maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/Transforms/SeparateConstOffsetFromGEP/NVPTX/split-gep-and-gvn.ll

197 lines
9.2 KiB
LLVM
Raw Normal View History

Add an optimization that does CSE in a group of similar GEPs. This optimization merges the common part of a group of GEPs, so we can compute each pointer address by adding a simple offset to the common part. The optimization is currently only enabled for the NVPTX backend, where it has a large payoff on some benchmarks. Review: http://reviews.llvm.org/D3462 Patch by Jingyue Wu. llvm-svn: 207783
2014-05-02 02:38:36 +08:00
; RUN: llc < %s -march=nvptx64 -mcpu=sm_20 | FileCheck %s --check-prefix=PTX
; RUN: opt < %s -S -separate-const-offset-from-gep -gvn -dce | FileCheck %s --check-prefix=IR
; Verifies the SeparateConstOffsetFromGEP pass.
; The following code computes
; *output = array[x][y] + array[x][y+1] + array[x+1][y] + array[x+1][y+1]
;
; We expect SeparateConstOffsetFromGEP to transform it to
;
; float *base = &a[x][y];
; *output = base[0] + base[1] + base[32] + base[33];
;
; so the backend can emit PTX that uses fewer virtual registers.
target datalayout = "e-i64:64-v16:16-v32:32-n16:32:64"
target triple = "nvptx64-unknown-unknown"
@array = internal addrspace(3) constant [32 x [32 x float]] zeroinitializer, align 4
define void @sum_of_array(i32 %x, i32 %y, float* nocapture %output) {
.preheader:
Fixed several correctness issues in SeparateConstOffsetFromGEP Most issues are on mishandling s/zext. Fixes: 1. When rebuilding new indices, s/zext should be distributed to sub-expressions. e.g., sext(a +nsw (b +nsw 5)) = sext(a) + sext(b) + 5 but not sext(a + b) + 5. This also affects the logic of recursively looking for a constant offset, we need to include s/zext into the context of the searching. 2. Function find should return the bitwidth of the constant offset instead of always sign-extending it to i64. 3. Stop shortcutting zext'ed GEP indices. LLVM conceptually sign-extends GEP indices to pointer-size before computing the address. Therefore, gep base, zext(a + b) != gep base, a + b Improvements: 1. Add an optimization for splitting sext(a + b): if a + b is proven non-negative (e.g., used as an index of an inbound GEP) and one of a, b is non-negative, sext(a + b) = sext(a) + sext(b) 2. Function Distributable checks whether both sext and zext can be distributed to operands of a binary operator. This helps us split zext(sext(a + b)) to zext(sext(a) + zext(sext(b)) when a + b does not signed or unsigned overflow. Refactoring: Merge some common logic of handling add/sub/or in find. Testing: Add many tests in split-gep.ll and split-gep-and-gvn.ll to verify the changes we made. llvm-svn: 210291
2014-06-06 06:07:33 +08:00
%0 = sext i32 %y to i64
%1 = sext i32 %x to i64
[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
%2 = getelementptr inbounds [32 x [32 x float]], [32 x [32 x float]] addrspace(3)* @array, i64 0, i64 %1, i64 %0
Add an optimization that does CSE in a group of similar GEPs. This optimization merges the common part of a group of GEPs, so we can compute each pointer address by adding a simple offset to the common part. The optimization is currently only enabled for the NVPTX backend, where it has a large payoff on some benchmarks. Review: http://reviews.llvm.org/D3462 Patch by Jingyue Wu. llvm-svn: 207783
2014-05-02 02:38:36 +08:00
%3 = addrspacecast float addrspace(3)* %2 to float*
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%4 = load float, float* %3, align 4
Add an optimization that does CSE in a group of similar GEPs. This optimization merges the common part of a group of GEPs, so we can compute each pointer address by adding a simple offset to the common part. The optimization is currently only enabled for the NVPTX backend, where it has a large payoff on some benchmarks. Review: http://reviews.llvm.org/D3462 Patch by Jingyue Wu. llvm-svn: 207783
2014-05-02 02:38:36 +08:00
%5 = fadd float %4, 0.000000e+00
%6 = add i32 %y, 1
Fixed several correctness issues in SeparateConstOffsetFromGEP Most issues are on mishandling s/zext. Fixes: 1. When rebuilding new indices, s/zext should be distributed to sub-expressions. e.g., sext(a +nsw (b +nsw 5)) = sext(a) + sext(b) + 5 but not sext(a + b) + 5. This also affects the logic of recursively looking for a constant offset, we need to include s/zext into the context of the searching. 2. Function find should return the bitwidth of the constant offset instead of always sign-extending it to i64. 3. Stop shortcutting zext'ed GEP indices. LLVM conceptually sign-extends GEP indices to pointer-size before computing the address. Therefore, gep base, zext(a + b) != gep base, a + b Improvements: 1. Add an optimization for splitting sext(a + b): if a + b is proven non-negative (e.g., used as an index of an inbound GEP) and one of a, b is non-negative, sext(a + b) = sext(a) + sext(b) 2. Function Distributable checks whether both sext and zext can be distributed to operands of a binary operator. This helps us split zext(sext(a + b)) to zext(sext(a) + zext(sext(b)) when a + b does not signed or unsigned overflow. Refactoring: Merge some common logic of handling add/sub/or in find. Testing: Add many tests in split-gep.ll and split-gep-and-gvn.ll to verify the changes we made. llvm-svn: 210291
2014-06-06 06:07:33 +08:00
%7 = sext i32 %6 to i64
[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
%8 = getelementptr inbounds [32 x [32 x float]], [32 x [32 x float]] addrspace(3)* @array, i64 0, i64 %1, i64 %7
Add an optimization that does CSE in a group of similar GEPs. This optimization merges the common part of a group of GEPs, so we can compute each pointer address by adding a simple offset to the common part. The optimization is currently only enabled for the NVPTX backend, where it has a large payoff on some benchmarks. Review: http://reviews.llvm.org/D3462 Patch by Jingyue Wu. llvm-svn: 207783
2014-05-02 02:38:36 +08:00
%9 = addrspacecast float addrspace(3)* %8 to float*
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%10 = load float, float* %9, align 4
Add an optimization that does CSE in a group of similar GEPs. This optimization merges the common part of a group of GEPs, so we can compute each pointer address by adding a simple offset to the common part. The optimization is currently only enabled for the NVPTX backend, where it has a large payoff on some benchmarks. Review: http://reviews.llvm.org/D3462 Patch by Jingyue Wu. llvm-svn: 207783
2014-05-02 02:38:36 +08:00
%11 = fadd float %5, %10
%12 = add i32 %x, 1
Fixed several correctness issues in SeparateConstOffsetFromGEP Most issues are on mishandling s/zext. Fixes: 1. When rebuilding new indices, s/zext should be distributed to sub-expressions. e.g., sext(a +nsw (b +nsw 5)) = sext(a) + sext(b) + 5 but not sext(a + b) + 5. This also affects the logic of recursively looking for a constant offset, we need to include s/zext into the context of the searching. 2. Function find should return the bitwidth of the constant offset instead of always sign-extending it to i64. 3. Stop shortcutting zext'ed GEP indices. LLVM conceptually sign-extends GEP indices to pointer-size before computing the address. Therefore, gep base, zext(a + b) != gep base, a + b Improvements: 1. Add an optimization for splitting sext(a + b): if a + b is proven non-negative (e.g., used as an index of an inbound GEP) and one of a, b is non-negative, sext(a + b) = sext(a) + sext(b) 2. Function Distributable checks whether both sext and zext can be distributed to operands of a binary operator. This helps us split zext(sext(a + b)) to zext(sext(a) + zext(sext(b)) when a + b does not signed or unsigned overflow. Refactoring: Merge some common logic of handling add/sub/or in find. Testing: Add many tests in split-gep.ll and split-gep-and-gvn.ll to verify the changes we made. llvm-svn: 210291
2014-06-06 06:07:33 +08:00
%13 = sext i32 %12 to i64
[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
%14 = getelementptr inbounds [32 x [32 x float]], [32 x [32 x float]] addrspace(3)* @array, i64 0, i64 %13, i64 %0
