P4 to Git Change 1201490 by xcui@merged_opencl_jxcwin on 2015/10/20 10:47:18

SWDEV-41018 -  if system is equal to or less than 2GB memory, disable CPU and APU for OpenCL, and force the device to be 1.2 if it is a discrete GPU.

	precheckin:
	http://ocltc.amd.com:8111/viewModification.html?modId=60516&personal=true&buildTypeId=&tab=vcsModificationBuilds&show_all_builds=true
	code review:
	http://ocltc.amd.com/reviews/r/8738/

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/device/cpu/cpudevice.cpp#277 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpudevice.cpp#531 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpusettings.cpp#331 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpusettings.hpp#95 edit

rocclr/runtime/device/cpu/cpudevice.cpp

@@ -235,6 +235,10 @@ Device::init()
     virtualMemSize =
         (uintptr_t) std::min(statex.ullTotalPageFile, statex.ullTotalVirtual);
 #endif
+    //disable CPU device if system memory is equal to or less than 2GB
+    if (info.globalMemSize_ <= 2 * Gi) {
+        return true;
+    }
 
     maxWorkerThreads_ = (size_t) (virtualMemSize /
         (uintptr_t) ((CPU_WORKER_THREAD_STACK_SIZE +

rocclr/runtime/device/gpu/gpudevice.cpp

@@ -762,9 +762,14 @@ Device::create(CALuint ordinal, CALuint numOfDevices)
 {
     appProfile_.init();
 
+    bool smallMemSystem = false;
+    if (amd::Os::hostTotalPhysicalMemory() < 2 * Gi) {
+        smallMemSystem = true;
+    }
+
     // Open GSL device
     if (!open(ordinal, appProfile_.enableHighPerformanceState(),
-        appProfile_.reportAsOCL12Device() || (OPENCL_VERSION < 200))) {
+        smallMemSystem || appProfile_.reportAsOCL12Device() || (OPENCL_VERSION < 200))) {
         return false;
     }
 
@@ -776,7 +781,7 @@ Device::create(CALuint ordinal, CALuint numOfDevices)
     settings_ = new gpu::Settings();
     gpu::Settings* gpuSettings = reinterpret_cast<gpu::Settings*>(settings_);
     if ((gpuSettings == NULL) || !gpuSettings->create(getAttribs()
-          , appProfile_.reportAsOCL12Device()
+        , appProfile_.reportAsOCL12Device(), smallMemSystem
         )) {
         return false;
     }

rocclr/runtime/device/gpu/gpusettings.cpp

@@ -146,6 +146,7 @@ bool
 Settings::create(
     const CALdeviceattribs& calAttr
   , bool reportAsOCL12Device
+  , bool smallMemSystem
 )
 {
     CALuint target = calAttr.target;
@@ -228,6 +229,14 @@ Settings::create(
                 oclVersion_ = !reportAsOCL12Device && calAttr.isOpenCL200Device ?
                     XCONCAT(OpenCL, XCONCAT(OPENCL_MAJOR, OPENCL_MINOR)) : OpenCL12;
             }
+            if (smallMemSystem) {         //force the dGPU to be 1.2 device for small memory system.
+                if (apuSystem_) {
+                    return false;
+                }
+                else {
+                    oclVersion_ = OpenCL12;
+                }
+            }
             if (GPU_FORCE_OCL20_32BIT) {
                 force32BitOcl20_ = true;
                 oclVersion_ = !reportAsOCL12Device && calAttr.isOpenCL200Device ?

rocclr/runtime/device/gpu/gpusettings.hpp

@@ -110,6 +110,7 @@ public:
     bool create(
         const CALdeviceattribs& calAttr //!< CAL attributes structure
       , bool reportAsOCL12Device = false //!< Report As OpenCL1.2 Device
+      , bool smallMemSystem = false //!< report the sys memory is small
         );
 
 private: