interp_read:
|
|
|
|
|
; GFX9: ; %bb.0:
|
|
|
|
|
; GFX9-NEXT: ds_read2_b32 v[0:1], v0 offset1:4
|
|
|
|
|
; GFX9-NEXT: s_mov_b32 m0, s0
|
|
|
|
|
; GFX9-NEXT: s_nop 0
|
|
|
|
|
; GFX9-NEXT: v_interp_mov_f32_e32 v2, p10, attr0.x
|
|
|
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-NEXT: v_add_f32_e32 v1, v1, v2
|
|
|
|
|
; GFX9-NEXT: ; return to shader part epilog
|
2018-02-23 18:45:56 +08:00
|
|
|
%v0 = load float, float addrspace(3)* %inptr, align 4
|
|
|
|
|
%intrp = call float @llvm.amdgcn.interp.mov(i32 0, i32 0, i32 0, i32 %prims)
|
|
|
|
|
%ptr1 = getelementptr float, float addrspace(3)* %inptr, i32 4
|
|
|
|
|
%v1 = load float, float addrspace(3)* %ptr1, align 4
|
|
|
|
|
%v1b = fadd float %v1, %intrp
|
|
|
|
|
%r0 = insertelement <2 x float> undef, float %v0, i32 0
|
|
|
|
|
%r1 = insertelement <2 x float> %r0, float %v1b, i32 1
|
|
|
|
|
ret <2 x float> %r1
|
|
|
|
|
}
|
|
|
|
|
|
2020-08-26 22:45:07 +08:00
|
|
|
@v2i32_align1 = internal addrspace(3) global [100 x <2 x i32>] undef, align 1
|
|
|
|
|
|
|
|
|
|
define amdgpu_kernel void @read2_v2i32_align1_odd_offset(<2 x i32> addrspace(1)* %out) {
|
2020-11-02 20:50:16 +08:00
|
|
|
; CI-LABEL: read2_v2i32_align1_odd_offset:
|
|
|
|
|
; CI: ; %bb.0: ; %entry
|
|
|
|
|
; CI-NEXT: v_mov_b32_e32 v0, 0
|
|
|
|
|
; CI-NEXT: s_mov_b32 m0, -1
|
|
|
|
|
; CI-NEXT: ds_read_u8 v1, v0 offset:72
|
|
|
|
|
; CI-NEXT: ds_read_u8 v2, v0 offset:71
|
|
|
|
|
; CI-NEXT: ds_read_u8 v3, v0 offset:70
|
|
|
|
|
; CI-NEXT: ds_read_u8 v4, v0 offset:69
|
|
|
|
|
; CI-NEXT: ds_read_u8 v5, v0 offset:68
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(4)
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v1, 8, v1
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(3)
|
|
|
|
|
; CI-NEXT: v_or_b32_e32 v1, v1, v2
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(2)
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v3, 8, v3
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(1)
|
|
|
|
|
; CI-NEXT: v_or_b32_e32 v3, v3, v4
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v1, 16, v1
|
|
|
|
|
; CI-NEXT: v_or_b32_e32 v1, v1, v3
|
|
|
|
|
; CI-NEXT: ds_read_u8 v2, v0 offset:67
|
|
|
|
|
; CI-NEXT: ds_read_u8 v3, v0 offset:66
|
|
|
|
|
; CI-NEXT: ds_read_u8 v0, v0 offset:65
|
|
|
|
|
; CI-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x9
|
|
|
|
|
; CI-NEXT: s_mov_b32 s3, 0xf000
|
|
|
|
|
; CI-NEXT: s_mov_b32 s2, -1
|
|
|
|
|
; CI-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v3, 8, v3
|
|
|
|
|
; CI-NEXT: v_or_b32_e32 v0, v3, v0
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v3, 8, v5
|
|
|
|
|
; CI-NEXT: v_or_b32_e32 v2, v3, v2
|
|
|
|
|
; CI-NEXT: v_lshlrev_b32_e32 v2, 16, v2
|
|
|
|
|
; CI-NEXT: v_or_b32_e32 v0, v2, v0
|
|
|
|
|
; CI-NEXT: buffer_store_dwordx2 v[0:1], off, s[0:3], 0
|
|
|
|
|
; CI-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-ALIGNED-LABEL: read2_v2i32_align1_odd_offset:
|
|
|
|
|
; GFX9-ALIGNED: ; %bb.0: ; %entry
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_mov_b32_e32 v0, 0
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read_u8 v2, v0 offset:65
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read_u8 v3, v0 offset:66
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read_u8 v4, v0 offset:67
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read_u8 v5, v0 offset:68
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read_u8 v1, v0 offset:69
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read_u8 v6, v0 offset:70
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read_u8 v7, v0 offset:71
|
|
|
|
|
; GFX9-ALIGNED-NEXT: ds_read_u8 v0, v0 offset:72
|
|
|
|
|
; GFX9-ALIGNED-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24
|
|
|
|
|
; GFX9-ALIGNED-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_lshlrev_b32_e32 v6, 8, v6
