[llvm-commits] [llvm] r158175 - in /llvm/trunk: lib/Target/X86/X86ISelLowering.cpp test/CodeGen/X86/iabs.ll

[Joerg Sonnenberger]

On Thu, Jun 07, 2012 at 10:39:11PM -0000, Manman Ren wrote:
> Log:
> X86: optimize generated code for integer ABS
> 
> This patch will generate the following for integer ABS:
>       movl    %edi, %eax
>       negl    %eax
>       cmovll  %edi, %eax
> INSTEAD OF
>       movl    %edi, %ecx
>       sarl    $31, %ecx
>       leal    (%rdi,%rcx), %eax
>       xorl    %ecx, %eax

Does this work correctly (e.g. not do the transform) on CPUs without
CMOV support?

Joerg