Add an optimization that does CSE in a group of similar GEPs. This optimization merges the common part of a group of GEPs, so we can compute each pointer address by adding a simple offset to the common part. The optimization is currently only enabled for the NVPTX backend, where it has a large payoff on some benchmarks. Review: http://reviews.llvm.org/D3462 Patch by Jingyue Wu. llvm-svn: 207783
2014-05-02 02:38:36 +08:00
%15 = addrspacecast float addrspace(3)* %14 to float*
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%16 = load float, float* %15, align 4
Add an optimization that does CSE in a group of similar GEPs. This optimization merges the common part of a group of GEPs, so we can compute each pointer address by adding a simple offset to the common part. The optimization is currently only enabled for the NVPTX backend, where it has a large payoff on some benchmarks. Review: http://reviews.llvm.org/D3462 Patch by Jingyue Wu. llvm-svn: 207783
2014-05-02 02:38:36 +08:00
%17 = fadd float %11, %16
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction One of several parallel first steps to remove the target type of pointers, replacing them with a single opaque pointer type. This adds an explicit type parameter to the gep instruction so that when the first parameter becomes an opaque pointer type, the type to gep through is still available to the instructions. * This doesn't modify gep operators, only instructions (operators will be handled separately) * Textual IR changes only. Bitcode (including upgrade) and changing the in-memory representation will be in separate changes. * geps of vectors are transformed as: getelementptr <4 x float*> %x, ... ->getelementptr float, <4 x float*> %x, ... Then, once the opaque pointer type is introduced, this will ultimately look like: getelementptr float, <4 x ptr> %x with the unambiguous interpretation that it is a vector of pointers to float. * address spaces remain on the pointer, not the type: getelementptr float addrspace(1)* %x ->getelementptr float, float addrspace(1)* %x Then, eventually: getelementptr float, ptr addrspace(1) %x Importantly, the massive amount of test case churn has been automated by same crappy python code. I had to manually update a few test cases that wouldn't fit the script's model (r228970,r229196,r229197,r229198). The python script just massages stdin and writes the result to stdout, I then wrapped that in a shell script to handle replacing files, then using the usual find+xargs to migrate all the files. update.py: import fileinput import sys import re ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") def conv(match, line): if not match: return line line = match.groups()[0] if len(match.groups()[5]) == 0: line += match.groups()[2] line += match.groups()[3] line += ", " line += match.groups()[1] line += "\n" return line for line in sys.stdin: if line.find("getelementptr ") == line.find("getelementptr inbounds"): if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("): line = conv(re.match(ibrep, line), line) elif line.find("getelementptr ") != line.find("getelementptr ("): line = conv(re.match(normrep, line), line) sys.stdout.write(line) apply.sh: for name in "$@" do python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name" rm -f "$name.tmp" done The actual commands: From llvm/src: find test/ -name *.ll | xargs ./apply.sh From llvm/src/tools/clang: find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}" From llvm/src/tools/polly: find test/ -name *.ll | xargs ./apply.sh After that, check-all (with llvm, clang, clang-tools-extra, lld, compiler-rt, and polly all checked out). The extra 'rm' in the apply.sh script is due to a few files in clang's test suite using interesting unicode stuff that my python script was throwing exceptions on. None of those files needed to be migrated, so it seemed sufficient to ignore those cases. Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7636 llvm-svn: 230786
2015-02-28 03:29:02 +08:00
%18 = getelementptr inbounds [32 x [32 x float]], [32 x [32 x float]] addrspace(3)* @array, i64 0, i64 %13, i64 %7
Add an optimization that does CSE in a group of similar GEPs. This optimization merges the common part of a group of GEPs, so we can compute each pointer address by adding a simple offset to the common part. The optimization is currently only enabled for the NVPTX backend, where it has a large payoff on some benchmarks. Review: http://reviews.llvm.org/D3462 Patch by Jingyue Wu. llvm-svn: 207783
2014-05-02 02:38:36 +08:00
%19 = addrspacecast float addrspace(3)* %18 to float*
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%20 = load float, float* %19, align 4
Add an optimization that does CSE in a group of similar GEPs. This optimization merges the common part of a group of GEPs, so we can compute each pointer address by adding a simple offset to the common part. The optimization is currently only enabled for the NVPTX backend, where it has a large payoff on some benchmarks. Review: http://reviews.llvm.org/D3462 Patch by Jingyue Wu. llvm-svn: 207783
2014-05-02 02:38:36 +08:00
%21 = fadd float %17, %20
store float %21, float* %output, align 4
ret void
}
; PTX-LABEL: sum_of_array(
[NVPTX] Rename registers %fl -> %fd and %rl -> %rd This matches the internal behavior of NVIDIA tools like libnvvm. llvm-svn: 213168
2014-07-17 00:26:58 +08:00
; PTX: ld.shared.f32 {{%f[0-9]+}}, {{\[}}[[BASE_REG:%(rd|r)[0-9]+]]{{\]}}
Add an optimization that does CSE in a group of similar GEPs. This optimization merges the common part of a group of GEPs, so we can compute each pointer address by adding a simple offset to the common part. The optimization is currently only enabled for the NVPTX backend, where it has a large payoff on some benchmarks. Review: http://reviews.llvm.org/D3462 Patch by Jingyue Wu. llvm-svn: 207783
2014-05-02 02:38:36 +08:00
; PTX: ld.shared.f32 {{%f[0-9]+}}, {{\[}}[[BASE_REG]]+4{{\]}}
; PTX: ld.shared.f32 {{%f[0-9]+}}, {{\[}}[[BASE_REG]]+128{{\]}}
; PTX: ld.shared.f32 {{%f[0-9]+}}, {{\[}}[[BASE_REG]]+132{{\]}}
; IR-LABEL: @sum_of_array(
[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
; IR: [[BASE_PTR:%[a-zA-Z0-9]+]] = getelementptr inbounds [32 x [32 x float]], [32 x [32 x float]] addrspace(3)* @array, i64 0, i64 %{{[a-zA-Z0-9]+}}, i64 %{{[a-zA-Z0-9]+}}
; IR: getelementptr float, float addrspace(3)* [[BASE_PTR]], i64 1
; IR: getelementptr float, float addrspace(3)* [[BASE_PTR]], i64 32
; IR: getelementptr float, float addrspace(3)* [[BASE_PTR]], i64 33
Fixed several correctness issues in SeparateConstOffsetFromGEP Most issues are on mishandling s/zext. Fixes: 1. When rebuilding new indices, s/zext should be distributed to sub-expressions. e.g., sext(a +nsw (b +nsw 5)) = sext(a) + sext(b) + 5 but not sext(a + b) + 5. This also affects the logic of recursively looking for a constant offset, we need to include s/zext into the context of the searching. 2. Function find should return the bitwidth of the constant offset instead of always sign-extending it to i64. 3. Stop shortcutting zext'ed GEP indices. LLVM conceptually sign-extends GEP indices to pointer-size before computing the address. Therefore, gep base, zext(a + b) != gep base, a + b Improvements: 1. Add an optimization for splitting sext(a + b): if a + b is proven non-negative (e.g., used as an index of an inbound GEP) and one of a, b is non-negative, sext(a + b) = sext(a) + sext(b) 2. Function Distributable checks whether both sext and zext can be distributed to operands of a binary operator. This helps us split zext(sext(a + b)) to zext(sext(a) + zext(sext(b)) when a + b does not signed or unsigned overflow. Refactoring: Merge some common logic of handling add/sub/or in find. Testing: Add many tests in split-gep.ll and split-gep-and-gvn.ll to verify the changes we made. llvm-svn: 210291
2014-06-06 06:07:33 +08:00
; @sum_of_array2 is very similar to @sum_of_array. The only difference is in
; the order of "sext" and "add" when computing the array indices. @sum_of_array
; computes add before sext, e.g., array[sext(x + 1)][sext(y + 1)], while
; @sum_of_array2 computes sext before add,
; e.g., array[sext(x) + 1][sext(y) + 1]. SeparateConstOffsetFromGEP should be
; able to extract constant offsets from both forms.
