maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/CodeGen/AMDGPU/fma-combine.ll

678 lines
34 KiB
LLVM
Raw Normal View History

AMDGPU: Fix -enable-var-scope violations llvm-svn: 318004
2017-11-13 07:53:44 +08:00
; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=tahiti -verify-machineinstrs -fp-contract=fast < %s | FileCheck -enable-var-scope -check-prefix=SI-NOFMA -check-prefix=SI-SAFE -check-prefix=SI -check-prefix=FUNC %s
; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=verde -verify-machineinstrs -fp-contract=fast < %s | FileCheck -enable-var-scope -check-prefix=SI-NOFMA -check-prefix=SI-SAFE -check-prefix=SI -check-prefix=FUNC %s
; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=tahiti -verify-machineinstrs -fp-contract=fast -enable-no-infs-fp-math -enable-unsafe-fp-math -mattr=+fp32-denormals < %s | FileCheck -enable-var-scope -check-prefix=SI-FMA -check-prefix=SI-UNSAFE -check-prefix=SI -check-prefix=FUNC %s
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
; Note: The SI-FMA conversions of type x * (y + 1) --> x * y + x would be
; beneficial even without fp32 denormals, but they do require no-infs-fp-math
; for correctness.
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
AMDGPU: Remove some old intrinsic uses from tests llvm-svn: 260493
2016-02-11 14:02:01 +08:00
declare i32 @llvm.amdgcn.workitem.id.x() #0
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
declare double @llvm.fabs.f64(double) #0
declare double @llvm.fma.f64(double, double, double) #0
declare float @llvm.fma.f32(float, float, float) #0
; (fadd (fmul x, y), z) -> (fma x, y, z)
; FUNC-LABEL: {{^}}combine_to_fma_f64_0:
; SI-DAG: buffer_load_dwordx2 [[A:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64{{$}}
; SI-DAG: buffer_load_dwordx2 [[B:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:8{{$}}
; SI-DAG: buffer_load_dwordx2 [[C:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:16{{$}}
; SI: v_fma_f64 [[RESULT:v\[[0-9]+:[0-9]+\]]], [[A]], [[B]], [[C]]
; SI: buffer_store_dwordx2 [[RESULT]]
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @combine_to_fma_f64_0(double addrspace(1)* noalias %out, double addrspace(1)* noalias %in) #1 {
AMDGPU: Remove some old intrinsic uses from tests llvm-svn: 260493
2016-02-11 14:02:01 +08:00
%tid = tail call i32 @llvm.amdgcn.workitem.id.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
%gep.0 = getelementptr double, double addrspace(1)* %in, i32 %tid
%gep.1 = getelementptr double, double addrspace(1)* %gep.0, i32 1
%gep.2 = getelementptr double, double addrspace(1)* %gep.0, i32 2
%gep.out = getelementptr double, double addrspace(1)* %out, i32 %tid
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
AMDGPU: Add volatile to test loads and stores When the memory vectorizer is enabled, these tests break. These tests don't really care about the memory instructions, and it's easier to write check lines with the unmerged loads. llvm-svn: 266071
2016-04-12 21:38:18 +08:00
%a = load volatile double, double addrspace(1)* %gep.0
%b = load volatile double, double addrspace(1)* %gep.1
%c = load volatile double, double addrspace(1)* %gep.2
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
%mul = fmul double %a, %b
%fma = fadd double %mul, %c
store double %fma, double addrspace(1)* %gep.out
ret void
}
; (fadd (fmul x, y), z) -> (fma x, y, z)
; FUNC-LABEL: {{^}}combine_to_fma_f64_0_2use:
; SI-DAG: buffer_load_dwordx2 [[A:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64{{$}}
; SI-DAG: buffer_load_dwordx2 [[B:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:8{{$}}
; SI-DAG: buffer_load_dwordx2 [[C:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:16{{$}}
; SI-DAG: buffer_load_dwordx2 [[D:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:24{{$}}
; SI-DAG: v_fma_f64 [[RESULT0:v\[[0-9]+:[0-9]+\]]], [[A]], [[B]], [[C]]
; SI-DAG: v_fma_f64 [[RESULT1:v\[[0-9]+:[0-9]+\]]], [[A]], [[B]], [[D]]
; SI-DAG: buffer_store_dwordx2 [[RESULT0]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64{{$}}
; SI-DAG: buffer_store_dwordx2 [[RESULT1]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:8{{$}}
; SI: s_endpgm
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @combine_to_fma_f64_0_2use(double addrspace(1)* noalias %out, double addrspace(1)* noalias %in) #1 {
AMDGPU: Remove some old intrinsic uses from tests llvm-svn: 260493
2016-02-11 14:02:01 +08:00
%tid = tail call i32 @llvm.amdgcn.workitem.id.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
%gep.0 = getelementptr double, double addrspace(1)* %in, i32 %tid
%gep.1 = getelementptr double, double addrspace(1)* %gep.0, i32 1
%gep.2 = getelementptr double, double addrspace(1)* %gep.0, i32 2
%gep.3 = getelementptr double, double addrspace(1)* %gep.0, i32 3
%gep.out.0 = getelementptr double, double addrspace(1)* %out, i32 %tid
%gep.out.1 = getelementptr double, double addrspace(1)* %gep.out.0, i32 1
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
AMDGPU: Add volatile to test loads and stores When the memory vectorizer is enabled, these tests break. These tests don't really care about the memory instructions, and it's easier to write check lines with the unmerged loads. llvm-svn: 266071
2016-04-12 21:38:18 +08:00
%a = load volatile double, double addrspace(1)* %gep.0
%b = load volatile double, double addrspace(1)* %gep.1
%c = load volatile double, double addrspace(1)* %gep.2
%d = load volatile double, double addrspace(1)* %gep.3
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
%mul = fmul double %a, %b
%fma0 = fadd double %mul, %c
%fma1 = fadd double %mul, %d
AMDGPU: Add volatile to test loads and stores When the memory vectorizer is enabled, these tests break. These tests don't really care about the memory instructions, and it's easier to write check lines with the unmerged loads. llvm-svn: 266071
2016-04-12 21:38:18 +08:00
store volatile double %fma0, double addrspace(1)* %gep.out.0
store volatile double %fma1, double addrspace(1)* %gep.out.1
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
ret void
}
; (fadd x, (fmul y, z)) -> (fma y, z, x)
; FUNC-LABEL: {{^}}combine_to_fma_f64_1:
; SI-DAG: buffer_load_dwordx2 [[A:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64{{$}}
; SI-DAG: buffer_load_dwordx2 [[B:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:8{{$}}
; SI-DAG: buffer_load_dwordx2 [[C:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:16{{$}}
; SI: v_fma_f64 [[RESULT:v\[[0-9]+:[0-9]+\]]], [[A]], [[B]], [[C]]
; SI: buffer_store_dwordx2 [[RESULT]]
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @combine_to_fma_f64_1(double addrspace(1)* noalias %out, double addrspace(1)* noalias %in) #1 {
AMDGPU: Remove some old intrinsic uses from tests llvm-svn: 260493
2016-02-11 14:02:01 +08:00
%tid = tail call i32 @llvm.amdgcn.workitem.id.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
