diff options
| author | Jakob Stoklund Olesen <stoklund@2pi.dk> | 2012-08-09 22:49:46 +0000 |
|---|---|---|
| committer | Jakob Stoklund Olesen <stoklund@2pi.dk> | 2012-08-09 22:49:46 +0000 |
| commit | 81a6995243380668e6f991fa4e11dd0a6e37e030 (patch) | |
| tree | 44dcd3571c6e0280d0ed7fbb9047d71d703ad1a3 /test/CodeGen | |
| parent | 46f4c35372062eaf097922b5683bc6639ccf342b (diff) | |
| download | external_llvm-81a6995243380668e6f991fa4e11dd0a6e37e030.zip external_llvm-81a6995243380668e6f991fa4e11dd0a6e37e030.tar.gz external_llvm-81a6995243380668e6f991fa4e11dd0a6e37e030.tar.bz2 | |
Partition use lists so defs always come before uses.
This makes it possible to speed up def_iterator by stopping at the first
use. This makes def_empty() and getUniqueVRegDef() much faster when
there are many uses.
In a +Asserts build, LiveVariables is 100x faster in one case because
getVRegDef() has an assertion that would scan to the end of a
def_iterator chain.
Spill weight calculation is significantly faster (300x in one case)
because isTriviallyReMaterializable() calls MRI->isConstantPhysReg(%RIP)
which calls def_empty(%RIP).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161634 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test/CodeGen')
| -rw-r--r-- | test/CodeGen/X86/apm.ll | 4 | ||||
| -rw-r--r-- | test/CodeGen/X86/xor.ll | 2 |
2 files changed, 3 insertions, 3 deletions
diff --git a/test/CodeGen/X86/apm.ll b/test/CodeGen/X86/apm.ll index 9f4b0f4..aaedf18 100644 --- a/test/CodeGen/X86/apm.ll +++ b/test/CodeGen/X86/apm.ll @@ -3,8 +3,8 @@ ; PR8573 ; CHECK: foo: -; CHECK: movl %esi, %ecx -; CHECK-NEXT: leaq (%rdi), %rax +; CHECK: leaq (%rdi), %rax +; CHECK-NEXT: movl %esi, %ecx ; CHECK-NEXT: monitor ; WIN64: foo: ; WIN64: leaq (%rcx), %rax diff --git a/test/CodeGen/X86/xor.ll b/test/CodeGen/X86/xor.ll index 73416ef..996bfc4 100644 --- a/test/CodeGen/X86/xor.ll +++ b/test/CodeGen/X86/xor.ll @@ -31,7 +31,7 @@ entry: ; X64: test3: ; X64: notl ; X64: andl -; X64: shrl %eax +; X64: shrl ; X64: ret ; X32: test3: |