define void @sum_of_array2(i32 %x, i32 %y, float* nocapture %output) {
.preheader:
%0 = sext i32 %y to i64
%1 = sext i32 %x to i64
[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
%2 = getelementptr inbounds [32 x [32 x float]], [32 x [32 x float]] addrspace(3)* @array, i64 0, i64 %1, i64 %0
Fixed several correctness issues in SeparateConstOffsetFromGEP Most issues are on mishandling s/zext. Fixes: 1. When rebuilding new indices, s/zext should be distributed to sub-expressions. e.g., sext(a +nsw (b +nsw 5)) = sext(a) + sext(b) + 5 but not sext(a + b) + 5. This also affects the logic of recursively looking for a constant offset, we need to include s/zext into the context of the searching. 2. Function find should return the bitwidth of the constant offset instead of always sign-extending it to i64. 3. Stop shortcutting zext'ed GEP indices. LLVM conceptually sign-extends GEP indices to pointer-size before computing the address. Therefore, gep base, zext(a + b) != gep base, a + b Improvements: 1. Add an optimization for splitting sext(a + b): if a + b is proven non-negative (e.g., used as an index of an inbound GEP) and one of a, b is non-negative, sext(a + b) = sext(a) + sext(b) 2. Function Distributable checks whether both sext and zext can be distributed to operands of a binary operator. This helps us split zext(sext(a + b)) to zext(sext(a) + zext(sext(b)) when a + b does not signed or unsigned overflow. Refactoring: Merge some common logic of handling add/sub/or in find. Testing: Add many tests in split-gep.ll and split-gep-and-gvn.ll to verify the changes we made. llvm-svn: 210291
2014-06-06 06:07:33 +08:00
%3 = addrspacecast float addrspace(3)* %2 to float*
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%4 = load float, float* %3, align 4
Fixed several correctness issues in SeparateConstOffsetFromGEP Most issues are on mishandling s/zext. Fixes: 1. When rebuilding new indices, s/zext should be distributed to sub-expressions. e.g., sext(a +nsw (b +nsw 5)) = sext(a) + sext(b) + 5 but not sext(a + b) + 5. This also affects the logic of recursively looking for a constant offset, we need to include s/zext into the context of the searching. 2. Function find should return the bitwidth of the constant offset instead of always sign-extending it to i64. 3. Stop shortcutting zext'ed GEP indices. LLVM conceptually sign-extends GEP indices to pointer-size before computing the address. Therefore, gep base, zext(a + b) != gep base, a + b Improvements: 1. Add an optimization for splitting sext(a + b): if a + b is proven non-negative (e.g., used as an index of an inbound GEP) and one of a, b is non-negative, sext(a + b) = sext(a) + sext(b) 2. Function Distributable checks whether both sext and zext can be distributed to operands of a binary operator. This helps us split zext(sext(a + b)) to zext(sext(a) + zext(sext(b)) when a + b does not signed or unsigned overflow. Refactoring: Merge some common logic of handling add/sub/or in find. Testing: Add many tests in split-gep.ll and split-gep-and-gvn.ll to verify the changes we made. llvm-svn: 210291
2014-06-06 06:07:33 +08:00
%5 = fadd float %4, 0.000000e+00
%6 = add i64 %0, 1
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction One of several parallel first steps to remove the target type of pointers, replacing them with a single opaque pointer type. This adds an explicit type parameter to the gep instruction so that when the first parameter becomes an opaque pointer type, the type to gep through is still available to the instructions. * This doesn't modify gep operators, only instructions (operators will be handled separately) * Textual IR changes only. Bitcode (including upgrade) and changing the in-memory representation will be in separate changes. * geps of vectors are transformed as: getelementptr <4 x float*> %x, ... ->getelementptr float, <4 x float*> %x, ... Then, once the opaque pointer type is introduced, this will ultimately look like: getelementptr float, <4 x ptr> %x with the unambiguous interpretation that it is a vector of pointers to float. * address spaces remain on the pointer, not the type: getelementptr float addrspace(1)* %x ->getelementptr float, float addrspace(1)* %x Then, eventually: getelementptr float, ptr addrspace(1) %x Importantly, the massive amount of test case churn has been automated by same crappy python code. I had to manually update a few test cases that wouldn't fit the script's model (r228970,r229196,r229197,r229198). The python script just massages stdin and writes the result to stdout, I then wrapped that in a shell script to handle replacing files, then using the usual find+xargs to migrate all the files. update.py: import fileinput import sys import re ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") def conv(match, line): if not match: return line line = match.groups()[0] if len(match.groups()[5]) == 0: line += match.groups()[2] line += match.groups()[3] line += ", " line += match.groups()[1] line += "\n" return line for line in sys.stdin: if line.find("getelementptr ") == line.find("getelementptr inbounds"): if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("): line = conv(re.match(ibrep, line), line) elif line.find("getelementptr ") != line.find("getelementptr ("): line = conv(re.match(normrep, line), line) sys.stdout.write(line) apply.sh: for name in "$@" do python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name" rm -f "$name.tmp" done The actual commands: From llvm/src: find test/ -name *.ll | xargs ./apply.sh From llvm/src/tools/clang: find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}" From llvm/src/tools/polly: find test/ -name *.ll | xargs ./apply.sh After that, check-all (with llvm, clang, clang-tools-extra, lld, compiler-rt, and polly all checked out). The extra 'rm' in the apply.sh script is due to a few files in clang's test suite using interesting unicode stuff that my python script was throwing exceptions on. None of those files needed to be migrated, so it seemed sufficient to ignore those cases. Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7636 llvm-svn: 230786
2015-02-28 03:29:02 +08:00
%7 = getelementptr inbounds [32 x [32 x float]], [32 x [32 x float]] addrspace(3)* @array, i64 0, i64 %1, i64 %6
Fixed several correctness issues in SeparateConstOffsetFromGEP Most issues are on mishandling s/zext. Fixes: 1. When rebuilding new indices, s/zext should be distributed to sub-expressions. e.g., sext(a +nsw (b +nsw 5)) = sext(a) + sext(b) + 5 but not sext(a + b) + 5. This also affects the logic of recursively looking for a constant offset, we need to include s/zext into the context of the searching. 2. Function find should return the bitwidth of the constant offset instead of always sign-extending it to i64. 3. Stop shortcutting zext'ed GEP indices. LLVM conceptually sign-extends GEP indices to pointer-size before computing the address. Therefore, gep base, zext(a + b) != gep base, a + b Improvements: 1. Add an optimization for splitting sext(a + b): if a + b is proven non-negative (e.g., used as an index of an inbound GEP) and one of a, b is non-negative, sext(a + b) = sext(a) + sext(b) 2. Function Distributable checks whether both sext and zext can be distributed to operands of a binary operator. This helps us split zext(sext(a + b)) to zext(sext(a) + zext(sext(b)) when a + b does not signed or unsigned overflow. Refactoring: Merge some common logic of handling add/sub/or in find. Testing: Add many tests in split-gep.ll and split-gep-and-gvn.ll to verify the changes we made. llvm-svn: 210291
2014-06-06 06:07:33 +08:00
%8 = addrspacecast float addrspace(3)* %7 to float*
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%9 = load float, float* %8, align 4
Fixed several correctness issues in SeparateConstOffsetFromGEP Most issues are on mishandling s/zext. Fixes: 1. When rebuilding new indices, s/zext should be distributed to sub-expressions. e.g., sext(a +nsw (b +nsw 5)) = sext(a) + sext(b) + 5 but not sext(a + b) + 5. This also affects the logic of recursively looking for a constant offset, we need to include s/zext into the context of the searching. 2. Function find should return the bitwidth of the constant offset instead of always sign-extending it to i64. 3. Stop shortcutting zext'ed GEP indices. LLVM conceptually sign-extends GEP indices to pointer-size before computing the address. Therefore, gep base, zext(a + b) != gep base, a + b Improvements: 1. Add an optimization for splitting sext(a + b): if a + b is proven non-negative (e.g., used as an index of an inbound GEP) and one of a, b is non-negative, sext(a + b) = sext(a) + sext(b) 2. Function Distributable checks whether both sext and zext can be distributed to operands of a binary operator. This helps us split zext(sext(a + b)) to zext(sext(a) + zext(sext(b)) when a + b does not signed or unsigned overflow. Refactoring: Merge some common logic of handling add/sub/or in find. Testing: Add many tests in split-gep.ll and split-gep-and-gvn.ll to verify the changes we made. llvm-svn: 210291
