]> granicus.if.org Git - clang/commit
[NVPTX, CUDA] Added support for m8n32k16 and m32n8k16 variants of wmma instructions.
authorArtem Belevich <tra@google.com>
Wed, 18 Apr 2018 21:51:48 +0000 (21:51 +0000)
committerArtem Belevich <tra@google.com>
Wed, 18 Apr 2018 21:51:48 +0000 (21:51 +0000)
commitc4d3d32435b0867c3e518acd4004249b7cd63b71
treee6c8aaed8323b701be0dcebf0c7d3e703558b368
parent202f222f94f9ca45d6c6896ba397c4646d7f99cb
[NVPTX, CUDA] Added support for m8n32k16 and m32n8k16 variants of wmma instructions.

The new instructions were added added for sm_70+ GPUs in CUDA-9.1.

Differential Revision: https://reviews.llvm.org/D45068

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@330296 91177308-0d34-0410-b5e6-96231b3b80d8
include/clang/Basic/BuiltinsNVPTX.def
lib/CodeGen/CGBuiltin.cpp
lib/Driver/ToolChains/Cuda.cpp
test/CodeGen/builtins-nvptx-sm_70.cu