projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 258da8a) | patch
[X86] Rename the VEX scalar fma builtins to end with a '3' to match gcc
authorCraig Topper <craig.topper@intel.com>
Thu, 9 Nov 2017 04:10:46 +0000 (04:10 +0000)
committerCraig Topper <craig.topper@intel.com>
Thu, 9 Nov 2017 04:10:46 +0000 (04:10 +0000)
commite5b84ec2a19be202474c72f8d57ff8ac12b1e111
treeb4b7ee89107beb532b2b24a497754845c191ec58tree | snapshot
parent258da8a249e25b86e7b9ef0049b1883898a899d3commit | diff
[X86] Rename the VEX scalar fma builtins to end with a '3' to match gcc

I think we need to use different builtins for the FMA4 instructions since those instructions zero the upper bits and FMA3 instructions pass the bits through.

So this moves the existing builtins to be the FMA3 versions. New versions will be added for FMA4.

llvm-svn: 317766
clang/include/clang/Basic/BuiltinsX86.def diff | blob | history
clang/lib/Headers/fma4intrin.h diff | blob | history
clang/lib/Headers/fmaintrin.h diff | blob | history
Domain: System / Toolchain;