2014-06-06 06:07:33 +08:00
%10 = fadd float %5, %9
%11 = add i64 %1, 1
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction One of several parallel first steps to remove the target type of pointers, replacing them with a single opaque pointer type. This adds an explicit type parameter to the gep instruction so that when the first parameter becomes an opaque pointer type, the type to gep through is still available to the instructions. * This doesn't modify gep operators, only instructions (operators will be handled separately) * Textual IR changes only. Bitcode (including upgrade) and changing the in-memory representation will be in separate changes. * geps of vectors are transformed as: getelementptr <4 x float*> %x, ... ->getelementptr float, <4 x float*> %x, ... Then, once the opaque pointer type is introduced, this will ultimately look like: getelementptr float, <4 x ptr> %x with the unambiguous interpretation that it is a vector of pointers to float. * address spaces remain on the pointer, not the type: getelementptr float addrspace(1)* %x ->getelementptr float, float addrspace(1)* %x Then, eventually: getelementptr float, ptr addrspace(1) %x Importantly, the massive amount of test case churn has been automated by same crappy python code. I had to manually update a few test cases that wouldn't fit the script's model (r228970,r229196,r229197,r229198). The python script just massages stdin and writes the result to stdout, I then wrapped that in a shell script to handle replacing files, then using the usual find+xargs to migrate all the files. update.py: import fileinput import sys import re ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") def conv(match, line): if not match: return line line = match.groups()[0] if len(match.groups()[5]) == 0: line += match.groups()[2] line += match.groups()[3] line += ", " line += match.groups()[1] line += "\n" return line for line in sys.stdin: if line.find("getelementptr ") == line.find("getelementptr inbounds"): if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("): line = conv(re.match(ibrep, line), line) elif line.find("getelementptr ") != line.find("getelementptr ("): line = conv(re.match(normrep, line), line) sys.stdout.write(line) apply.sh: for name in "$@" do python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name" rm -f "$name.tmp" done The actual commands: From llvm/src: find test/ -name *.ll | xargs ./apply.sh From llvm/src/tools/clang: find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}" From llvm/src/tools/polly: find test/ -name *.ll | xargs ./apply.sh After that, check-all (with llvm, clang, clang-tools-extra, lld, compiler-rt, and polly all checked out). The extra 'rm' in the apply.sh script is due to a few files in clang's test suite using interesting unicode stuff that my python script was throwing exceptions on. None of those files needed to be migrated, so it seemed sufficient to ignore those cases. Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7636 llvm-svn: 230786
2015-02-28 03:29:02 +08:00
%12 = getelementptr inbounds [32 x [32 x float]], [32 x [32 x float]] addrspace(3)* @array, i64 0, i64 %11, i64 %0
Fixed several correctness issues in SeparateConstOffsetFromGEP Most issues are on mishandling s/zext. Fixes: 1. When rebuilding new indices, s/zext should be distributed to sub-expressions. e.g., sext(a +nsw (b +nsw 5)) = sext(a) + sext(b) + 5 but not sext(a + b) + 5. This also affects the logic of recursively looking for a constant offset, we need to include s/zext into the context of the searching. 2. Function find should return the bitwidth of the constant offset instead of always sign-extending it to i64. 3. Stop shortcutting zext'ed GEP indices. LLVM conceptually sign-extends GEP indices to pointer-size before computing the address. Therefore, gep base, zext(a + b) != gep base, a + b Improvements: 1. Add an optimization for splitting sext(a + b): if a + b is proven non-negative (e.g., used as an index of an inbound GEP) and one of a, b is non-negative, sext(a + b) = sext(a) + sext(b) 2. Function Distributable checks whether both sext and zext can be distributed to operands of a binary operator. This helps us split zext(sext(a + b)) to zext(sext(a) + zext(sext(b)) when a + b does not signed or unsigned overflow. Refactoring: Merge some common logic of handling add/sub/or in find. Testing: Add many tests in split-gep.ll and split-gep-and-gvn.ll to verify the changes we made. llvm-svn: 210291
2014-06-06 06:07:33 +08:00
%13 = addrspacecast float addrspace(3)* %12 to float*
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%14 = load float, float* %13, align 4
Fixed several correctness issues in SeparateConstOffsetFromGEP Most issues are on mishandling s/zext. Fixes: 1. When rebuilding new indices, s/zext should be distributed to sub-expressions. e.g., sext(a +nsw (b +nsw 5)) = sext(a) + sext(b) + 5 but not sext(a + b) + 5. This also affects the logic of recursively looking for a constant offset, we need to include s/zext into the context of the searching. 2. Function find should return the bitwidth of the constant offset instead of always sign-extending it to i64. 3. Stop shortcutting zext'ed GEP indices. LLVM conceptually sign-extends GEP indices to pointer-size before computing the address. Therefore, gep base, zext(a + b) != gep base, a + b Improvements: 1. Add an optimization for splitting sext(a + b): if a + b is proven non-negative (e.g., used as an index of an inbound GEP) and one of a, b is non-negative, sext(a + b) = sext(a) + sext(b) 2. Function Distributable checks whether both sext and zext can be distributed to operands of a binary operator. This helps us split zext(sext(a + b)) to zext(sext(a) + zext(sext(b)) when a + b does not signed or unsigned overflow. Refactoring: Merge some common logic of handling add/sub/or in find. Testing: Add many tests in split-gep.ll and split-gep-and-gvn.ll to verify the changes we made. llvm-svn: 210291
2014-06-06 06:07:33 +08:00
%15 = fadd float %10, %14
[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
%16 = getelementptr inbounds [32 x [32 x float]], [32 x [32 x float]] addrspace(3)* @array, i64 0, i64 %11, i64 %6
Fixed several correctness issues in SeparateConstOffsetFromGEP Most issues are on mishandling s/zext. Fixes: 1. When rebuilding new indices, s/zext should be distributed to sub-expressions. e.g., sext(a +nsw (b +nsw 5)) = sext(a) + sext(b) + 5 but not sext(a + b) + 5. This also affects the logic of recursively looking for a constant offset, we need to include s/zext into the context of the searching. 2. Function find should return the bitwidth of the constant offset instead of always sign-extending it to i64. 3. Stop shortcutting zext'ed GEP indices. LLVM conceptually sign-extends GEP indices to pointer-size before computing the address. Therefore, gep base, zext(a + b) != gep base, a + b Improvements: 1. Add an optimization for splitting sext(a + b): if a + b is proven non-negative (e.g., used as an index of an inbound GEP) and one of a, b is non-negative, sext(a + b) = sext(a) + sext(b) 2. Function Distributable checks whether both sext and zext can be distributed to operands of a binary operator. This helps us split zext(sext(a + b)) to zext(sext(a) + zext(sext(b)) when a + b does not signed or unsigned overflow. Refactoring: Merge some common logic of handling add/sub/or in find. Testing: Add many tests in split-gep.ll and split-gep-and-gvn.ll to verify the changes we made. llvm-svn: 210291
2014-06-06 06:07:33 +08:00
%17 = addrspacecast float addrspace(3)* %16 to float*
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%18 = load float, float* %17, align 4
