RPCS3/llvm
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm.git synced 2024-12-27 14:45:50 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity
38c6b58eec
llvm/test/CodeGen/X86/shrink-compare.ll

92 lines
1.9 KiB
LLVM
Raw Normal View History

DAGCombine: fold "(zext x) == C" into "x == (trunc C)" if the trunc is lossless. On x86 this allows to fold a load into the cmp, greatly reducing register pressure. movzbl (%rdi), %eax cmpl $47, %eax -> cmpb $47, (%rdi) This shaves 8k off gcc.o on i386. I'll leave applying the patch in README.txt to Chris :) git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@130005 91177308-0d34-0410-b5e6-96231b3b80d8
2011-04-22 18:47:44 +00:00
; RUN: llc < %s -march=x86-64 | FileCheck %s
declare void @bar()
X86: Perform integer comparisons at i32 or larger. Utilizing the 8 and 16 bit comparison instructions, even when an input can be folded into the comparison instruction itself, is typically not worth it. There are too many partial register stalls as a result, leading to significant slowdowns. By always performing comparisons on at least 32-bit registers, performance of the calculation chain leading to the comparison improves. Continue to use the smaller comparisons when minimizing size, as that allows better folding of loads into the comparison instructions. rdar://15386341 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195496 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-22 19:57:47 +00:00
define void @test1(i32* nocapture %X) nounwind minsize {
DAGCombine: fold "(zext x) == C" into "x == (trunc C)" if the trunc is lossless. On x86 this allows to fold a load into the cmp, greatly reducing register pressure. movzbl (%rdi), %eax cmpl $47, %eax -> cmpb $47, (%rdi) This shaves 8k off gcc.o on i386. I'll leave applying the patch in README.txt to Chris :) git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@130005 91177308-0d34-0410-b5e6-96231b3b80d8
2011-04-22 18:47:44 +00:00
entry:
%tmp1 = load i32* %X, align 4
%and = and i32 %tmp1, 255
%cmp = icmp eq i32 %and, 47
br i1 %cmp, label %if.then, label %if.end
if.then:
tail call void @bar() nounwind
br label %if.end
if.end:
ret void
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@186258 91177308-0d34-0410-b5e6-96231b3b80d8
2013-07-13 20:38:47 +00:00
; CHECK-LABEL: test1:
test/CodeGen/X86/shrink-compare.ll: Relax expressions for Win64. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@130039 91177308-0d34-0410-b5e6-96231b3b80d8
2011-04-23 00:15:45 +00:00
; CHECK: cmpb $47, (%{{rdi|rcx}})
DAGCombine: fold "(zext x) == C" into "x == (trunc C)" if the trunc is lossless. On x86 this allows to fold a load into the cmp, greatly reducing register pressure. movzbl (%rdi), %eax cmpl $47, %eax -> cmpb $47, (%rdi) This shaves 8k off gcc.o on i386. I'll leave applying the patch in README.txt to Chris :) git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@130005 91177308-0d34-0410-b5e6-96231b3b80d8
2011-04-22 18:47:44 +00:00
}
X86: Perform integer comparisons at i32 or larger. Utilizing the 8 and 16 bit comparison instructions, even when an input can be folded into the comparison instruction itself, is typically not worth it. There are too many partial register stalls as a result, leading to significant slowdowns. By always performing comparisons on at least 32-bit registers, performance of the calculation chain leading to the comparison improves. Continue to use the smaller comparisons when minimizing size, as that allows better folding of loads into the comparison instructions. rdar://15386341 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195496 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-22 19:57:47 +00:00
define void @test2(i32 %X) nounwind minsize {
DAGCombine: fold "(zext x) == C" into "x == (trunc C)" if the trunc is lossless. On x86 this allows to fold a load into the cmp, greatly reducing register pressure. movzbl (%rdi), %eax cmpl $47, %eax -> cmpb $47, (%rdi) This shaves 8k off gcc.o on i386. I'll leave applying the patch in README.txt to Chris :) git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@130005 91177308-0d34-0410-b5e6-96231b3b80d8
2011-04-22 18:47:44 +00:00
entry:
%and = and i32 %X, 255
%cmp = icmp eq i32 %and, 47
br i1 %cmp, label %if.then, label %if.end
if.then:
tail call void @bar() nounwind
br label %if.end
if.end:
ret void
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@186258 91177308-0d34-0410-b5e6-96231b3b80d8
2013-07-13 20:38:47 +00:00
; CHECK-LABEL: test2:
test/CodeGen/X86/shrink-compare.ll: Relax expressions for Win64. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@130039 91177308-0d34-0410-b5e6-96231b3b80d8
2011-04-23 00:15:45 +00:00
; CHECK: cmpb $47, %{{dil|cl}}
DAGCombine: fold "(zext x) == C" into "x == (trunc C)" if the trunc is lossless. On x86 this allows to fold a load into the cmp, greatly reducing register pressure. movzbl (%rdi), %eax cmpl $47, %eax -> cmpb $47, (%rdi) This shaves 8k off gcc.o on i386. I'll leave applying the patch in README.txt to Chris :) git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@130005 91177308-0d34-0410-b5e6-96231b3b80d8
2011-04-22 18:47:44 +00:00
}
DAGCombine: Also shrink eq compares where the constant is exactly as large as the smaller type. if ((x & 255) == 255) before: movzbl %al, %eax cmpl $255, %eax after: cmpb $-1, %al git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@182038 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-16 18:47:58 +00:00
