P4 to Git Change 1177889 by weizhang@msdnweizhang-opencl on 2015/08/06 17:16:54

EPR #421017 - IOMMU2/SVM on CZ Win10, the bit INST_ATC of  COMPUTE_PGM_HI needs to be set for device enqueue.

Affected files ...

... //depot/stg/opencl/drivers/opencl/library/hsa/hsail/src/devenq/schedule.cl#9 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpusched.hpp#16 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuvirtual.cpp#370 edit


[ROCm/clr commit: 7a54d367f3]

projects/clr/rocclr/runtime/device/gpu/gpusched.hpp

@@ -62,7 +62,7 @@ struct SchedulerParam {
     uint32_t    eng_clk;        //!< Engine clock in Mhz
     uint64_t    hw_queue;       //!< Address to HW queue
     uint64_t    hsa_queue;      //!< Address to HSA dummy queue
-    uint32_t    reserved;       //!< Reserved
+    uint32_t    useATC;         //!< GPU access to shader program by ATC.
     uint32_t    scratchSize;    //!< Scratch buffer size
     uint64_t    scratch;        //!< GPU address to the scratch buffer
     uint32_t    numMaxWaves;    //!< The max number of possible waves

projects/clr/rocclr/runtime/device/gpu/gpuvirtual.cpp

@@ -1926,6 +1926,7 @@ VirtualGPU::submitKernelInternalHSA(
             param->releaseHostCP = 0;
             param->parentAQL = vmParentWrap;
             param->dedicatedQueue = dev().settings().useDeviceQueue_;
+            param->useATC = dev().settings().svmFineGrainSystem_;
 
             // Fill the scratch buffer information
             if (hsaKernel.prog().maxScratchRegs() > 0) {