Fixed several correctness issues in SeparateConstOffsetFromGEP Most issues are on mishandling s/zext. Fixes: 1. When rebuilding new indices, s/zext should be distributed to sub-expressions. e.g., sext(a +nsw (b +nsw 5)) = sext(a) + sext(b) + 5 but not sext(a + b) + 5. This also affects the logic of recursively looking for a constant offset, we need to include s/zext into the context of the searching. 2. Function find should return the bitwidth of the constant offset instead of always sign-extending it to i64. 3. Stop shortcutting zext'ed GEP indices. LLVM conceptually sign-extends GEP indices to pointer-size before computing the address. Therefore, gep base, zext(a + b) != gep base, a + b Improvements: 1. Add an optimization for splitting sext(a + b): if a + b is proven non-negative (e.g., used as an index of an inbound GEP) and one of a, b is non-negative, sext(a + b) = sext(a) + sext(b) 2. Function Distributable checks whether both sext and zext can be distributed to operands of a binary operator. This helps us split zext(sext(a + b)) to zext(sext(a) + zext(sext(b)) when a + b does not signed or unsigned overflow. Refactoring: Merge some common logic of handling add/sub/or in find. Testing: Add many tests in split-gep.ll and split-gep-and-gvn.ll to verify the changes we made. llvm-svn: 210291
2014-06-06 06:07:33 +08:00
%19 = fadd float %15, %18
store float %19, float* %output, align 4
ret void
}
; PTX-LABEL: sum_of_array2(
[NVPTX] Rename registers %fl -> %fd and %rl -> %rd This matches the internal behavior of NVIDIA tools like libnvvm. llvm-svn: 213168
2014-07-17 00:26:58 +08:00
; PTX: ld.shared.f32 {{%f[0-9]+}}, {{\[}}[[BASE_REG:%(rd|r)[0-9]+]]{{\]}}
Fixed several correctness issues in SeparateConstOffsetFromGEP Most issues are on mishandling s/zext. Fixes: 1. When rebuilding new indices, s/zext should be distributed to sub-expressions. e.g., sext(a +nsw (b +nsw 5)) = sext(a) + sext(b) + 5 but not sext(a + b) + 5. This also affects the logic of recursively looking for a constant offset, we need to include s/zext into the context of the searching. 2. Function find should return the bitwidth of the constant offset instead of always sign-extending it to i64. 3. Stop shortcutting zext'ed GEP indices. LLVM conceptually sign-extends GEP indices to pointer-size before computing the address. Therefore, gep base, zext(a + b) != gep base, a + b Improvements: 1. Add an optimization for splitting sext(a + b): if a + b is proven non-negative (e.g., used as an index of an inbound GEP) and one of a, b is non-negative, sext(a + b) = sext(a) + sext(b) 2. Function Distributable checks whether both sext and zext can be distributed to operands of a binary operator. This helps us split zext(sext(a + b)) to zext(sext(a) + zext(sext(b)) when a + b does not signed or unsigned overflow. Refactoring: Merge some common logic of handling add/sub/or in find. Testing: Add many tests in split-gep.ll and split-gep-and-gvn.ll to verify the changes we made. llvm-svn: 210291
2014-06-06 06:07:33 +08:00
; PTX: ld.shared.f32 {{%f[0-9]+}}, {{\[}}[[BASE_REG]]+4{{\]}}
; PTX: ld.shared.f32 {{%f[0-9]+}}, {{\[}}[[BASE_REG]]+128{{\]}}
; PTX: ld.shared.f32 {{%f[0-9]+}}, {{\[}}[[BASE_REG]]+132{{\]}}
; IR-LABEL: @sum_of_array2(
[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
; IR: [[BASE_PTR:%[a-zA-Z0-9]+]] = getelementptr inbounds [32 x [32 x float]], [32 x [32 x float]] addrspace(3)* @array, i64 0, i64 %{{[a-zA-Z0-9]+}}, i64 %{{[a-zA-Z0-9]+}}
; IR: getelementptr float, float addrspace(3)* [[BASE_PTR]], i64 1
; IR: getelementptr float, float addrspace(3)* [[BASE_PTR]], i64 32
; IR: getelementptr float, float addrspace(3)* [[BASE_PTR]], i64 33
[SeparateConstOffsetFromGEP] inbounds zext => sext for better splitting For each array index that is in the form of zext(a), convert it to sext(a) if we can prove zext(a) <= max signed value of typeof(a). The conversion helps to split zext(x + y) into sext(x) + sext(y). Reviewed in http://reviews.llvm.org/D4060 llvm-svn: 210444
2014-06-09 07:49:34 +08:00
Partially revert r210444 due to performance regression Summary: Converting outermost zext(a) to sext(a) causes worse code when the computation of zext(a) could be reused. For example, after converting ... = array[zext(a)] ... = array[zext(a) + 1] to ... = array[sext(a)] ... = array[zext(a) + 1], the program computes sext(a), which is actually unnecessary. I added one test in split-gep-and-gvn.ll to illustrate this scenario. Also, with r211281 and r211084, we annotate more "nuw" tags to computation involving CUDA intrinsics such as threadIdx.x. These annotations help with splitting GEP a lot, rendering the benefit we get from this reverted optimization only marginal. Test Plan: make check-all Reviewers: eliben, meheff Reviewed By: meheff Subscribers: jholewinski, llvm-commits Differential Revision: http://reviews.llvm.org/D4542 llvm-svn: 213209
2014-07-17 07:25:00 +08:00
; This function loads
; array[zext(x)][zext(y)]
; array[zext(x)][zext(y +nuw 1)]
; array[zext(x +nuw 1)][zext(y)]
; array[zext(x +nuw 1)][zext(y +nuw 1)].
;
; This function is similar to @sum_of_array, but it
; 1) extends array indices using zext instead of sext;
; 2) annotates the addition with "nuw"; otherwise, zext(x + 1) => zext(x) + 1
; may be invalid.
[SeparateConstOffsetFromGEP] inbounds zext => sext for better splitting For each array index that is in the form of zext(a), convert it to sext(a) if we can prove zext(a) <= max signed value of typeof(a). The conversion helps to split zext(x + y) into sext(x) + sext(y). Reviewed in http://reviews.llvm.org/D4060 llvm-svn: 210444
2014-06-09 07:49:34 +08:00
define void @sum_of_array3(i32 %x, i32 %y, float* nocapture %output) {
.preheader:
%0 = zext i32 %y to i64
%1 = zext i32 %x to i64
[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
%2 = getelementptr inbounds [32 x [32 x float]], [32 x [32 x float]] addrspace(3)* @array, i64 0, i64 %1, i64 %0
[SeparateConstOffsetFromGEP] inbounds zext => sext for better splitting For each array index that is in the form of zext(a), convert it to sext(a) if we can prove zext(a) <= max signed value of typeof(a). The conversion helps to split zext(x + y) into sext(x) + sext(y). Reviewed in http://reviews.llvm.org/D4060 llvm-svn: 210444
2014-06-09 07:49:34 +08:00
%3 = addrspacecast float addrspace(3)* %2 to float*
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%4 = load float, float* %3, align 4
[SeparateConstOffsetFromGEP] inbounds zext => sext for better splitting For each array index that is in the form of zext(a), convert it to sext(a) if we can prove zext(a) <= max signed value of typeof(a). The conversion helps to split zext(x + y) into sext(x) + sext(y). Reviewed in http://reviews.llvm.org/D4060 llvm-svn: 210444
2014-06-09 07:49:34 +08:00
%5 = fadd float %4, 0.000000e+00
Partially revert r210444 due to performance regression Summary: Converting outermost zext(a) to sext(a) causes worse code when the computation of zext(a) could be reused. For example, after converting ... = array[zext(a)] ... = array[zext(a) + 1] to ... = array[sext(a)] ... = array[zext(a) + 1], the program computes sext(a), which is actually unnecessary. I added one test in split-gep-and-gvn.ll to illustrate this scenario. Also, with r211281 and r211084, we annotate more "nuw" tags to computation involving CUDA intrinsics such as threadIdx.x. These annotations help with splitting GEP a lot, rendering the benefit we get from this reverted optimization only marginal. Test Plan: make check-all Reviewers: eliben, meheff Reviewed By: meheff Subscribers: jholewinski, llvm-commits Differential Revision: http://reviews.llvm.org/D4542 llvm-svn: 213209
2014-07-17 07:25:00 +08:00
%6 = add nuw i32 %y, 1
[SeparateConstOffsetFromGEP] inbounds zext => sext for better splitting For each array index that is in the form of zext(a), convert it to sext(a) if we can prove zext(a) <= max signed value of typeof(a). The conversion helps to split zext(x + y) into sext(x) + sext(y). Reviewed in http://reviews.llvm.org/D4060 llvm-svn: 210444
2014-06-09 07:49:34 +08:00
%7 = zext i32 %6 to i64
[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
%8 = getelementptr inbounds [32 x [32 x float]], [32 x [32 x float]] addrspace(3)* @array, i64 0, i64 %1, i64 %7
[SeparateConstOffsetFromGEP] inbounds zext => sext for better splitting For each array index that is in the form of zext(a), convert it to sext(a) if we can prove zext(a) <= max signed value of typeof(a). The conversion helps to split zext(x + y) into sext(x) + sext(y). Reviewed in http://reviews.llvm.org/D4060 llvm-svn: 210444
2014-06-09 07:49:34 +08:00
%9 = addrspacecast float addrspace(3)* %8 to float*
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%10 = load float, float* %9, align 4
[SeparateConstOffsetFromGEP] inbounds zext => sext for better splitting For each array index that is in the form of zext(a), convert it to sext(a) if we can prove zext(a) <= max signed value of typeof(a). The conversion helps to split zext(x + y) into sext(x) + sext(y). Reviewed in http://reviews.llvm.org/D4060 llvm-svn: 210444
2014-06-09 07:49:34 +08:00
%11 = fadd float %5, %10
