P4 to Git Change 1524559 by asalmanp@asalmanp-ocl-stg on 2018/03/08 14:18:24

SWDEV-132899 - [gfx10][OCL]- Adding support for forcing WaveSize32 from runtime for testing on gfx10 HW emulator

	Motivation: During testing ocltst on Windows on PAL/HSAIL/SC path on gfx10 HW emulator, it was found that SC uses WaveSize64 by default for compute kernels.
	SC also has an interface that can be used for forcing the WaveSize to 32 or 64.

	- Adding the "-force-wave-size-32" into compiler to be passed down to Finalizer/SC
	- Adding environment variable "GPU_FORCE_WAVE_SIZE_32" that can be used from runtime to force WaveSize32 compilation in HSAIL/SC path

	ReviewBoardURL = http://ocltc.amd.com/reviews/r/14364/

Affected files ...

... //depot/stg/opencl/drivers/opencl/compiler/lib/backends/gpu/hsail_be.cpp#69 edit
... //depot/stg/opencl/drivers/opencl/compiler/lib/utils/OPTIONS.def#138 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palprogram.cpp#55 edit
... //depot/stg/opencl/drivers/opencl/runtime/utils/flags.hpp#284 edit


[ROCm/clr commit: 3c0e80074b]

projects/clr/rocclr/runtime/device/pal/palprogram.cpp

@@ -598,6 +598,10 @@ bool HSAILProgram::linkImpl(amd::option::Options* options) {
     if (dev().settings().svmFineGrainSystem_) {
       fin_options.append(" -sc-xnack-iommu");
     }
+    if (dev().settings().gfx10Plus_ && GPU_FORCE_WAVE_SIZE_32) {
+      fin_options.append(" -force-wave-size-32");
+    }
+
     errorCode = aclCompile(dev().compiler(), binaryElf_, fin_options.c_str(), ACL_TYPE_CG,
                            ACL_TYPE_ISA, nullptr);
     buildLog_ += aclGetCompilerLog(dev().compiler());