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_or_b32_e32 v1, v6, v1
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_lshlrev_b32_e32 v0, 8, v0
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_or_b32_sdwa v0, v0, v7 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_or_b32_e32 v1, v0, v1
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_lshlrev_b32_e32 v0, 8, v3
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_or_b32_e32 v0, v0, v2
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_lshlrev_b32_e32 v2, 8, v5
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_or_b32_sdwa v2, v2, v4 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_or_b32_e32 v0, v2, v0
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_mov_b32_e32 v3, s1
|
|
|
|
|
; GFX9-ALIGNED-NEXT: v_mov_b32_e32 v2, s0
|
|
|
|
|
; GFX9-ALIGNED-NEXT: global_store_dwordx2 v[2:3], v[0:1], off
|
|
|
|
|
; GFX9-ALIGNED-NEXT: s_endpgm
|
|
|
|
|
;
|
|
|
|
|
; GFX9-UNALIGNED-LABEL: read2_v2i32_align1_odd_offset:
|
|
|
|
|
; GFX9-UNALIGNED: ; %bb.0: ; %entry
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: v_mov_b32_e32 v0, 0x41
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: ds_read2_b32 v[0:1], v0 offset1:1
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: s_waitcnt lgkmcnt(0)
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: v_mov_b32_e32 v3, s1
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: v_mov_b32_e32 v2, s0
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: global_store_dwordx2 v[2:3], v[0:1], off
|
|
|
|
|
; GFX9-UNALIGNED-NEXT: s_endpgm
|
2020-08-26 22:45:07 +08:00
|
|
|
entry:
|
|
|
|
|
%load = load <2 x i32>, <2 x i32> addrspace(3)* bitcast (i8 addrspace(3)* getelementptr (i8, i8 addrspace(3)* bitcast ([100 x <2 x i32>] addrspace(3)* @v2i32_align1 to i8 addrspace(3)*), i32 65) to <2 x i32> addrspace(3)*), align 1
|
|
|
|
|
store <2 x i32> %load, <2 x i32> addrspace(1)* %out
|
|
|
|
|
ret void
|
|
|
|
|
}
|
|
|
|
|
|
AMDGPU: Do not combine loads/store across physreg defs
Summary:
Since this pass operates on machine SSA form, this should only really
affect M0 in practice.
Fixes various piglit variable-indexing/vs-varying-array-mat4-index-*
Change-Id: Ib2a1dc3a8d7b08225a8da49a86f533faa0986aa8
Fixes: r317751 ("AMDGPU: Merge S_BUFFER_LOAD_DWORD_IMM into x2, x4")
Reviewers: arsenm, mareko, rampitec
Subscribers: kzhuravl, wdng, yaxunl, dstuttard, tpr, t-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D40343
llvm-svn: 325677
2018-02-21 21:31:35 +08:00
|
|
|
declare void @void_func_void() #3
|
|
|
|
|
|
2016-02-11 14:02:01 +08:00
|
|
|
declare i32 @llvm.amdgcn.workgroup.id.x() #1
|
|
|
|
|
declare i32 @llvm.amdgcn.workgroup.id.y() #1
|
|
|
|
|
declare i32 @llvm.amdgcn.workitem.id.x() #1
|
|
|
|
|
declare i32 @llvm.amdgcn.workitem.id.y() #1
|
2017-12-01 06:51:26 +08:00
|
|
|
|
2018-02-23 18:45:56 +08:00
|
|
|
declare float @llvm.amdgcn.interp.mov(i32, i32, i32, i32) nounwind readnone
|
|
|
|
|
|
2016-02-11 14:02:01 +08:00
|
|
|
declare void @llvm.amdgcn.s.barrier() #2
|
2014-10-11 06:01:59 +08:00
|
|
|
|
2016-07-12 07:35:48 +08:00
|
|
|
attributes #0 = { nounwind }
|
2017-12-01 06:51:26 +08:00
|
|
|
attributes #1 = { nounwind readnone speculatable }
|
2015-12-19 09:46:41 +08:00
|
|
|
attributes #2 = { convergent nounwind }
|
AMDGPU: Do not combine loads/store across physreg defs
Summary:
Since this pass operates on machine SSA form, this should only really
affect M0 in practice.
Fixes various piglit variable-indexing/vs-varying-array-mat4-index-*
Change-Id: Ib2a1dc3a8d7b08225a8da49a86f533faa0986aa8
Fixes: r317751 ("AMDGPU: Merge S_BUFFER_LOAD_DWORD_IMM into x2, x4")
Reviewers: arsenm, mareko, rampitec
Subscribers: kzhuravl, wdng, yaxunl, dstuttard, tpr, t-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D40343
llvm-svn: 325677
2018-02-21 21:31:35 +08:00
|
|
|
attributes #3 = { nounwind noinline }
|