Partially revert r210444 due to performance regression Summary: Converting outermost zext(a) to sext(a) causes worse code when the computation of zext(a) could be reused. For example, after converting ... = array[zext(a)] ... = array[zext(a) + 1] to ... = array[sext(a)] ... = array[zext(a) + 1], the program computes sext(a), which is actually unnecessary. I added one test in split-gep-and-gvn.ll to illustrate this scenario. Also, with r211281 and r211084, we annotate more "nuw" tags to computation involving CUDA intrinsics such as threadIdx.x. These annotations help with splitting GEP a lot, rendering the benefit we get from this reverted optimization only marginal. Test Plan: make check-all Reviewers: eliben, meheff Reviewed By: meheff Subscribers: jholewinski, llvm-commits Differential Revision: http://reviews.llvm.org/D4542 llvm-svn: 213209
2014-07-17 07:25:00 +08:00
%12 = add nuw i32 %x, 1
[SeparateConstOffsetFromGEP] inbounds zext => sext for better splitting For each array index that is in the form of zext(a), convert it to sext(a) if we can prove zext(a) <= max signed value of typeof(a). The conversion helps to split zext(x + y) into sext(x) + sext(y). Reviewed in http://reviews.llvm.org/D4060 llvm-svn: 210444
2014-06-09 07:49:34 +08:00
%13 = zext i32 %12 to i64
[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
%14 = getelementptr inbounds [32 x [32 x float]], [32 x [32 x float]] addrspace(3)* @array, i64 0, i64 %13, i64 %0
[SeparateConstOffsetFromGEP] inbounds zext => sext for better splitting For each array index that is in the form of zext(a), convert it to sext(a) if we can prove zext(a) <= max signed value of typeof(a). The conversion helps to split zext(x + y) into sext(x) + sext(y). Reviewed in http://reviews.llvm.org/D4060 llvm-svn: 210444
2014-06-09 07:49:34 +08:00
%15 = addrspacecast float addrspace(3)* %14 to float*
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%16 = load float, float* %15, align 4
[SeparateConstOffsetFromGEP] inbounds zext => sext for better splitting For each array index that is in the form of zext(a), convert it to sext(a) if we can prove zext(a) <= max signed value of typeof(a). The conversion helps to split zext(x + y) into sext(x) + sext(y). Reviewed in http://reviews.llvm.org/D4060 llvm-svn: 210444
2014-06-09 07:49:34 +08:00
%17 = fadd float %11, %16
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction One of several parallel first steps to remove the target type of pointers, replacing them with a single opaque pointer type. This adds an explicit type parameter to the gep instruction so that when the first parameter becomes an opaque pointer type, the type to gep through is still available to the instructions. * This doesn't modify gep operators, only instructions (operators will be handled separately) * Textual IR changes only. Bitcode (including upgrade) and changing the in-memory representation will be in separate changes. * geps of vectors are transformed as: getelementptr <4 x float*> %x, ... ->getelementptr float, <4 x float*> %x, ... Then, once the opaque pointer type is introduced, this will ultimately look like: getelementptr float, <4 x ptr> %x with the unambiguous interpretation that it is a vector of pointers to float. * address spaces remain on the pointer, not the type: getelementptr float addrspace(1)* %x ->getelementptr float, float addrspace(1)* %x Then, eventually: getelementptr float, ptr addrspace(1) %x Importantly, the massive amount of test case churn has been automated by same crappy python code. I had to manually update a few test cases that wouldn't fit the script's model (r228970,r229196,r229197,r229198). The python script just massages stdin and writes the result to stdout, I then wrapped that in a shell script to handle replacing files, then using the usual find+xargs to migrate all the files. update.py: import fileinput import sys import re ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") def conv(match, line): if not match: return line line = match.groups()[0] if len(match.groups()[5]) == 0: line += match.groups()[2] line += match.groups()[3] line += ", " line += match.groups()[1] line += "\n" return line for line in sys.stdin: if line.find("getelementptr ") == line.find("getelementptr inbounds"): if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("): line = conv(re.match(ibrep, line), line) elif line.find("getelementptr ") != line.find("getelementptr ("): line = conv(re.match(normrep, line), line) sys.stdout.write(line) apply.sh: for name in "$@" do python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name" rm -f "$name.tmp" done The actual commands: From llvm/src: find test/ -name *.ll | xargs ./apply.sh From llvm/src/tools/clang: find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}" From llvm/src/tools/polly: find test/ -name *.ll | xargs ./apply.sh After that, check-all (with llvm, clang, clang-tools-extra, lld, compiler-rt, and polly all checked out). The extra 'rm' in the apply.sh script is due to a few files in clang's test suite using interesting unicode stuff that my python script was throwing exceptions on. None of those files needed to be migrated, so it seemed sufficient to ignore those cases. Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7636 llvm-svn: 230786
2015-02-28 03:29:02 +08:00
%18 = getelementptr inbounds [32 x [32 x float]], [32 x [32 x float]] addrspace(3)* @array, i64 0, i64 %13, i64 %7
[SeparateConstOffsetFromGEP] inbounds zext => sext for better splitting For each array index that is in the form of zext(a), convert it to sext(a) if we can prove zext(a) <= max signed value of typeof(a). The conversion helps to split zext(x + y) into sext(x) + sext(y). Reviewed in http://reviews.llvm.org/D4060 llvm-svn: 210444
2014-06-09 07:49:34 +08:00
%19 = addrspacecast float addrspace(3)* %18 to float*
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%20 = load float, float* %19, align 4
[SeparateConstOffsetFromGEP] inbounds zext => sext for better splitting For each array index that is in the form of zext(a), convert it to sext(a) if we can prove zext(a) <= max signed value of typeof(a). The conversion helps to split zext(x + y) into sext(x) + sext(y). Reviewed in http://reviews.llvm.org/D4060 llvm-svn: 210444
2014-06-09 07:49:34 +08:00
%21 = fadd float %17, %20
store float %21, float* %output, align 4
ret void
}
; PTX-LABEL: sum_of_array3(
[NVPTX] Rename registers %fl -> %fd and %rl -> %rd This matches the internal behavior of NVIDIA tools like libnvvm. llvm-svn: 213168
2014-07-17 00:26:58 +08:00
; PTX: ld.shared.f32 {{%f[0-9]+}}, {{\[}}[[BASE_REG:%(rd|r)[0-9]+]]{{\]}}
[SeparateConstOffsetFromGEP] inbounds zext => sext for better splitting For each array index that is in the form of zext(a), convert it to sext(a) if we can prove zext(a) <= max signed value of typeof(a). The conversion helps to split zext(x + y) into sext(x) + sext(y). Reviewed in http://reviews.llvm.org/D4060 llvm-svn: 210444
2014-06-09 07:49:34 +08:00
; PTX: ld.shared.f32 {{%f[0-9]+}}, {{\[}}[[BASE_REG]]+4{{\]}}
; PTX: ld.shared.f32 {{%f[0-9]+}}, {{\[}}[[BASE_REG]]+128{{\]}}
; PTX: ld.shared.f32 {{%f[0-9]+}}, {{\[}}[[BASE_REG]]+132{{\]}}
; IR-LABEL: @sum_of_array3(
[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
; IR: [[BASE_PTR:%[a-zA-Z0-9]+]] = getelementptr inbounds [32 x [32 x float]], [32 x [32 x float]] addrspace(3)* @array, i64 0, i64 %{{[a-zA-Z0-9]+}}, i64 %{{[a-zA-Z0-9]+}}
; IR: getelementptr float, float addrspace(3)* [[BASE_PTR]], i64 1
; IR: getelementptr float, float addrspace(3)* [[BASE_PTR]], i64 32
; IR: getelementptr float, float addrspace(3)* [[BASE_PTR]], i64 33
Partially revert r210444 due to performance regression Summary: Converting outermost zext(a) to sext(a) causes worse code when the computation of zext(a) could be reused. For example, after converting ... = array[zext(a)] ... = array[zext(a) + 1] to ... = array[sext(a)] ... = array[zext(a) + 1], the program computes sext(a), which is actually unnecessary. I added one test in split-gep-and-gvn.ll to illustrate this scenario. Also, with r211281 and r211084, we annotate more "nuw" tags to computation involving CUDA intrinsics such as threadIdx.x. These annotations help with splitting GEP a lot, rendering the benefit we get from this reverted optimization only marginal. Test Plan: make check-all Reviewers: eliben, meheff Reviewed By: meheff Subscribers: jholewinski, llvm-commits Differential Revision: http://reviews.llvm.org/D4542 llvm-svn: 213209
2014-07-17 07:25:00 +08:00
; This function loads
; array[zext(x)][zext(y)]
; array[zext(x)][zext(y)]
; array[zext(x) + 1][zext(y) + 1]
; array[zext(x) + 1][zext(y) + 1].