%gep.0 = getelementptr double, double addrspace(1)* %in, i32 %tid
%gep.1 = getelementptr double, double addrspace(1)* %gep.0, i32 1
%gep.2 = getelementptr double, double addrspace(1)* %gep.0, i32 2
%gep.out = getelementptr double, double addrspace(1)* %out, i32 %tid
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
AMDGPU: Add volatile to test loads and stores When the memory vectorizer is enabled, these tests break. These tests don't really care about the memory instructions, and it's easier to write check lines with the unmerged loads. llvm-svn: 266071
2016-04-12 21:38:18 +08:00
%a = load volatile double, double addrspace(1)* %gep.0
%b = load volatile double, double addrspace(1)* %gep.1
%c = load volatile double, double addrspace(1)* %gep.2
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
%mul = fmul double %a, %b
%fma = fadd double %c, %mul
store double %fma, double addrspace(1)* %gep.out
ret void
}
; (fsub (fmul x, y), z) -> (fma x, y, (fneg z))
; FUNC-LABEL: {{^}}combine_to_fma_fsub_0_f64:
; SI-DAG: buffer_load_dwordx2 [[A:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64{{$}}
; SI-DAG: buffer_load_dwordx2 [[B:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:8{{$}}
; SI-DAG: buffer_load_dwordx2 [[C:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:16{{$}}
; SI: v_fma_f64 [[RESULT:v\[[0-9]+:[0-9]+\]]], [[A]], [[B]], -[[C]]
; SI: buffer_store_dwordx2 [[RESULT]]
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @combine_to_fma_fsub_0_f64(double addrspace(1)* noalias %out, double addrspace(1)* noalias %in) #1 {
AMDGPU: Remove some old intrinsic uses from tests llvm-svn: 260493
2016-02-11 14:02:01 +08:00
%tid = tail call i32 @llvm.amdgcn.workitem.id.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
%gep.0 = getelementptr double, double addrspace(1)* %in, i32 %tid
%gep.1 = getelementptr double, double addrspace(1)* %gep.0, i32 1
%gep.2 = getelementptr double, double addrspace(1)* %gep.0, i32 2
%gep.out = getelementptr double, double addrspace(1)* %out, i32 %tid
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
AMDGPU: Add volatile to test loads and stores When the memory vectorizer is enabled, these tests break. These tests don't really care about the memory instructions, and it's easier to write check lines with the unmerged loads. llvm-svn: 266071
2016-04-12 21:38:18 +08:00
%a = load volatile double, double addrspace(1)* %gep.0
%b = load volatile double, double addrspace(1)* %gep.1
%c = load volatile double, double addrspace(1)* %gep.2
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
%mul = fmul double %a, %b
%fma = fsub double %mul, %c
store double %fma, double addrspace(1)* %gep.out
ret void
}
; (fsub (fmul x, y), z) -> (fma x, y, (fneg z))
; FUNC-LABEL: {{^}}combine_to_fma_fsub_f64_0_2use:
; SI-DAG: buffer_load_dwordx2 [[A:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64{{$}}
; SI-DAG: buffer_load_dwordx2 [[B:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:8{{$}}
; SI-DAG: buffer_load_dwordx2 [[C:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:16{{$}}
; SI-DAG: buffer_load_dwordx2 [[D:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:24{{$}}
; SI-DAG: v_fma_f64 [[RESULT0:v\[[0-9]+:[0-9]+\]]], [[A]], [[B]], -[[C]]
; SI-DAG: v_fma_f64 [[RESULT1:v\[[0-9]+:[0-9]+\]]], [[A]], [[B]], -[[D]]
; SI-DAG: buffer_store_dwordx2 [[RESULT0]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64{{$}}
; SI-DAG: buffer_store_dwordx2 [[RESULT1]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:8{{$}}
; SI: s_endpgm
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @combine_to_fma_fsub_f64_0_2use(double addrspace(1)* noalias %out, double addrspace(1)* noalias %in) #1 {
AMDGPU: Remove some old intrinsic uses from tests llvm-svn: 260493
2016-02-11 14:02:01 +08:00
%tid = tail call i32 @llvm.amdgcn.workitem.id.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
%gep.0 = getelementptr double, double addrspace(1)* %in, i32 %tid
%gep.1 = getelementptr double, double addrspace(1)* %gep.0, i32 1
%gep.2 = getelementptr double, double addrspace(1)* %gep.0, i32 2
%gep.3 = getelementptr double, double addrspace(1)* %gep.0, i32 3
%gep.out.0 = getelementptr double, double addrspace(1)* %out, i32 %tid
%gep.out.1 = getelementptr double, double addrspace(1)* %gep.out.0, i32 1
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
AMDGPU: Add volatile to test loads and stores When the memory vectorizer is enabled, these tests break. These tests don't really care about the memory instructions, and it's easier to write check lines with the unmerged loads. llvm-svn: 266071
2016-04-12 21:38:18 +08:00
%a = load volatile double, double addrspace(1)* %gep.0
%b = load volatile double, double addrspace(1)* %gep.1
%c = load volatile double, double addrspace(1)* %gep.2
%d = load volatile double, double addrspace(1)* %gep.3
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
%mul = fmul double %a, %b
%fma0 = fsub double %mul, %c
%fma1 = fsub double %mul, %d
AMDGPU: Add volatile to test loads and stores When the memory vectorizer is enabled, these tests break. These tests don't really care about the memory instructions, and it's easier to write check lines with the unmerged loads. llvm-svn: 266071
2016-04-12 21:38:18 +08:00
store volatile double %fma0, double addrspace(1)* %gep.out.0
store volatile double %fma1, double addrspace(1)* %gep.out.1
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
ret void
}
; (fsub x, (fmul y, z)) -> (fma (fneg y), z, x)
; FUNC-LABEL: {{^}}combine_to_fma_fsub_1_f64:
; SI-DAG: buffer_load_dwordx2 [[A:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64{{$}}
; SI-DAG: buffer_load_dwordx2 [[B:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:8{{$}}
; SI-DAG: buffer_load_dwordx2 [[C:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:16{{$}}
; SI: v_fma_f64 [[RESULT:v\[[0-9]+:[0-9]+\]]], -[[A]], [[B]], [[C]]
; SI: buffer_store_dwordx2 [[RESULT]]
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @combine_to_fma_fsub_1_f64(double addrspace(1)* noalias %out, double addrspace(1)* noalias %in) #1 {
AMDGPU: Remove some old intrinsic uses from tests llvm-svn: 260493
2016-02-11 14:02:01 +08:00
%tid = tail call i32 @llvm.amdgcn.workitem.id.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
%gep.0 = getelementptr double, double addrspace(1)* %in, i32 %tid
%gep.1 = getelementptr double, double addrspace(1)* %gep.0, i32 1
%gep.2 = getelementptr double, double addrspace(1)* %gep.0, i32 2
%gep.out = getelementptr double, double addrspace(1)* %out, i32 %tid
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
AMDGPU: Add volatile to test loads and stores When the memory vectorizer is enabled, these tests break. These tests don't really care about the memory instructions, and it's easier to write check lines with the unmerged loads. llvm-svn: 266071
2016-04-12 21:38:18 +08:00
%a = load volatile double, double addrspace(1)* %gep.0
