Stephan Herhut 545c3e489f Port mlir-cuda-runner to use dialect conversion framework. ... Instead of lowering the program in two steps (Standard->LLVM followed by GPU->NVVM), leading to invalid IR inbetween, the runner now uses one pattern based rewrite step to go directly from Standard+GPU to LLVM+NVVM. PiperOrigin-RevId: 265861934		2019-08-28 01:50:57 -07:00
..
mlir-cpu-runner	Move shared cpu runner library to Support/JitRunner.	2019-07-16 13:45:16 -07:00
mlir-cuda-runner	Port mlir-cuda-runner to use dialect conversion framework.	2019-08-28 01:50:57 -07:00
mlir-opt	InitLLVM already initializes PrettyStackTraceProgram	2019-08-18 11:32:52 -07:00
mlir-tblgen	NFC: Avoid reconstructing the OpInterface methods.	2019-08-22 11:31:27 -07:00
mlir-translate	InitLLVM already initializes PrettyStackTraceProgram	2019-08-18 11:32:52 -07:00
CMakeLists.txt	Add an mlir-cuda-runner tool.	2019-07-04 07:53:54 -07:00