diff options
| author | Benjamin Kramer <benny.kra@googlemail.com> | 2013-07-12 18:06:44 +0000 |
|---|---|---|
| committer | Benjamin Kramer <benny.kra@googlemail.com> | 2013-07-12 18:06:44 +0000 |
| commit | b619dd5d5b69ba9f4571a96e1a96d09d8aed03a7 (patch) | |
| tree | 79411ae481a896f857d5a454961479a75289eac9 /test/CodeGen/X86/widen_conv-2.ll | |
| parent | 2494dfcf1795a9bd28030d04cf2fb15e5c3b0f33 (diff) | |
| download | llvm-b619dd5d5b69ba9f4571a96e1a96d09d8aed03a7.tar.gz llvm-b619dd5d5b69ba9f4571a96e1a96d09d8aed03a7.tar.bz2 llvm-b619dd5d5b69ba9f4571a96e1a96d09d8aed03a7.tar.xz | |
X86: Shrink certain forms of movsx.
In particular:
movsbw %al, %ax --> cbtw
movswl %ax, %eax --> cwtl
movslq %eax, %rax --> cltq
According to Intel's manual those have the same performance characteristics but
come with a smaller encoding.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@186174 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test/CodeGen/X86/widen_conv-2.ll')
| -rw-r--r-- | test/CodeGen/X86/widen_conv-2.ll | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/test/CodeGen/X86/widen_conv-2.ll b/test/CodeGen/X86/widen_conv-2.ll index 969cb512be..db8fa93dc7 100644 --- a/test/CodeGen/X86/widen_conv-2.ll +++ b/test/CodeGen/X86/widen_conv-2.ll @@ -1,6 +1,6 @@ ; RUN: llc < %s -march=x86 -mattr=+sse42 | FileCheck %s -; CHECK: movswl -; CHECK: movswl +; CHECK: cwtl +; CHECK: cwtl ; sign extension v2i32 to v2i16 |