%b = load volatile double, double addrspace(1)* %gep.1
%c = load volatile double, double addrspace(1)* %gep.2
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
%mul = fmul double %a, %b
%fma = fsub double %c, %mul
store double %fma, double addrspace(1)* %gep.out
ret void
}
; (fsub x, (fmul y, z)) -> (fma (fneg y), z, x)
; FUNC-LABEL: {{^}}combine_to_fma_fsub_1_f64_2use:
; SI-DAG: buffer_load_dwordx2 [[A:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64{{$}}
; SI-DAG: buffer_load_dwordx2 [[B:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:8{{$}}
; SI-DAG: buffer_load_dwordx2 [[C:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:16{{$}}
; SI-DAG: buffer_load_dwordx2 [[D:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:24{{$}}
; SI-DAG: v_fma_f64 [[RESULT0:v\[[0-9]+:[0-9]+\]]], -[[A]], [[B]], [[C]]
; SI-DAG: v_fma_f64 [[RESULT1:v\[[0-9]+:[0-9]+\]]], -[[A]], [[B]], [[D]]
; SI-DAG: buffer_store_dwordx2 [[RESULT0]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64{{$}}
; SI-DAG: buffer_store_dwordx2 [[RESULT1]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:8{{$}}
; SI: s_endpgm
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @combine_to_fma_fsub_1_f64_2use(double addrspace(1)* noalias %out, double addrspace(1)* noalias %in) #1 {
AMDGPU: Remove some old intrinsic uses from tests llvm-svn: 260493
2016-02-11 14:02:01 +08:00
%tid = tail call i32 @llvm.amdgcn.workitem.id.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
%gep.0 = getelementptr double, double addrspace(1)* %in, i32 %tid
%gep.1 = getelementptr double, double addrspace(1)* %gep.0, i32 1
%gep.2 = getelementptr double, double addrspace(1)* %gep.0, i32 2
%gep.3 = getelementptr double, double addrspace(1)* %gep.0, i32 3
%gep.out.0 = getelementptr double, double addrspace(1)* %out, i32 %tid
%gep.out.1 = getelementptr double, double addrspace(1)* %gep.out.0, i32 1
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
AMDGPU: Add volatile to test loads and stores When the memory vectorizer is enabled, these tests break. These tests don't really care about the memory instructions, and it's easier to write check lines with the unmerged loads. llvm-svn: 266071
2016-04-12 21:38:18 +08:00
%a = load volatile double, double addrspace(1)* %gep.0
%b = load volatile double, double addrspace(1)* %gep.1
%c = load volatile double, double addrspace(1)* %gep.2
%d = load volatile double, double addrspace(1)* %gep.3
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
%mul = fmul double %a, %b
%fma0 = fsub double %c, %mul
%fma1 = fsub double %d, %mul
AMDGPU: Add volatile to test loads and stores When the memory vectorizer is enabled, these tests break. These tests don't really care about the memory instructions, and it's easier to write check lines with the unmerged loads. llvm-svn: 266071
2016-04-12 21:38:18 +08:00
store volatile double %fma0, double addrspace(1)* %gep.out.0
store volatile double %fma1, double addrspace(1)* %gep.out.1
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
ret void
}
; (fsub (fneg (fmul x, y)), z) -> (fma (fneg x), y, (fneg z))
; FUNC-LABEL: {{^}}combine_to_fma_fsub_2_f64:
; SI-DAG: buffer_load_dwordx2 [[A:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64{{$}}
; SI-DAG: buffer_load_dwordx2 [[B:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:8{{$}}
; SI-DAG: buffer_load_dwordx2 [[C:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:16{{$}}
; SI: v_fma_f64 [[RESULT:v\[[0-9]+:[0-9]+\]]], -[[A]], [[B]], -[[C]]
; SI: buffer_store_dwordx2 [[RESULT]]
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @combine_to_fma_fsub_2_f64(double addrspace(1)* noalias %out, double addrspace(1)* noalias %in) #1 {
AMDGPU: Remove some old intrinsic uses from tests llvm-svn: 260493
2016-02-11 14:02:01 +08:00
%tid = tail call i32 @llvm.amdgcn.workitem.id.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
%gep.0 = getelementptr double, double addrspace(1)* %in, i32 %tid
%gep.1 = getelementptr double, double addrspace(1)* %gep.0, i32 1
%gep.2 = getelementptr double, double addrspace(1)* %gep.0, i32 2
%gep.out = getelementptr double, double addrspace(1)* %out, i32 %tid
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
AMDGPU: Add volatile to test loads and stores When the memory vectorizer is enabled, these tests break. These tests don't really care about the memory instructions, and it's easier to write check lines with the unmerged loads. llvm-svn: 266071
2016-04-12 21:38:18 +08:00
%a = load volatile double, double addrspace(1)* %gep.0
%b = load volatile double, double addrspace(1)* %gep.1
%c = load volatile double, double addrspace(1)* %gep.2
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
%mul = fmul double %a, %b
%mul.neg = fsub double -0.0, %mul
%fma = fsub double %mul.neg, %c
store double %fma, double addrspace(1)* %gep.out
ret void
}
; (fsub (fneg (fmul x, y)), z) -> (fma (fneg x), y, (fneg z))
; FUNC-LABEL: {{^}}combine_to_fma_fsub_2_f64_2uses_neg:
; SI-DAG: buffer_load_dwordx2 [[A:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64{{$}}
; SI-DAG: buffer_load_dwordx2 [[B:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:8{{$}}
; SI-DAG: buffer_load_dwordx2 [[C:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:16{{$}}
AMDGPU: Fix -enable-var-scope violations llvm-svn: 318004
2017-11-13 07:53:44 +08:00
; SI-DAG: buffer_load_dwordx2 [[D:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:24{{$}}
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
; SI-DAG: v_fma_f64 [[RESULT0:v\[[0-9]+:[0-9]+\]]], -[[A]], [[B]], -[[C]]
; SI-DAG: v_fma_f64 [[RESULT1:v\[[0-9]+:[0-9]+\]]], -[[A]], [[B]], -[[D]]
; SI-DAG: buffer_store_dwordx2 [[RESULT0]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64{{$}}
; SI-DAG: buffer_store_dwordx2 [[RESULT1]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:8{{$}}
; SI: s_endpgm
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @combine_to_fma_fsub_2_f64_2uses_neg(double addrspace(1)* noalias %out, double addrspace(1)* noalias %in) #1 {
AMDGPU: Remove some old intrinsic uses from tests llvm-svn: 260493
2016-02-11 14:02:01 +08:00
%tid = tail call i32 @llvm.amdgcn.workitem.id.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
%gep.0 = getelementptr double, double addrspace(1)* %in, i32 %tid
%gep.1 = getelementptr double, double addrspace(1)* %gep.0, i32 1
%gep.2 = getelementptr double, double addrspace(1)* %gep.0, i32 2
%gep.3 = getelementptr double, double addrspace(1)* %gep.0, i32 3
%gep.out.0 = getelementptr double, double addrspace(1)* %out, i32 %tid
%gep.out.1 = getelementptr double, double addrspace(1)* %gep.out.0, i32 1
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
AMDGPU: Add volatile to test loads and stores When the memory vectorizer is enabled, these tests break. These tests don't really care about the memory instructions, and it's easier to write check lines with the unmerged loads. llvm-svn: 266071
