Tracks device where memory is allocated, pinned-host or device, and
more.
Uses memory-range-based lookups - so pointers that exist anywhere in
the range of hostPtr + size will find the associated AmPointerInfo.
The insertions and lookups use a self-balancing binary tree and
should support O(logN) lookup speed.
Few more counters are now available in GFX8 register specs. So adding
them. Also for gfx700 and gfx801 report correct number of SQ perf counter slots
Change-Id: I9e6b4b10238230aabeccbfaa5e491a28b5e54f2d
SWDEV-2 - Change OpenCL version number from 2033 to 2034.
Affected files ...
... //depot/stg/opencl/drivers/opencl/runtime/utils/versions.hpp#1780 edit
SWDEV-78299 - Back out changelist 1236441 since OCLCreateBuffer fails.
Affected files ...
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpudevice.cpp#541 edit
Device property MaxSharedMemoryPerMultiprocessor set equal to totalGlobalMem (HIP path).
Reason: MaxSharedMemoryPerMultiprocessor should be as the same as group memory size. Group memory will not be paged out, so, the physical memory size = total shared memory size = group region size. NVCC path remains untouched: CUDA's device property MaxSharedMemoryPerMultiprocessor is reported.
hipify is updated as well.
Device property MaxSharedMemoryPerMultiprocessor set equal to totalGlobalMem (HIP path).
Reason: MaxSharedMemoryPerMultiprocessor should be as the same as group memory size. Group memory will not be paged out, so, the physical memory size = total shared memory size = group region size. NVCC path remains untouched: CUDA's device property MaxSharedMemoryPerMultiprocessor is reported.
hipify is updated as well.
Reason: maxThreadsPerMultiProcessor should be as the same as group memory size. Group memory will not be paged out, so, the physical memory size = total shared memory size = group region size.
NVCC path remains untouched: CUDA's device property maxThreadsPerMultiProcessor is reported.
Reason: maxThreadsPerMultiProcessor should be as the same as group memory size. Group memory will not be paged out, so, the physical memory size = total shared memory size = group region size.
NVCC path remains untouched: CUDA's device property maxThreadsPerMultiProcessor is reported.
SWDEV-78299 - [Brahma] Setting max single allocation size by comparing cardMemAvailableBytes with cardExtMemAvailableBytes on Brahma.
Affected files ...
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpudevice.cpp#540 edit
SWDEV-2 - Change OpenCL version number from 2032 to 2033.
Affected files ...
... //depot/stg/opencl/drivers/opencl/runtime/utils/versions.hpp#1779 edit
Allocations from GPU nodes will return VRAM, not system memory.
Only non-paged allocation from GPU nodes is supported. System
memory can only be allocated from CPU nodes (usually node 0).
The HostAccess flag is no longer used to distinguish the memory
type. It only indicates, whether the memory is mapped for CPU
access.
Maintain compatibility with broken KfdTests by returning system
memory for paged-memory requested from GPU nodes.
Change-Id: I514defede735f55e6de436f41944125b6f2c4ccf
SWDEV-2 - Change OpenCL version number from 2031 to 2032.
Affected files ...
... //depot/stg/opencl/drivers/opencl/runtime/utils/versions.hpp#1778 edit
Ben Sander
Harish Kasiviswanathan
Ramesh Errabolu (xN/A) TX
Ben Goz
Kent Russell
Maneesh Gupta
foreman
Evgeny Mankov
sunway513
dfukalov
Felix Kuehling
streamhsa
Yair Shachar
gargrahul