Changed wrong ROCDL instructions in GPU lowering.
authorJulian Gross <julian.gross@dfki.de>
Tue, 28 Jan 2020 10:09:13 +0000 (11:09 +0100)
committerJulian Gross <julian.gross@dfki.de>
Thu, 30 Jan 2020 14:37:00 +0000 (15:37 +0100)
commitaddc27bc437d2fb1f31d88294b227ac32be63cc5
tree9b7965a92270b00108e384ca7c524c4babf3f1d2
parent6cc6e89c11de514fc9dfff99ca093f7013003e7f
Changed wrong ROCDL instructions in GPU lowering.

Summary:
In the scope of the lowering phase from GPU to ROCDL, the intructions for the conversion patterns seems to be wrong.
According to https://github.com/ROCm-Developer-Tools/HIP/blob/master/include/hip/hcc_detail/math_fwd.h the instructions need two underscores in the beginning instead of one.

Reviewers: nicolasvasilache, herhut, rriddle

Reviewed By: herhut, rriddle

Subscribers: merge_guards_bot, mehdi_amini, rriddle, jpienaar, burmako, shauheen, antiagainst, csigg, arpith-jacob, mgester, lucyrfox, herhut, liufengdb, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D73535
mlir/lib/Conversion/GPUToROCDL/LowerGpuOpsToROCDLOps.cpp
mlir/test/Conversion/GPUToROCDL/gpu-to-rocdl.mlir