projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: ea3a687) | patch
[ARM] Match fminnum/fmaxnum for vector vminnm/vmaxnm instead of an intrinsic
authorJames Molloy <james.molloy@arm.com>
Tue, 11 Aug 2015 12:06:25 +0000 (12:06 +0000)
committerJames Molloy <james.molloy@arm.com>
Tue, 11 Aug 2015 12:06:25 +0000 (12:06 +0000)
commitee868b2a3e5180d075e08a7466a501c1d3b98711
treeb4f9389de8970395982b8ae671e7118ca1ef5785tree | snapshot
parentea3a687a334c801ffeb7062730dfa87d1570b270commit | diff
[ARM] Match fminnum/fmaxnum for vector vminnm/vmaxnm instead of an intrinsic

Lower the intrinsic to a FMINNUM/FMAXNUM node and select that instead. This is important because soon SDAG will be able to select FMINNUM/FMAXNUM itself, so we need an integrated lowering path between SDAG and intrinsics.

NFCI.

llvm-svn: 244592
llvm/lib/Target/ARM/ARMISelLowering.cpp diff | blob | history
llvm/lib/Target/ARM/ARMInstrNEON.td diff | blob | history
Domain: System / Toolchain;