2016-04-12 21:38:18 +08:00
%a = load volatile double, double addrspace(1)* %gep.0
%b = load volatile double, double addrspace(1)* %gep.1
%c = load volatile double, double addrspace(1)* %gep.2
%d = load volatile double, double addrspace(1)* %gep.3
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
%mul = fmul double %a, %b
%mul.neg = fsub double -0.0, %mul
%fma0 = fsub double %mul.neg, %c
%fma1 = fsub double %mul.neg, %d
AMDGPU: Add volatile to test loads and stores When the memory vectorizer is enabled, these tests break. These tests don't really care about the memory instructions, and it's easier to write check lines with the unmerged loads. llvm-svn: 266071
2016-04-12 21:38:18 +08:00
store volatile double %fma0, double addrspace(1)* %gep.out.0
store volatile double %fma1, double addrspace(1)* %gep.out.1
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
ret void
}
; (fsub (fneg (fmul x, y)), z) -> (fma (fneg x), y, (fneg z))
; FUNC-LABEL: {{^}}combine_to_fma_fsub_2_f64_2uses_mul:
; SI-DAG: buffer_load_dwordx2 [[A:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64{{$}}
; SI-DAG: buffer_load_dwordx2 [[B:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:8{{$}}
; SI-DAG: buffer_load_dwordx2 [[C:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:16{{$}}
AMDGPU: Fix -enable-var-scope violations llvm-svn: 318004
2017-11-13 07:53:44 +08:00
; SI-DAG: buffer_load_dwordx2 [[D:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:24{{$}}
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
; SI-DAG: v_fma_f64 [[RESULT0:v\[[0-9]+:[0-9]+\]]], -[[A]], [[B]], -[[C]]
; SI-DAG: v_fma_f64 [[RESULT1:v\[[0-9]+:[0-9]+\]]], [[A]], [[B]], -[[D]]
; SI-DAG: buffer_store_dwordx2 [[RESULT0]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64{{$}}
; SI-DAG: buffer_store_dwordx2 [[RESULT1]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:8{{$}}
; SI: s_endpgm
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @combine_to_fma_fsub_2_f64_2uses_mul(double addrspace(1)* noalias %out, double addrspace(1)* noalias %in) #1 {
AMDGPU: Remove some old intrinsic uses from tests llvm-svn: 260493
2016-02-11 14:02:01 +08:00
%tid = tail call i32 @llvm.amdgcn.workitem.id.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
%gep.0 = getelementptr double, double addrspace(1)* %in, i32 %tid
%gep.1 = getelementptr double, double addrspace(1)* %gep.0, i32 1
%gep.2 = getelementptr double, double addrspace(1)* %gep.0, i32 2
%gep.3 = getelementptr double, double addrspace(1)* %gep.0, i32 3
%gep.out.0 = getelementptr double, double addrspace(1)* %out, i32 %tid
%gep.out.1 = getelementptr double, double addrspace(1)* %gep.out.0, i32 1
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
AMDGPU: Add volatile to test loads and stores When the memory vectorizer is enabled, these tests break. These tests don't really care about the memory instructions, and it's easier to write check lines with the unmerged loads. llvm-svn: 266071
2016-04-12 21:38:18 +08:00
%a = load volatile double, double addrspace(1)* %gep.0
%b = load volatile double, double addrspace(1)* %gep.1
%c = load volatile double, double addrspace(1)* %gep.2
%d = load volatile double, double addrspace(1)* %gep.3
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
%mul = fmul double %a, %b
%mul.neg = fsub double -0.0, %mul
%fma0 = fsub double %mul.neg, %c
%fma1 = fsub double %mul, %d
AMDGPU: Add volatile to test loads and stores When the memory vectorizer is enabled, these tests break. These tests don't really care about the memory instructions, and it's easier to write check lines with the unmerged loads. llvm-svn: 266071
2016-04-12 21:38:18 +08:00
store volatile double %fma0, double addrspace(1)* %gep.out.0
store volatile double %fma1, double addrspace(1)* %gep.out.1
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
ret void
}
; fold (fsub (fma x, y, (fmul u, v)), z) -> (fma x, y (fma u, v, (fneg z)))
; FUNC-LABEL: {{^}}aggressive_combine_to_fma_fsub_0_f64:
; SI-DAG: buffer_load_dwordx2 [[X:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64{{$}}
; SI-DAG: buffer_load_dwordx2 [[Y:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:8{{$}}
; SI-DAG: buffer_load_dwordx2 [[Z:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:16{{$}}
; SI-DAG: buffer_load_dwordx2 [[U:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:24{{$}}
; SI-DAG: buffer_load_dwordx2 [[V:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:32{{$}}
[DAGCombine] require UnsafeFPMath for re-association of addition Summary: The affected transforms all implicitly use associativity of addition, for which we usually require unsafe math to be enabled. The "Aggressive" flag is only meant to convey information about the performance of the fused ops relative to a fmul+fadd sequence. Fixes Bug 31626. Reviewers: spatel, hfinkel, mehdi_amini, arsenm, tstellarAMD Subscribers: jholewinski, nemanjai, wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D28675 llvm-svn: 293635
2017-01-31 22:35:37 +08:00
; SI-SAFE: v_mul_f64 [[TMP0:v\[[0-9]+:[0-9]+\]]], [[U]], [[V]]
; SI-SAFE: v_fma_f64 [[TMP1:v\[[0-9]+:[0-9]+\]]], [[X]], [[Y]], [[TMP0]]
; SI-SAFE: v_add_f64 [[RESULT:v\[[0-9]+:[0-9]+\]]], [[TMP1]], -[[Z]]
; SI-UNSAFE: v_fma_f64 [[FMA0:v\[[0-9]+:[0-9]+\]]], [[U]], [[V]], -[[Z]]
; SI-UNSAFE: v_fma_f64 [[RESULT:v\[[0-9]+:[0-9]+\]]], [[X]], [[Y]], [[FMA0]]
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
; SI: buffer_store_dwordx2 [[RESULT]]
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @aggressive_combine_to_fma_fsub_0_f64(double addrspace(1)* noalias %out, double addrspace(1)* noalias %in) #1 {
AMDGPU: Remove some old intrinsic uses from tests llvm-svn: 260493
2016-02-11 14:02:01 +08:00
%tid = tail call i32 @llvm.amdgcn.workitem.id.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
%gep.0 = getelementptr double, double addrspace(1)* %in, i32 %tid
%gep.1 = getelementptr double, double addrspace(1)* %gep.0, i32 1
%gep.2 = getelementptr double, double addrspace(1)* %gep.0, i32 2
%gep.3 = getelementptr double, double addrspace(1)* %gep.0, i32 3
%gep.4 = getelementptr double, double addrspace(1)* %gep.0, i32 4
%gep.out = getelementptr double, double addrspace(1)* %out, i32 %tid
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
AMDGPU: Add volatile to test loads and stores When the memory vectorizer is enabled, these tests break. These tests don't really care about the memory instructions, and it's easier to write check lines with the unmerged loads. llvm-svn: 266071
2016-04-12 21:38:18 +08:00
%x = load volatile double, double addrspace(1)* %gep.0
%y = load volatile double, double addrspace(1)* %gep.1
%z = load volatile double, double addrspace(1)* %gep.2
%u = load volatile double, double addrspace(1)* %gep.3
%v = load volatile double, double addrspace(1)* %gep.4
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
%tmp0 = fmul double %u, %v
