Formalize the notion that AVX and SSE are non-overlapping extensions from the compile...
authorNate Begeman <natebegeman@mac.com>
Fri, 10 Dec 2010 00:26:57 +0000 (00:26 +0000)
committerNate Begeman <natebegeman@mac.com>
Fri, 10 Dec 2010 00:26:57 +0000 (00:26 +0000)
commit2ea8ee7c76b8d8754d81072e691caf25d23289e3
treeb3d1820d852b71777b8db8f87672b044555d6e5c
parent1c952b9cc98e84b28f68f0f6cf11197263f89863
Formalize the notion that AVX and SSE are non-overlapping extensions from the compiler's point of view.  Per email discussion, we either want to always use VEX-prefixed instructions or never use them, and are taking "HasAVX" to mean "Always use VEX".  Passing -mattr=-avx,+sse42 should serve to restore legacy SSE support when desirable.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121439 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Target/X86/X86.td
lib/Target/X86/X86CallingConv.td
lib/Target/X86/X86ISelLowering.cpp
lib/Target/X86/X86InstrInfo.td
lib/Target/X86/X86InstrSSE.td
lib/Target/X86/X86Subtarget.cpp
lib/Target/X86/X86Subtarget.h
test/CodeGen/X86/avx-128.ll