P4 to Git Change 1546657 by gandryey@gera-w8 on 2018/04/26 10:59:34

SWDEV-151739 - [CQE OCL][DTB][Perf][QR][DTB-BLOCKER][VEGA10] Upto 18% performance drop observed while running Video Composition test sub test of Compubench due to faulty CL#1544622
	- Implement customized TS tracking for managed buffers. The common TS tracking mechanism saves the event of the last command, assuming SDMA and compute operations occur in order, but for managed buffers it's not the case. Also managed buffer doesn't have to validate TS for the parent resource.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palblit.cpp#21 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palconstbuf.cpp#11 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palconstbuf.hpp#9 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palmemory.hpp#6 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palresource.hpp#22 edit


[ROCm/clr commit: 79ba5904dc]

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

@@ -267,7 +267,7 @@ bool DmaBlitManager::writeMemoryStaged(const void* srcHost, Memory& dstMemory, M
     chunkSize = gpu().xferWrite().MaxSize();
   } else {
     chunkSize = std::min(amd::alignUp(xferSize / 4, 256), gpu().xferWrite().MaxSize());
-    chunkSize = std::max(chunkSize, 128 * Ki);
+    chunkSize = std::max(chunkSize, 64 * Ki);
     bool flushDMA = true;
   }