%tmp1 = call double @llvm.fma.f64(double %x, double %y, double %tmp0) #0
%tmp2 = fsub double %tmp1, %z
store double %tmp2, double addrspace(1)* %gep.out
ret void
}
; fold (fsub x, (fma y, z, (fmul u, v)))
; -> (fma (fneg y), z, (fma (fneg u), v, x))
; FUNC-LABEL: {{^}}aggressive_combine_to_fma_fsub_1_f64:
; SI-DAG: buffer_load_dwordx2 [[X:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64{{$}}
; SI-DAG: buffer_load_dwordx2 [[Y:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:8{{$}}
; SI-DAG: buffer_load_dwordx2 [[Z:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:16{{$}}
; SI-DAG: buffer_load_dwordx2 [[U:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:24{{$}}
; SI-DAG: buffer_load_dwordx2 [[V:v\[[0-9]+:[0-9]+\]]], v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:32{{$}}
[DAGCombine] require UnsafeFPMath for re-association of addition Summary: The affected transforms all implicitly use associativity of addition, for which we usually require unsafe math to be enabled. The "Aggressive" flag is only meant to convey information about the performance of the fused ops relative to a fmul+fadd sequence. Fixes Bug 31626. Reviewers: spatel, hfinkel, mehdi_amini, arsenm, tstellarAMD Subscribers: jholewinski, nemanjai, wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D28675 llvm-svn: 293635
2017-01-31 22:35:37 +08:00
; SI-SAFE: v_mul_f64 [[TMP0:v\[[0-9]+:[0-9]+\]]], [[U]], [[V]]
; SI-SAFE: v_fma_f64 [[TMP1:v\[[0-9]+:[0-9]+\]]], [[Y]], [[Z]], [[TMP0]]
; SI-SAFE: v_add_f64 [[RESULT:v\[[0-9]+:[0-9]+\]]], [[X]], -[[TMP1]]
; SI-UNSAFE: v_fma_f64 [[FMA0:v\[[0-9]+:[0-9]+\]]], -[[U]], [[V]], [[X]]
; SI-UNSAFE: v_fma_f64 [[RESULT:v\[[0-9]+:[0-9]+\]]], -[[Y]], [[Z]], [[FMA0]]
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
; SI: buffer_store_dwordx2 [[RESULT]]
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @aggressive_combine_to_fma_fsub_1_f64(double addrspace(1)* noalias %out, double addrspace(1)* noalias %in) #1 {
AMDGPU: Remove some old intrinsic uses from tests llvm-svn: 260493
2016-02-11 14:02:01 +08:00
%tid = tail call i32 @llvm.amdgcn.workitem.id.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
%gep.0 = getelementptr double, double addrspace(1)* %in, i32 %tid
%gep.1 = getelementptr double, double addrspace(1)* %gep.0, i32 1
%gep.2 = getelementptr double, double addrspace(1)* %gep.0, i32 2
%gep.3 = getelementptr double, double addrspace(1)* %gep.0, i32 3
%gep.4 = getelementptr double, double addrspace(1)* %gep.0, i32 4
%gep.out = getelementptr double, double addrspace(1)* %out, i32 %tid
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
AMDGPU: Add volatile to test loads and stores When the memory vectorizer is enabled, these tests break. These tests don't really care about the memory instructions, and it's easier to write check lines with the unmerged loads. llvm-svn: 266071
2016-04-12 21:38:18 +08:00
%x = load volatile double, double addrspace(1)* %gep.0
%y = load volatile double, double addrspace(1)* %gep.1
%z = load volatile double, double addrspace(1)* %gep.2
%u = load volatile double, double addrspace(1)* %gep.3
%v = load volatile double, double addrspace(1)* %gep.4
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
%tmp0 = fmul double %u, %v
%tmp1 = call double @llvm.fma.f64(double %y, double %z, double %tmp0) #0
%tmp2 = fsub double %x, %tmp1
store double %tmp2, double addrspace(1)* %gep.out
ret void
}
[DAGCombiner] Improve FMA support for interpolation patterns This patch adds support for combining patterns such as (FMUL(FADD(1.0, x), y)) and (FMUL(FSUB(x, 1.0), y)) to their FMA equivalents. This is useful in particular for linear interpolation cases such as (FADD(FMUL(x, t), FMUL(y, FSUB(1.0, t)))) Differential Revision: http://reviews.llvm.org/D13003 llvm-svn: 248210
2015-09-22 04:32:48 +08:00
;
; Patterns (+ fneg variants): mul(add(1.0,x),y), mul(sub(1.0,x),y), mul(sub(x,1.0),y)
;
; FUNC-LABEL: {{^}}test_f32_mul_add_x_one_y:
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
; SI-NOFMA: v_add_f32_e32 [[VS:v[0-9]]], 1.0, [[VX:v[0-9]]]
AMDGPU: Allow SIShrinkInstructions to work in non-SSA Immediates can be folded as long as the immediate is a vreg. Also undo commuting instructions if it didn't fold an immediate. llvm-svn: 307575
2017-07-11 03:53:57 +08:00
; SI-NOFMA: v_mul_f32_e32 {{v[0-9]}}, [[VS]], [[VY:v[0-9]]]
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
;
; SI-FMA: v_fma_f32 {{v[0-9]}}, [[VX:v[0-9]]], [[VY:v[0-9]]], [[VY:v[0-9]]]
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @test_f32_mul_add_x_one_y(float addrspace(1)* %out,
[DAGCombiner] Improve FMA support for interpolation patterns This patch adds support for combining patterns such as (FMUL(FADD(1.0, x), y)) and (FMUL(FSUB(x, 1.0), y)) to their FMA equivalents. This is useful in particular for linear interpolation cases such as (FADD(FMUL(x, t), FMUL(y, FSUB(1.0, t)))) Differential Revision: http://reviews.llvm.org/D13003 llvm-svn: 248210
2015-09-22 04:32:48 +08:00
float addrspace(1)* %in1,
float addrspace(1)* %in2) {
AMDGPU: Add volatile to test loads and stores When the memory vectorizer is enabled, these tests break. These tests don't really care about the memory instructions, and it's easier to write check lines with the unmerged loads. llvm-svn: 266071
2016-04-12 21:38:18 +08:00
%x = load volatile float, float addrspace(1)* %in1
%y = load volatile float, float addrspace(1)* %in2
[DAGCombiner] Improve FMA support for interpolation patterns This patch adds support for combining patterns such as (FMUL(FADD(1.0, x), y)) and (FMUL(FSUB(x, 1.0), y)) to their FMA equivalents. This is useful in particular for linear interpolation cases such as (FADD(FMUL(x, t), FMUL(y, FSUB(1.0, t)))) Differential Revision: http://reviews.llvm.org/D13003 llvm-svn: 248210
2015-09-22 04:32:48 +08:00
%a = fadd float %x, 1.0
%m = fmul float %a, %y
store float %m, float addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}test_f32_mul_y_add_x_one:
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
; SI-NOFMA: v_add_f32_e32 [[VS:v[0-9]]], 1.0, [[VX:v[0-9]]]
AMDGPU: Allow SIShrinkInstructions to work in non-SSA Immediates can be folded as long as the immediate is a vreg. Also undo commuting instructions if it didn't fold an immediate. llvm-svn: 307575
2017-07-11 03:53:57 +08:00
; SI-NOFMA: v_mul_f32_e32 {{v[0-9]}}, [[VY:v[0-9]]], [[VS]]
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
;
; SI-FMA: v_fma_f32 {{v[0-9]}}, [[VX:v[0-9]]], [[VY:v[0-9]]], [[VY:v[0-9]]]
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @test_f32_mul_y_add_x_one(float addrspace(1)* %out,
[DAGCombiner] Improve FMA support for interpolation patterns This patch adds support for combining patterns such as (FMUL(FADD(1.0, x), y)) and (FMUL(FSUB(x, 1.0), y)) to their FMA equivalents. This is useful in particular for linear interpolation cases such as (FADD(FMUL(x, t), FMUL(y, FSUB(1.0, t)))) Differential Revision: http://reviews.llvm.org/D13003 llvm-svn: 248210