;
; We expect the generated code to reuse the computation of
; &array[zext(x)][zext(y)]. See the expected IR and PTX for details.
define void @sum_of_array4(i32 %x, i32 %y, float* nocapture %output) {
.preheader:
%0 = zext i32 %y to i64
%1 = zext i32 %x to i64
[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
%2 = getelementptr inbounds [32 x [32 x float]], [32 x [32 x float]] addrspace(3)* @array, i64 0, i64 %1, i64 %0
Partially revert r210444 due to performance regression Summary: Converting outermost zext(a) to sext(a) causes worse code when the computation of zext(a) could be reused. For example, after converting ... = array[zext(a)] ... = array[zext(a) + 1] to ... = array[sext(a)] ... = array[zext(a) + 1], the program computes sext(a), which is actually unnecessary. I added one test in split-gep-and-gvn.ll to illustrate this scenario. Also, with r211281 and r211084, we annotate more "nuw" tags to computation involving CUDA intrinsics such as threadIdx.x. These annotations help with splitting GEP a lot, rendering the benefit we get from this reverted optimization only marginal. Test Plan: make check-all Reviewers: eliben, meheff Reviewed By: meheff Subscribers: jholewinski, llvm-commits Differential Revision: http://reviews.llvm.org/D4542 llvm-svn: 213209
2014-07-17 07:25:00 +08:00
%3 = addrspacecast float addrspace(3)* %2 to float*
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%4 = load float, float* %3, align 4
Partially revert r210444 due to performance regression Summary: Converting outermost zext(a) to sext(a) causes worse code when the computation of zext(a) could be reused. For example, after converting ... = array[zext(a)] ... = array[zext(a) + 1] to ... = array[sext(a)] ... = array[zext(a) + 1], the program computes sext(a), which is actually unnecessary. I added one test in split-gep-and-gvn.ll to illustrate this scenario. Also, with r211281 and r211084, we annotate more "nuw" tags to computation involving CUDA intrinsics such as threadIdx.x. These annotations help with splitting GEP a lot, rendering the benefit we get from this reverted optimization only marginal. Test Plan: make check-all Reviewers: eliben, meheff Reviewed By: meheff Subscribers: jholewinski, llvm-commits Differential Revision: http://reviews.llvm.org/D4542 llvm-svn: 213209
2014-07-17 07:25:00 +08:00
%5 = fadd float %4, 0.000000e+00
%6 = add i64 %0, 1
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction One of several parallel first steps to remove the target type of pointers, replacing them with a single opaque pointer type. This adds an explicit type parameter to the gep instruction so that when the first parameter becomes an opaque pointer type, the type to gep through is still available to the instructions. * This doesn't modify gep operators, only instructions (operators will be handled separately) * Textual IR changes only. Bitcode (including upgrade) and changing the in-memory representation will be in separate changes. * geps of vectors are transformed as: getelementptr <4 x float*> %x, ... ->getelementptr float, <4 x float*> %x, ... Then, once the opaque pointer type is introduced, this will ultimately look like: getelementptr float, <4 x ptr> %x with the unambiguous interpretation that it is a vector of pointers to float. * address spaces remain on the pointer, not the type: getelementptr float addrspace(1)* %x ->getelementptr float, float addrspace(1)* %x Then, eventually: getelementptr float, ptr addrspace(1) %x Importantly, the massive amount of test case churn has been automated by same crappy python code. I had to manually update a few test cases that wouldn't fit the script's model (r228970,r229196,r229197,r229198). The python script just massages stdin and writes the result to stdout, I then wrapped that in a shell script to handle replacing files, then using the usual find+xargs to migrate all the files. update.py: import fileinput import sys import re ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") def conv(match, line): if not match: return line line = match.groups()[0] if len(match.groups()[5]) == 0: line += match.groups()[2] line += match.groups()[3] line += ", " line += match.groups()[1] line += "\n" return line for line in sys.stdin: if line.find("getelementptr ") == line.find("getelementptr inbounds"): if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("): line = conv(re.match(ibrep, line), line) elif line.find("getelementptr ") != line.find("getelementptr ("): line = conv(re.match(normrep, line), line) sys.stdout.write(line) apply.sh: for name in "$@" do python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name" rm -f "$name.tmp" done The actual commands: From llvm/src: find test/ -name *.ll | xargs ./apply.sh From llvm/src/tools/clang: find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}" From llvm/src/tools/polly: find test/ -name *.ll | xargs ./apply.sh After that, check-all (with llvm, clang, clang-tools-extra, lld, compiler-rt, and polly all checked out). The extra 'rm' in the apply.sh script is due to a few files in clang's test suite using interesting unicode stuff that my python script was throwing exceptions on. None of those files needed to be migrated, so it seemed sufficient to ignore those cases. Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7636 llvm-svn: 230786
2015-02-28 03:29:02 +08:00
%7 = getelementptr inbounds [32 x [32 x float]], [32 x [32 x float]] addrspace(3)* @array, i64 0, i64 %1, i64 %6
Partially revert r210444 due to performance regression Summary: Converting outermost zext(a) to sext(a) causes worse code when the computation of zext(a) could be reused. For example, after converting ... = array[zext(a)] ... = array[zext(a) + 1] to ... = array[sext(a)] ... = array[zext(a) + 1], the program computes sext(a), which is actually unnecessary. I added one test in split-gep-and-gvn.ll to illustrate this scenario. Also, with r211281 and r211084, we annotate more "nuw" tags to computation involving CUDA intrinsics such as threadIdx.x. These annotations help with splitting GEP a lot, rendering the benefit we get from this reverted optimization only marginal. Test Plan: make check-all Reviewers: eliben, meheff Reviewed By: meheff Subscribers: jholewinski, llvm-commits Differential Revision: http://reviews.llvm.org/D4542 llvm-svn: 213209
2014-07-17 07:25:00 +08:00
%8 = addrspacecast float addrspace(3)* %7 to float*
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%9 = load float, float* %8, align 4
Partially revert r210444 due to performance regression Summary: Converting outermost zext(a) to sext(a) causes worse code when the computation of zext(a) could be reused. For example, after converting ... = array[zext(a)] ... = array[zext(a) + 1] to ... = array[sext(a)] ... = array[zext(a) + 1], the program computes sext(a), which is actually unnecessary. I added one test in split-gep-and-gvn.ll to illustrate this scenario. Also, with r211281 and r211084, we annotate more "nuw" tags to computation involving CUDA intrinsics such as threadIdx.x. These annotations help with splitting GEP a lot, rendering the benefit we get from this reverted optimization only marginal. Test Plan: make check-all Reviewers: eliben, meheff Reviewed By: meheff Subscribers: jholewinski, llvm-commits Differential Revision: http://reviews.llvm.org/D4542 llvm-svn: 213209
2014-07-17 07:25:00 +08:00
%10 = fadd float %5, %9
%11 = add i64 %1, 1