X86: Perform integer comparisons at i32 or larger. Utilizing the 8 and 16 bit comparison instructions, even when an input can be folded into the comparison instruction itself, is typically not worth it. There are too many partial register stalls as a result, leading to significant slowdowns. By always performing comparisons on at least 32-bit registers, performance of the calculation chain leading to the comparison improves. Continue to use the smaller comparisons when minimizing size, as that allows better folding of loads into the comparison instructions. rdar://15386341 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195496 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-22 19:57:47 +00:00
define void @test3(i32 %X) nounwind minsize {
DAGCombine: Also shrink eq compares where the constant is exactly as large as the smaller type. if ((x & 255) == 255) before: movzbl %al, %eax cmpl $255, %eax after: cmpb $-1, %al git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@182038 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-16 18:47:58 +00:00
entry:
%and = and i32 %X, 255
%cmp = icmp eq i32 %and, 255
br i1 %cmp, label %if.then, label %if.end
if.then:
tail call void @bar() nounwind
br label %if.end
if.end:
ret void
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@186258 91177308-0d34-0410-b5e6-96231b3b80d8
2013-07-13 20:38:47 +00:00
; CHECK-LABEL: test3:
DAGCombine: Also shrink eq compares where the constant is exactly as large as the smaller type. if ((x & 255) == 255) before: movzbl %al, %eax cmpl $255, %eax after: cmpb $-1, %al git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@182038 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-16 18:47:58 +00:00
; CHECK: cmpb $-1, %{{dil|cl}}
}
DAGCombine: Avoid an edge case where it tried to create an i0 type for (x & 0) == 0. Fixes PR16083. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@182357 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-21 08:51:09 +00:00
; PR16083
define i1 @test4(i64 %a, i32 %b) {
entry:
%tobool = icmp ne i32 %b, 0
br i1 %tobool, label %lor.end, label %lor.rhs
lor.rhs: ; preds = %entry
%and = and i64 0, %a
%tobool1 = icmp ne i64 %and, 0
br label %lor.end
lor.end: ; preds = %lor.rhs, %entry
%p = phi i1 [ true, %entry ], [ %tobool1, %lor.rhs ]
ret i1 %p
}
DAGCombiner: Don't drop extension behavior when shrinking a load when unsafe. ReduceLoadWidth unconditionally drops extensions from loads. Limit it to the case when all of the bits the extension would otherwise produce are dropped by the shrink. It would be possible to shrink the load in more cases by merging the extensions, but this isn't trivial and a very rare case. I left a TODO for that case. Fixes PR16551. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185755 91177308-0d34-0410-b5e6-96231b3b80d8
2013-07-06 14:05:09 +00:00
@x = global { i8, i8, i8, i8, i8, i8, i8, i8 } { i8 1, i8 0, i8 0, i8 0, i8 1, i8 0, i8 0, i8 1 }, align 4
; PR16551
X86: Perform integer comparisons at i32 or larger. Utilizing the 8 and 16 bit comparison instructions, even when an input can be folded into the comparison instruction itself, is typically not worth it. There are too many partial register stalls as a result, leading to significant slowdowns. By always performing comparisons on at least 32-bit registers, performance of the calculation chain leading to the comparison improves. Continue to use the smaller comparisons when minimizing size, as that allows better folding of loads into the comparison instructions. rdar://15386341 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195496 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-22 19:57:47 +00:00
define void @test5(i32 %X) nounwind minsize {
DAGCombiner: Don't drop extension behavior when shrinking a load when unsafe. ReduceLoadWidth unconditionally drops extensions from loads. Limit it to the case when all of the bits the extension would otherwise produce are dropped by the shrink. It would be possible to shrink the load in more cases by merging the extensions, but this isn't trivial and a very rare case. I left a TODO for that case. Fixes PR16551. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185755 91177308-0d34-0410-b5e6-96231b3b80d8
2013-07-06 14:05:09 +00:00
entry:
%bf.load = load i56* bitcast ({ i8, i8, i8, i8, i8, i8, i8, i8 }* @x to i56*), align 4
%bf.lshr = lshr i56 %bf.load, 32
%bf.cast = trunc i56 %bf.lshr to i32
%cmp = icmp ne i32 %bf.cast, 1
br i1 %cmp, label %if.then, label %if.end
if.then:
tail call void @bar() nounwind
br label %if.end
if.end:
ret void
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@186258 91177308-0d34-0410-b5e6-96231b3b80d8
2013-07-13 20:38:47 +00:00
; CHECK-LABEL: test5:
DAGCombiner: Don't drop extension behavior when shrinking a load when unsafe. ReduceLoadWidth unconditionally drops extensions from loads. Limit it to the case when all of the bits the extension would otherwise produce are dropped by the shrink. It would be possible to shrink the load in more cases by merging the extensions, but this isn't trivial and a very rare case. I left a TODO for that case. Fixes PR16551. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185755 91177308-0d34-0410-b5e6-96231b3b80d8
2013-07-06 14:05:09 +00:00
; CHECK-NOT: cmpl $1,{{.*}}x+4
; CHECK: ret
}
Reference in New Issue Copy Permalink