2015-09-22 04:32:48 +08:00
float addrspace(1)* %in1,
float addrspace(1)* %in2) {
AMDGPU: Add volatile to test loads and stores When the memory vectorizer is enabled, these tests break. These tests don't really care about the memory instructions, and it's easier to write check lines with the unmerged loads. llvm-svn: 266071
2016-04-12 21:38:18 +08:00
%x = load volatile float, float addrspace(1)* %in1
%y = load volatile float, float addrspace(1)* %in2
[DAGCombiner] Improve FMA support for interpolation patterns This patch adds support for combining patterns such as (FMUL(FADD(1.0, x), y)) and (FMUL(FSUB(x, 1.0), y)) to their FMA equivalents. This is useful in particular for linear interpolation cases such as (FADD(FMUL(x, t), FMUL(y, FSUB(1.0, t)))) Differential Revision: http://reviews.llvm.org/D13003 llvm-svn: 248210
2015-09-22 04:32:48 +08:00
%a = fadd float %x, 1.0
%m = fmul float %y, %a
store float %m, float addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}test_f32_mul_add_x_negone_y:
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
; SI-NOFMA: v_add_f32_e32 [[VS:v[0-9]]], -1.0, [[VX:v[0-9]]]
AMDGPU: Allow SIShrinkInstructions to work in non-SSA Immediates can be folded as long as the immediate is a vreg. Also undo commuting instructions if it didn't fold an immediate. llvm-svn: 307575
2017-07-11 03:53:57 +08:00
; SI-NOFMA: v_mul_f32_e32 {{v[0-9]}}, [[VS]], [[VY:v[0-9]]]
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
;
; SI-FMA: v_fma_f32 {{v[0-9]}}, [[VX:v[0-9]]], [[VY:v[0-9]]], -[[VY:v[0-9]]]
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @test_f32_mul_add_x_negone_y(float addrspace(1)* %out,
[DAGCombiner] Improve FMA support for interpolation patterns This patch adds support for combining patterns such as (FMUL(FADD(1.0, x), y)) and (FMUL(FSUB(x, 1.0), y)) to their FMA equivalents. This is useful in particular for linear interpolation cases such as (FADD(FMUL(x, t), FMUL(y, FSUB(1.0, t)))) Differential Revision: http://reviews.llvm.org/D13003 llvm-svn: 248210
2015-09-22 04:32:48 +08:00
float addrspace(1)* %in1,
float addrspace(1)* %in2) {
%x = load float, float addrspace(1)* %in1
%y = load float, float addrspace(1)* %in2
%a = fadd float %x, -1.0
%m = fmul float %a, %y
store float %m, float addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}test_f32_mul_y_add_x_negone:
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
; SI-NOFMA: v_add_f32_e32 [[VS:v[0-9]]], -1.0, [[VX:v[0-9]]]
AMDGPU: Allow SIShrinkInstructions to work in non-SSA Immediates can be folded as long as the immediate is a vreg. Also undo commuting instructions if it didn't fold an immediate. llvm-svn: 307575
2017-07-11 03:53:57 +08:00
; SI-NOFMA: v_mul_f32_e32 {{v[0-9]}}, [[VY:v[0-9]]], [[VS]]
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
;
; SI-FMA: v_fma_f32 {{v[0-9]}}, [[VX:v[0-9]]], [[VY:v[0-9]]], -[[VY:v[0-9]]]
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @test_f32_mul_y_add_x_negone(float addrspace(1)* %out,
[DAGCombiner] Improve FMA support for interpolation patterns This patch adds support for combining patterns such as (FMUL(FADD(1.0, x), y)) and (FMUL(FSUB(x, 1.0), y)) to their FMA equivalents. This is useful in particular for linear interpolation cases such as (FADD(FMUL(x, t), FMUL(y, FSUB(1.0, t)))) Differential Revision: http://reviews.llvm.org/D13003 llvm-svn: 248210
2015-09-22 04:32:48 +08:00
float addrspace(1)* %in1,
float addrspace(1)* %in2) {
%x = load float, float addrspace(1)* %in1
%y = load float, float addrspace(1)* %in2
%a = fadd float %x, -1.0
%m = fmul float %y, %a
store float %m, float addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}test_f32_mul_sub_one_x_y:
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
; SI-NOFMA: v_sub_f32_e32 [[VS:v[0-9]]], 1.0, [[VX:v[0-9]]]
AMDGPU: Allow SIShrinkInstructions to work in non-SSA Immediates can be folded as long as the immediate is a vreg. Also undo commuting instructions if it didn't fold an immediate. llvm-svn: 307575
2017-07-11 03:53:57 +08:00
; SI-NOFMA: v_mul_f32_e32 {{v[0-9]}}, [[VS]], [[VY:v[0-9]]]
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
;
; SI-FMA: v_fma_f32 {{v[0-9]}}, -[[VX:v[0-9]]], [[VY:v[0-9]]], [[VY:v[0-9]]]
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @test_f32_mul_sub_one_x_y(float addrspace(1)* %out,
[DAGCombiner] Improve FMA support for interpolation patterns This patch adds support for combining patterns such as (FMUL(FADD(1.0, x), y)) and (FMUL(FSUB(x, 1.0), y)) to their FMA equivalents. This is useful in particular for linear interpolation cases such as (FADD(FMUL(x, t), FMUL(y, FSUB(1.0, t)))) Differential Revision: http://reviews.llvm.org/D13003 llvm-svn: 248210
2015-09-22 04:32:48 +08:00
float addrspace(1)* %in1,
float addrspace(1)* %in2) {
%x = load float, float addrspace(1)* %in1
%y = load float, float addrspace(1)* %in2
%s = fsub float 1.0, %x
%m = fmul float %s, %y
store float %m, float addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}test_f32_mul_y_sub_one_x:
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
; SI-NOFMA: v_sub_f32_e32 [[VS:v[0-9]]], 1.0, [[VX:v[0-9]]]
AMDGPU: Allow SIShrinkInstructions to work in non-SSA Immediates can be folded as long as the immediate is a vreg. Also undo commuting instructions if it didn't fold an immediate. llvm-svn: 307575
2017-07-11 03:53:57 +08:00
; SI-NOFMA: v_mul_f32_e32 {{v[0-9]}}, [[VY:v[0-9]]], [[VS]]
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
;
; SI-FMA: v_fma_f32 {{v[0-9]}}, -[[VX:v[0-9]]], [[VY:v[0-9]]], [[VY:v[0-9]]]
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @test_f32_mul_y_sub_one_x(float addrspace(1)* %out,
[DAGCombiner] Improve FMA support for interpolation patterns This patch adds support for combining patterns such as (FMUL(FADD(1.0, x), y)) and (FMUL(FSUB(x, 1.0), y)) to their FMA equivalents. This is useful in particular for linear interpolation cases such as (FADD(FMUL(x, t), FMUL(y, FSUB(1.0, t)))) Differential Revision: http://reviews.llvm.org/D13003 llvm-svn: 248210
2015-09-22 04:32:48 +08:00
float addrspace(1)* %in1,
float addrspace(1)* %in2) {
%x = load float, float addrspace(1)* %in1
%y = load float, float addrspace(1)* %in2
%s = fsub float 1.0, %x
%m = fmul float %y, %s
store float %m, float addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}test_f32_mul_sub_negone_x_y:
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
; SI-NOFMA: v_sub_f32_e32 [[VS:v[0-9]]], -1.0, [[VX:v[0-9]]]
AMDGPU: Allow SIShrinkInstructions to work in non-SSA Immediates can be folded as long as the immediate is a vreg. Also undo commuting instructions if it didn't fold an immediate. llvm-svn: 307575
2017-07-11 03:53:57 +08:00
; SI-NOFMA: v_mul_f32_e32 {{v[0-9]}}, [[VS]], [[VY:v[0-9]]]
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
;
