- If fat binary doesn't contain code object for current gfx then inside FatBinaryInfo::ExtractFatBinary(), valid FatBinaryDeviceInfo address isn't stored in vector fatbin_dev_info_
- This raises segmentation fault during hipLaunchKernel
Change-Id: I21017338d91edbd5d9cc2d37277f66558198a129
[ROCm/clr commit: 348698ed30]
Build script was passing relative path for setting the RUNPATH. But clinfo is overriding it and using absolute paths
Removed the same from source code
Change-Id: I505e7fdc84c53d147f3afbc6c4133f906de2110f
[ROCm/clr commit: 067beca608]
OS can terminate unfinished queue thread from default stream at any
time. Potentially leaving the queue lock in a bad state and causing a
deadlock if runtime destroys VirtualGPU later from the host thread.
Change-Id: I247f102ee84e6b4dba947504933395071945c85d
[ROCm/clr commit: 28daf98f1f]
This reverts commit de1e612775.
Reason for revert: Need to wait for apps to make fixes first
Change-Id: I337a0b56d268093fc8689152002a6f8a8ceb258f
[ROCm/clr commit: c7576e2add]
This reverts commit dfa7790030.
Reason for revert: Deferred to a future release.
Change-Id: Ia66c37f0ab9734dee73c930d10d7469d5fd57254
[ROCm/clr commit: 5dc104b3ea]
This reverts commit e9ebd8b072.
Reason for revert: Need app fixes first
Change-Id: I1fea9a2ca8d57ef79a4f407bbc0d906976292eb1
[ROCm/clr commit: 2801be2372]
Remove alignment to granularity for IPC handles as ROCr has a patch that
will internally validate pointer sizes against requested size during
allocation instead of size aligned to page size. This patch is needed
together with this patch from ROCr:
f8a42a3a:Use user requested size for memory fragments
Change-Id: I28b25558ea03c836b44fafdb34b7330cf6887424
[ROCm/clr commit: fa1b7925cf]
Newer GCC's seem to require this.
Change-Id: I85926d4fa552b772f2eb9f8ede7863a546c47f54
Signed-off-by: Jeremy Newton <Jeremy.Newton@amd.com>
[ROCm/clr commit: 70bdb7a597]
This reverts commit 8da846db0a.
Reason for revert: Test failures with Luxmark, blender, and Indigobench. Need to investigate before re-applying
Change-Id: I6b08273a8f9c8bcaa4e7a06cd42d15048e52ca2a
[ROCm/clr commit: 5168485d23]
Heap initialization used device queue, but it shoudl be used for
cooperative launches only. Heap initialization must use the same queue
as the current dispatch.
Change-Id: I856621bf82bbdeb1c2d0fbc4970e90d09af805cb
[ROCm/clr commit: 67af16bdf3]
- params should be valid when used for default flag since we support
unified virtual address space
Signed-off-by: sdashmiz <shadi.dashmiz@amd.com>
Change-Id: I75d40e437b12ee58e72e423bb4818b484ce35b66
[ROCm/clr commit: 5c5b220561]
Newer GCC's seem to require this.
Signed-off-by: Jeremy Newton <Jeremy.Newton@amd.com>
Change-Id: I56ef934600729e892f1cd7ab11662760df48aed8
[ROCm/clr commit: 158e79358c]
No such wrapper or intrinsic would ever exist because there is no such
underlying instruction.
Change-Id: I6c3f64cd2df2a58edf32037da8f5712868f296ea
[ROCm/clr commit: bebbf3c4ec]
MemPool was designed to use hip::Stream, but graph implementation uses
amd::HostQueue. Hence switch graph to hip::Stream management.
Change-Id: Ia319389de45e4c3c6043d17473279a6f27a13140
[ROCm/clr commit: 0fe03b6494]
Xie,AlexBin
Yaxun (Sam) Liu
Ajay
Jason Tang
Todd tiantuo Li
Rakesh Roy
Konstantin Zhuravlyov
victzhan
Ranjith Ramakrishnan
German
Maneesh Gupta
Anusha GodavarthySurya
Ioannis Assiouras
kjayapra-amd
sdashmiz
Sourabh Betigeri
Saleel Kudchadker
David Yat Sin
Chauncey Hui
Jeremy Newton
Satyanvesh Dittakavi
Jacob Lambert
Jaydeep Patel
pghafari
Matt Arsenault