projects / oota-llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: d026168) | patch
NVPTX: support direct f16 <-> f64 conversions via intrinsics.
authorTim Northover <tnorthover@apple.com>
Fri, 18 Jul 2014 08:30:10 +0000 (08:30 +0000)
committerTim Northover <tnorthover@apple.com>
Fri, 18 Jul 2014 08:30:10 +0000 (08:30 +0000)
commit7bbf5786d76b3d862e30938ce56d3ea7a814e8a3
treeded84b6015d7d75dd1a6583973dae00867f34e27tree | snapshot
parentd0261683e8768df25e1c9e8c9cbe46abd74db737commit | diff
NVPTX: support direct f16 <-> f64 conversions via intrinsics.

Clang may well start emitting these soon, and while it may not be
directly relevant for OpenCL or GLSL, the instructions were just
sitting there waiting to be used.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213356 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Target/NVPTX/NVPTXIntrinsics.td diff | blob | history
test/CodeGen/NVPTX/fp16.ll [new file with mode: 0644] blob
C/C++ LLVM-based compilers that forbids OOTA behaviors