; SI-FMA: v_fma_f32 {{v[0-9]}}, -[[VX:v[0-9]]], [[VY:v[0-9]]], -[[VY:v[0-9]]]
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @test_f32_mul_sub_negone_x_y(float addrspace(1)* %out,
[DAGCombiner] Improve FMA support for interpolation patterns This patch adds support for combining patterns such as (FMUL(FADD(1.0, x), y)) and (FMUL(FSUB(x, 1.0), y)) to their FMA equivalents. This is useful in particular for linear interpolation cases such as (FADD(FMUL(x, t), FMUL(y, FSUB(1.0, t)))) Differential Revision: http://reviews.llvm.org/D13003 llvm-svn: 248210
2015-09-22 04:32:48 +08:00
float addrspace(1)* %in1,
float addrspace(1)* %in2) {
%x = load float, float addrspace(1)* %in1
%y = load float, float addrspace(1)* %in2
%s = fsub float -1.0, %x
%m = fmul float %s, %y
store float %m, float addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}test_f32_mul_y_sub_negone_x:
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
; SI-NOFMA: v_sub_f32_e32 [[VS:v[0-9]]], -1.0, [[VX:v[0-9]]]
AMDGPU: Allow SIShrinkInstructions to work in non-SSA Immediates can be folded as long as the immediate is a vreg. Also undo commuting instructions if it didn't fold an immediate. llvm-svn: 307575
2017-07-11 03:53:57 +08:00
; SI-NOFMA: v_mul_f32_e32 {{v[0-9]}}, [[VY:v[0-9]]], [[VS]]
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
;
; SI-FMA: v_fma_f32 {{v[0-9]}}, -[[VX:v[0-9]]], [[VY:v[0-9]]], -[[VY:v[0-9]]]
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @test_f32_mul_y_sub_negone_x(float addrspace(1)* %out,
[DAGCombiner] Improve FMA support for interpolation patterns This patch adds support for combining patterns such as (FMUL(FADD(1.0, x), y)) and (FMUL(FSUB(x, 1.0), y)) to their FMA equivalents. This is useful in particular for linear interpolation cases such as (FADD(FMUL(x, t), FMUL(y, FSUB(1.0, t)))) Differential Revision: http://reviews.llvm.org/D13003 llvm-svn: 248210
2015-09-22 04:32:48 +08:00
float addrspace(1)* %in1,
float addrspace(1)* %in2) {
%x = load float, float addrspace(1)* %in1
%y = load float, float addrspace(1)* %in2
%s = fsub float -1.0, %x
%m = fmul float %y, %s
store float %m, float addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}test_f32_mul_sub_x_one_y:
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
; SI-NOFMA: v_add_f32_e32 [[VS:v[0-9]]], -1.0, [[VX:v[0-9]]]
AMDGPU: Allow SIShrinkInstructions to work in non-SSA Immediates can be folded as long as the immediate is a vreg. Also undo commuting instructions if it didn't fold an immediate. llvm-svn: 307575
2017-07-11 03:53:57 +08:00
; SI-NOFMA: v_mul_f32_e32 {{v[0-9]}}, [[VS]], [[VY:v[0-9]]]
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
;
; SI-FMA: v_fma_f32 {{v[0-9]}}, [[VX:v[0-9]]], [[VY:v[0-9]]], -[[VY:v[0-9]]]
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @test_f32_mul_sub_x_one_y(float addrspace(1)* %out,
[DAGCombiner] Improve FMA support for interpolation patterns This patch adds support for combining patterns such as (FMUL(FADD(1.0, x), y)) and (FMUL(FSUB(x, 1.0), y)) to their FMA equivalents. This is useful in particular for linear interpolation cases such as (FADD(FMUL(x, t), FMUL(y, FSUB(1.0, t)))) Differential Revision: http://reviews.llvm.org/D13003 llvm-svn: 248210
2015-09-22 04:32:48 +08:00
float addrspace(1)* %in1,
float addrspace(1)* %in2) {
%x = load float, float addrspace(1)* %in1
%y = load float, float addrspace(1)* %in2
%s = fsub float %x, 1.0
%m = fmul float %s, %y
store float %m, float addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}test_f32_mul_y_sub_x_one:
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
; SI-NOFMA: v_add_f32_e32 [[VS:v[0-9]]], -1.0, [[VX:v[0-9]]]
AMDGPU: Allow SIShrinkInstructions to work in non-SSA Immediates can be folded as long as the immediate is a vreg. Also undo commuting instructions if it didn't fold an immediate. llvm-svn: 307575
2017-07-11 03:53:57 +08:00
; SI-NOFMA: v_mul_f32_e32 {{v[0-9]}}, [[VY:v[0-9]]], [[VS]]
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
;
; SI-FMA: v_fma_f32 {{v[0-9]}}, [[VX:v[0-9]]], [[VY:v[0-9]]], -[[VY:v[0-9]]]
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @test_f32_mul_y_sub_x_one(float addrspace(1)* %out,
[DAGCombiner] Improve FMA support for interpolation patterns This patch adds support for combining patterns such as (FMUL(FADD(1.0, x), y)) and (FMUL(FSUB(x, 1.0), y)) to their FMA equivalents. This is useful in particular for linear interpolation cases such as (FADD(FMUL(x, t), FMUL(y, FSUB(1.0, t)))) Differential Revision: http://reviews.llvm.org/D13003 llvm-svn: 248210
2015-09-22 04:32:48 +08:00
float addrspace(1)* %in1,
float addrspace(1)* %in2) {
%x = load float, float addrspace(1)* %in1
%y = load float, float addrspace(1)* %in2
%s = fsub float %x, 1.0
%m = fmul float %y, %s
store float %m, float addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}test_f32_mul_sub_x_negone_y:
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
; SI-NOFMA: v_add_f32_e32 [[VS:v[0-9]]], 1.0, [[VX:v[0-9]]]
AMDGPU: Allow SIShrinkInstructions to work in non-SSA Immediates can be folded as long as the immediate is a vreg. Also undo commuting instructions if it didn't fold an immediate. llvm-svn: 307575
2017-07-11 03:53:57 +08:00
; SI-NOFMA: v_mul_f32_e32 {{v[0-9]}}, [[VS]], [[VY:v[0-9]]]
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
;
; SI-FMA: v_fma_f32 {{v[0-9]}}, [[VX:v[0-9]]], [[VY:v[0-9]]], [[VY:v[0-9]]]
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @test_f32_mul_sub_x_negone_y(float addrspace(1)* %out,
[DAGCombiner] Improve FMA support for interpolation patterns This patch adds support for combining patterns such as (FMUL(FADD(1.0, x), y)) and (FMUL(FSUB(x, 1.0), y)) to their FMA equivalents. This is useful in particular for linear interpolation cases such as (FADD(FMUL(x, t), FMUL(y, FSUB(1.0, t)))) Differential Revision: http://reviews.llvm.org/D13003 llvm-svn: 248210
2015-09-22 04:32:48 +08:00
float addrspace(1)* %in1,
float addrspace(1)* %in2) {
%x = load float, float addrspace(1)* %in1
%y = load float, float addrspace(1)* %in2
%s = fsub float %x, -1.0
%m = fmul float %s, %y
store float %m, float addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}test_f32_mul_y_sub_x_negone:
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
; SI-NOFMA: v_add_f32_e32 [[VS:v[0-9]]], 1.0, [[VX:v[0-9]]]
AMDGPU: Allow SIShrinkInstructions to work in non-SSA Immediates can be folded as long as the immediate is a vreg. Also undo commuting instructions if it didn't fold an immediate. llvm-svn: 307575
2017-07-11 03:53:57 +08:00
; SI-NOFMA: v_mul_f32_e32 {{v[0-9]}}, [[VY:v[0-9]]], [[VS]]
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
;
; SI-FMA: v_fma_f32 {{v[0-9]}}, [[VX:v[0-9]]], [[VY:v[0-9]]], [[VY:v[0-9]]]
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @test_f32_mul_y_sub_x_negone(float addrspace(1)* %out,
