[NVPTX] Make tensor load/store intrinsics overloaded.
authorArtem Belevich <tra@google.com>
Tue, 20 Mar 2018 17:18:59 +0000 (17:18 +0000)
committerArtem Belevich <tra@google.com>
Tue, 20 Mar 2018 17:18:59 +0000 (17:18 +0000)
commit914d4babec358a845a31f30a50ee729393593f3d
treeaed987b58941c857aa3f68e4d5c81965ee8e99f6
parent3a9989361898266e5279d1d91055ff63b00d041d
[NVPTX] Make tensor load/store intrinsics overloaded.

This way we can support address-space specific variants without explicitly
encoding the space in the name of the intrinsic. Less intrinsics to deal with ->
less boilerplate.

Added a bit of tablegen magic to match/replace an intrinsics with a pointer
argument in particular address space with the space-specific instruction
variant.

Updated tests to use non-default address spaces.

Differential Revision: https://reviews.llvm.org/D43268

llvm-svn: 328006
clang/lib/CodeGen/CGBuiltin.cpp
llvm/include/llvm/IR/IntrinsicsNVVM.td
llvm/lib/Target/NVPTX/NVPTXISelLowering.cpp
llvm/lib/Target/NVPTX/NVPTXIntrinsics.td
llvm/test/CodeGen/NVPTX/wmma.py