With recent changes in ROCclr to take HSA cmake target it require
HIP to have CMAKE_PREFIX_PATH added for install location for HSA,
amd_comgr,llvm.
User can provide CMAKE_PREFIX_PATH if above
packages are built/installed at custom location as:
cmake -DCMAKE_PREFIX_PATH="<path_hsa> <path_amd_comgr> <path_llvm>"
For default ROCm installation adding a default path will help.
Change-Id: Iefa443dc198da76b57422a641c9821c63cde4922
[ROCm/hip commit: 6e50c0f755]
Cuda allows creating 2D textures from memory with unaligned size.
Aligning it will cause us to overshoot the size of the underlying
allocation.
Change-Id: Ia20dec2105e36ca4a8448c9c6299d0001b388d25
[ROCm/hip commit: 844dab3b22]
Submit explicit profile marker for hipEventRecord to record
timestamps. Enable explicit signal profiling if the API specifies
start and stop events.
Toggle this with env var HIP_FORCE_QUEUE_PROFILING=0
Change-Id: Iae449a63ec3ebf6c2880e65d7b1dd1031a29018f
[ROCm/hip commit: 2ef062480b]
Recently there were few OOM(out of Memory) issues poped. Adding trace will help know genuine OOM issue.
Change-Id: Ie2a56bba95cead24caf1af5c807e73c383c2f5d9
[ROCm/hip commit: d768406ec4]
If the queue device doesn't match the device on any memory object, use the queue device from the memory object.
Change-Id: I5fdcf00494f8391574f4443332c01788b8da44ef
[ROCm/hip commit: 678b3eeed5]
SWDEV-244798
If {src/dst} ptr is marked as hipMemoryTypeHost, check if the memory was
prepinned. If it was, upgrade the copy type to hipMemoryTypeDevice to
avoid extra pinning.
Change-Id: Id287ef5b14ae67dfbcf80c4caa1b08a311191948
[ROCm/hip commit: ed26013ec1]
Hip applications assume that hipEventRecord called from multiple
threads will contain exactly the last queued command to the stream.
Change-Id: I1da3259f143d7670d0870d9a47c08e32336b2222
[ROCm/hip commit: d491dbd796]
JIRA : SWDEV-234471
With this change compilation of the ibamdhip64 to be decied by arguments to build_hip_on_rocclr.sh file
build_hip_on_rocclr.sh -r (current) shall conitnue to create libamdhip64.so;
build_hip_on_rocclr.sh -r a (new) shall create libamdhip64.a
other options remain the same
Changed the targets according to requirement of stage2 dependencies.
Changed find_lib with find_package
Adpated the review comments
Change-Id: I719918f73ab33cd8031a32db6fe9b8adaa1b4b87
[ROCm/hip commit: ab88cbbb8a]
Host memory allocations use shared device context, which allows
to have access from any device. That breaks default queue detection.
Change-Id: I27ad1234d49ab0ea8f5edd6f2e3ffa93c47d4b1f
[ROCm/hip commit: b9b982247e]
Add hipHostMallocNumaUser flag to hipHostMalloc() in order to support
numa policy set by user.
Change-Id: I6d70ed539a5f97f27187f2242b68849c0e27e4d6
[ROCm/hip commit: 63051ca2e1]
hipChannelFormatDesc contains size of each channel in bits. There's 8 bits per byte, not 4, so we should divide by 8.
Change-Id: I38ae2d28ad54740823c98908ca2afecad0ba0183
[ROCm/hip commit: e9033edc10]
Aryan Salmanpour
Jatin Chaudhary
Todd tiantuo Li
kjayapra-amd
Freddy Paul
Jason Tang
German Andryeyev
Rahul Garg
Vlad Sytchenko
Saleel Kudchadker
Payam
Sarbojit Sarkar
Anusha Godavarthy Surya
Jason Tang
jujiang
Tao Sang
Christophe Paquot
Evgeny
Ashutosh Mishra