[DAGCombiner] Improve FMA support for interpolation patterns This patch adds support for combining patterns such as (FMUL(FADD(1.0, x), y)) and (FMUL(FSUB(x, 1.0), y)) to their FMA equivalents. This is useful in particular for linear interpolation cases such as (FADD(FMUL(x, t), FMUL(y, FSUB(1.0, t)))) Differential Revision: http://reviews.llvm.org/D13003 llvm-svn: 248210
2015-09-22 04:32:48 +08:00
float addrspace(1)* %in1,
float addrspace(1)* %in2) {
%x = load float, float addrspace(1)* %in1
%y = load float, float addrspace(1)* %in2
%s = fsub float %x, -1.0
%m = fmul float %y, %s
store float %m, float addrspace(1)* %out
ret void
}
;
; Interpolation Patterns: add(mul(x,t),mul(sub(1.0,t),y))
;
; FUNC-LABEL: {{^}}test_f32_interp:
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
; SI-NOFMA: v_sub_f32_e32 [[VT1:v[0-9]]], 1.0, [[VT:v[0-9]]]
AMDGPU: Allow SIShrinkInstructions to work in non-SSA Immediates can be folded as long as the immediate is a vreg. Also undo commuting instructions if it didn't fold an immediate. llvm-svn: 307575
2017-07-11 03:53:57 +08:00
; SI-NOFMA: v_mul_f32_e32 [[VTY:v[0-9]]], [[VY:v[0-9]]], [[VT1]]
; SI-NOFMA: v_mac_f32_e32 [[VTY]], [[VX:v[0-9]]], [[VT]]
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
;
; SI-FMA: v_fma_f32 [[VR:v[0-9]]], -[[VT:v[0-9]]], [[VY:v[0-9]]], [[VY]]
; SI-FMA: v_fma_f32 {{v[0-9]}}, [[VX:v[0-9]]], [[VT]], [[VR]]
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @test_f32_interp(float addrspace(1)* %out,
[DAGCombiner] Improve FMA support for interpolation patterns This patch adds support for combining patterns such as (FMUL(FADD(1.0, x), y)) and (FMUL(FSUB(x, 1.0), y)) to their FMA equivalents. This is useful in particular for linear interpolation cases such as (FADD(FMUL(x, t), FMUL(y, FSUB(1.0, t)))) Differential Revision: http://reviews.llvm.org/D13003 llvm-svn: 248210
2015-09-22 04:32:48 +08:00
float addrspace(1)* %in1,
float addrspace(1)* %in2,
float addrspace(1)* %in3) {
%x = load float, float addrspace(1)* %in1
%y = load float, float addrspace(1)* %in2
%t = load float, float addrspace(1)* %in3
%t1 = fsub float 1.0, %t
%tx = fmul float %x, %t
%ty = fmul float %y, %t1
%r = fadd float %tx, %ty
store float %r, float addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}test_f64_interp:
[DAGCombiner] do not fold (fmul (fadd X, 1), Y) -> (fmad X, Y, Y) by default Summary: When X = 0 and Y = inf, the original code produces inf, but the transformed code produces nan. So this transform (and its relatives) should only be used when the no-infs-fp-math flag is explicitly enabled. Also disable the transform using fmad (intermediate rounding) when unsafe-math is not enabled, since it can reduce the precision of the result; consider this example with binary floating point numbers with two bits of mantissa: x = 1.01 y = 111 x * (y + 1) = 1.01 * 1000 = 1010 (this is the exact result; no rounding occurs at any step) x * y + x = 1000.11 + 1.01 =r 1000 + 1.01 = 1001.01 =r 1000 (with rounding towards zero) The example relies on rounding towards zero at least in the second step. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98578 Reviewers: RKSimon, tstellarAMD, spatel, arsenm Subscribers: wdng, llvm-commits Differential Revision: https://reviews.llvm.org/D26602 llvm-svn: 288506
2016-12-03 00:06:18 +08:00
; SI-NOFMA: v_add_f64 [[VT1:v\[[0-9]+:[0-9]+\]]], -[[VT:v\[[0-9]+:[0-9]+\]]], 1.0
; SI-NOFMA: v_mul_f64 [[VTY:v\[[0-9]+:[0-9]+\]]], [[VY:v\[[0-9]+:[0-9]+\]]], [[VT1]]
; SI-NOFMA: v_fma_f64 v{{\[[0-9]+:[0-9]+\]}}, [[VX:v\[[0-9]+:[0-9]+\]]], [[VT]], [[VTY]]
;
; SI-FMA: v_fma_f64 [[VR:v\[[0-9]+:[0-9]+\]]], -[[VT:v\[[0-9]+:[0-9]+\]]], [[VY:v\[[0-9]+:[0-9]+\]]], [[VY]]
; SI-FMA: v_fma_f64 v{{\[[0-9]+:[0-9]+\]}}, [[VX:v\[[0-9]+:[0-9]+\]]], [[VT]], [[VR]]
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @test_f64_interp(double addrspace(1)* %out,
[DAGCombiner] Improve FMA support for interpolation patterns This patch adds support for combining patterns such as (FMUL(FADD(1.0, x), y)) and (FMUL(FSUB(x, 1.0), y)) to their FMA equivalents. This is useful in particular for linear interpolation cases such as (FADD(FMUL(x, t), FMUL(y, FSUB(1.0, t)))) Differential Revision: http://reviews.llvm.org/D13003 llvm-svn: 248210
2015-09-22 04:32:48 +08:00
double addrspace(1)* %in1,
double addrspace(1)* %in2,
double addrspace(1)* %in3) {
%x = load double, double addrspace(1)* %in1
%y = load double, double addrspace(1)* %in2
%t = load double, double addrspace(1)* %in3
%t1 = fsub double 1.0, %t
%tx = fmul double %x, %t
%ty = fmul double %y, %t1
%r = fadd double %tx, %ty
store double %r, double addrspace(1)* %out
ret void
}
DAG: Fold fma (fneg x), K, y -> fma x, -K, y llvm-svn: 316753
2017-10-27 17:06:07 +08:00
; Make sure negative constant cancels out fneg
; GCN-LABEL: {{^}}fma_neg_2.0_neg_a_b_f32:
; GCN: {{buffer|flat|global}}_load_dword [[A:v[0-9]+]]
; GCN: {{buffer|flat|global}}_load_dword [[B:v[0-9]+]]
; GCN-NOT: [[A]]
; GCN-NOT: [[B]]
; GCN: v_fma_f32 v{{[0-9]+}}, [[A]], 2.0, [[B]]
define amdgpu_kernel void @fma_neg_2.0_neg_a_b_f32(float addrspace(1)* %out, float addrspace(1)* %in) #0 {
%tid = call i32 @llvm.amdgcn.workitem.id.x()
%gep.0 = getelementptr float, float addrspace(1)* %out, i32 %tid
%gep.1 = getelementptr float, float addrspace(1)* %gep.0, i32 1
%gep.out = getelementptr float, float addrspace(1)* %out, i32 %tid
%r1 = load volatile float, float addrspace(1)* %gep.0
%r2 = load volatile float, float addrspace(1)* %gep.1
%r1.fneg = fsub float -0.000000e+00, %r1
%r3 = tail call float @llvm.fma.f32(float -2.0, float %r1.fneg, float %r2)
store float %r3, float addrspace(1)* %gep.out
ret void
}
; GCN-LABEL: {{^}}fma_2.0_neg_a_b_f32:
; GCN: {{buffer|flat|global}}_load_dword [[A:v[0-9]+]]
; GCN: {{buffer|flat|global}}_load_dword [[B:v[0-9]+]]
; GCN-NOT: [[A]]
; GCN-NOT: [[B]]
; GCN: v_fma_f32 v{{[0-9]+}}, [[A]], -2.0, [[B]]
define amdgpu_kernel void @fma_2.0_neg_a_b_f32(float addrspace(1)* %out, float addrspace(1)* %in) #0 {
%tid = call i32 @llvm.amdgcn.workitem.id.x()
%gep.0 = getelementptr float, float addrspace(1)* %out, i32 %tid
%gep.1 = getelementptr float, float addrspace(1)* %gep.0, i32 1
%gep.out = getelementptr float, float addrspace(1)* %out, i32 %tid
%r1 = load volatile float, float addrspace(1)* %gep.0
%r2 = load volatile float, float addrspace(1)* %gep.1
%r1.fneg = fsub float -0.000000e+00, %r1
%r3 = tail call float @llvm.fma.f32(float 2.0, float %r1.fneg, float %r2)
store float %r3, float addrspace(1)* %gep.out
ret void
}
R600/SI: Implement enableAggressiveFMAFusion Add tests for the various combines. This should always be at least cycle neutral on all subtargets for f64, and faster on some. For f32 we should prefer selecting v_mad_f32 over v_fma_f32. llvm-svn: 227484
2015-01-30 03:34:32 +08:00
attributes #0 = { nounwind readnone }
attributes #1 = { nounwind }
DAG: Fold fma (fneg x), K, y -> fma x, -K, y llvm-svn: 316753
2017-10-27 17:06:07 +08:00