Add the "gpu.launch_func" op to the GPU dialect.
authorMLIR Team <no-reply@google.com>
Mon, 6 May 2019 12:01:12 +0000 (05:01 -0700)
committerMehdi Amini <joker.eph@gmail.com>
Mon, 6 May 2019 15:30:07 +0000 (08:30 -0700)
commit465df0d916e706cd98641d33de9911b1fe3e98f9
tree891af03e69e35840d37393bc49a95c5b0b2c93bc
parentb7b75cf74cb07897e9d3148af66a6d26f00409cd
Add the "gpu.launch_func" op to the GPU dialect.

    The idea is to lower `gpu.launch` operations into `gpu.launch_func` operations by outlining the kernel body into a function, which is closer to the NVVM model.

--

PiperOrigin-RevId: 246806890
mlir/g3doc/Dialects/GPU.md
mlir/include/mlir/GPU/GPUDialect.h
mlir/lib/GPU/IR/DialectRegistration.cpp
mlir/lib/GPU/IR/GPUDialect.cpp
mlir/test/GPU/invalid.mlir
mlir/test/GPU/ops.mlir