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction One of several parallel first steps to remove the target type of pointers, replacing them with a single opaque pointer type. This adds an explicit type parameter to the gep instruction so that when the first parameter becomes an opaque pointer type, the type to gep through is still available to the instructions. * This doesn't modify gep operators, only instructions (operators will be handled separately) * Textual IR changes only. Bitcode (including upgrade) and changing the in-memory representation will be in separate changes. * geps of vectors are transformed as: getelementptr <4 x float*> %x, ... ->getelementptr float, <4 x float*> %x, ... Then, once the opaque pointer type is introduced, this will ultimately look like: getelementptr float, <4 x ptr> %x with the unambiguous interpretation that it is a vector of pointers to float. * address spaces remain on the pointer, not the type: getelementptr float addrspace(1)* %x ->getelementptr float, float addrspace(1)* %x Then, eventually: getelementptr float, ptr addrspace(1) %x Importantly, the massive amount of test case churn has been automated by same crappy python code. I had to manually update a few test cases that wouldn't fit the script's model (r228970,r229196,r229197,r229198). The python script just massages stdin and writes the result to stdout, I then wrapped that in a shell script to handle replacing files, then using the usual find+xargs to migrate all the files. update.py: import fileinput import sys import re ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") def conv(match, line): if not match: return line line = match.groups()[0] if len(match.groups()[5]) == 0: line += match.groups()[2] line += match.groups()[3] line += ", " line += match.groups()[1] line += "\n" return line for line in sys.stdin: if line.find("getelementptr ") == line.find("getelementptr inbounds"): if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("): line = conv(re.match(ibrep, line), line) elif line.find("getelementptr ") != line.find("getelementptr ("): line = conv(re.match(normrep, line), line) sys.stdout.write(line) apply.sh: for name in "$@" do python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name" rm -f "$name.tmp" done The actual commands: From llvm/src: find test/ -name *.ll | xargs ./apply.sh From llvm/src/tools/clang: find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}" From llvm/src/tools/polly: find test/ -name *.ll | xargs ./apply.sh After that, check-all (with llvm, clang, clang-tools-extra, lld, compiler-rt, and polly all checked out). The extra 'rm' in the apply.sh script is due to a few files in clang's test suite using interesting unicode stuff that my python script was throwing exceptions on. None of those files needed to be migrated, so it seemed sufficient to ignore those cases. Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7636 llvm-svn: 230786
2015-02-28 03:29:02 +08:00
%12 = getelementptr inbounds [32 x [32 x float]], [32 x [32 x float]] addrspace(3)* @array, i64 0, i64 %11, i64 %0
Partially revert r210444 due to performance regression Summary: Converting outermost zext(a) to sext(a) causes worse code when the computation of zext(a) could be reused. For example, after converting ... = array[zext(a)] ... = array[zext(a) + 1] to ... = array[sext(a)] ... = array[zext(a) + 1], the program computes sext(a), which is actually unnecessary. I added one test in split-gep-and-gvn.ll to illustrate this scenario. Also, with r211281 and r211084, we annotate more "nuw" tags to computation involving CUDA intrinsics such as threadIdx.x. These annotations help with splitting GEP a lot, rendering the benefit we get from this reverted optimization only marginal. Test Plan: make check-all Reviewers: eliben, meheff Reviewed By: meheff Subscribers: jholewinski, llvm-commits Differential Revision: http://reviews.llvm.org/D4542 llvm-svn: 213209
2014-07-17 07:25:00 +08:00
%13 = addrspacecast float addrspace(3)* %12 to float*
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%14 = load float, float* %13, align 4
Partially revert r210444 due to performance regression Summary: Converting outermost zext(a) to sext(a) causes worse code when the computation of zext(a) could be reused. For example, after converting ... = array[zext(a)] ... = array[zext(a) + 1] to ... = array[sext(a)] ... = array[zext(a) + 1], the program computes sext(a), which is actually unnecessary. I added one test in split-gep-and-gvn.ll to illustrate this scenario. Also, with r211281 and r211084, we annotate more "nuw" tags to computation involving CUDA intrinsics such as threadIdx.x. These annotations help with splitting GEP a lot, rendering the benefit we get from this reverted optimization only marginal. Test Plan: make check-all Reviewers: eliben, meheff Reviewed By: meheff Subscribers: jholewinski, llvm-commits Differential Revision: http://reviews.llvm.org/D4542 llvm-svn: 213209
2014-07-17 07:25:00 +08:00
%15 = fadd float %10, %14
[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
%16 = getelementptr inbounds [32 x [32 x float]], [32 x [32 x float]] addrspace(3)* @array, i64 0, i64 %11, i64 %6
Partially revert r210444 due to performance regression Summary: Converting outermost zext(a) to sext(a) causes worse code when the computation of zext(a) could be reused. For example, after converting ... = array[zext(a)] ... = array[zext(a) + 1] to ... = array[sext(a)] ... = array[zext(a) + 1], the program computes sext(a), which is actually unnecessary. I added one test in split-gep-and-gvn.ll to illustrate this scenario. Also, with r211281 and r211084, we annotate more "nuw" tags to computation involving CUDA intrinsics such as threadIdx.x. These annotations help with splitting GEP a lot, rendering the benefit we get from this reverted optimization only marginal. Test Plan: make check-all Reviewers: eliben, meheff Reviewed By: meheff Subscribers: jholewinski, llvm-commits Differential Revision: http://reviews.llvm.org/D4542 llvm-svn: 213209
2014-07-17 07:25:00 +08:00
%17 = addrspacecast float addrspace(3)* %16 to float*
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%18 = load float, float* %17, align 4
Partially revert r210444 due to performance regression Summary: Converting outermost zext(a) to sext(a) causes worse code when the computation of zext(a) could be reused. For example, after converting ... = array[zext(a)] ... = array[zext(a) + 1] to ... = array[sext(a)] ... = array[zext(a) + 1], the program computes sext(a), which is actually unnecessary. I added one test in split-gep-and-gvn.ll to illustrate this scenario. Also, with r211281 and r211084, we annotate more "nuw" tags to computation involving CUDA intrinsics such as threadIdx.x. These annotations help with splitting GEP a lot, rendering the benefit we get from this reverted optimization only marginal. Test Plan: make check-all Reviewers: eliben, meheff Reviewed By: meheff Subscribers: jholewinski, llvm-commits Differential Revision: http://reviews.llvm.org/D4542 llvm-svn: 213209
2014-07-17 07:25:00 +08:00
%19 = fadd float %15, %18
store float %19, float* %output, align 4
ret void
}
; PTX-LABEL: sum_of_array4(
; PTX: ld.shared.f32 {{%f[0-9]+}}, {{\[}}[[BASE_REG:%(rd|r)[0-9]+]]{{\]}}
; PTX: ld.shared.f32 {{%f[0-9]+}}, {{\[}}[[BASE_REG]]+4{{\]}}
; PTX: ld.shared.f32 {{%f[0-9]+}}, {{\[}}[[BASE_REG]]+128{{\]}}
; PTX: ld.shared.f32 {{%f[0-9]+}}, {{\[}}[[BASE_REG]]+132{{\]}}
; IR-LABEL: @sum_of_array4(
[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
; IR: [[BASE_PTR:%[a-zA-Z0-9]+]] = getelementptr inbounds [32 x [32 x float]], [32 x [32 x float]] addrspace(3)* @array, i64 0, i64 %{{[a-zA-Z0-9]+}}, i64 %{{[a-zA-Z0-9]+}}
; IR: getelementptr float, float addrspace(3)* [[BASE_PTR]], i64 1
; IR: getelementptr float, float addrspace(3)* [[BASE_PTR]], i64 32
; IR: getelementptr float, float addrspace(3)* [[BASE_PTR]], i64 33