projects / oota-llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: f85959e) | patch
R600: Permute operands when selecting legacy min/max
authorMatt Arsenault <Matthew.Arsenault@amd.com>
Sat, 15 Nov 2014 05:02:57 +0000 (05:02 +0000)
committerMatt Arsenault <Matthew.Arsenault@amd.com>
Sat, 15 Nov 2014 05:02:57 +0000 (05:02 +0000)
commitc093062447591991e3eacedaa111bb04b0f89fcb
treeac7791e93cab7dc3af648c35cf950a099494bd98tree | snapshot
parentf85959ebc34c0b7bd6df422b28d9ee342c72118fcommit | diff
R600: Permute operands when selecting legacy min/max

This gets the correct NaN behavior based on the compare type
the hardware uses. This now passes the new piglit test I have
for this on SI.

Add stricter tests for the operand order.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222079 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Target/R600/AMDGPUISelLowering.cpp diff | blob | history
test/CodeGen/R600/fmax_legacy.ll diff | blob | history
test/CodeGen/R600/fmin_legacy.ll diff | blob | history
test/CodeGen/R600/pv.ll diff | blob | history
C/C++ LLVM-based compilers that forbids OOTA behaviors