[llvm-commits] [llvm] r140332 - in /llvm/trunk: lib/Target/X86/X86ISelLowering.cpp lib/Target/X86/X86ISelLowering.h lib/Target/X86/X86InstrFragmentsSIMD.td lib/Target/X86/X86InstrSSE.td test/CodeGen/X86/haddsub.ll

[Duncan Sands]

Hi Chris,

>> Synthesize SSE3/AVX 128 bit horizontal add/sub instructions from
>> floating point add/sub of appropriate shuffle vectors.  Does not
>> synthesize the 256 bit AVX versions because they work differently.
>
> Very cool Duncan.  Would it make sense to legalize the intrinsics for these into the X86ISD nodes to avoid the duplicate patterns?

good idea!  I implemented this in commit 140383.

Ciao, Duncan.