To test ACCESS_IN_PLACE GPU mapping update to system memory.
Change-Id: I5b990215f39692e829128d848125e1ae0d571e03
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
To test ACCESS_IN_PLACE GPU mapping update to system memory.
Change-Id: I5b990215f39692e829128d848125e1ae0d571e03
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
[ROCm/ROCR-Runtime commit: 351a41ac76]
Corrects an invalid call to 'ar', which results in link failures for mixed shared
and static link invocations using hipcc.
Change-Id: I51c33438106af066a13e5ff9533721cdc326790d
[ROCm/hip commit: 6d20ee19c7]
Corrects an invalid call to 'ar', which results in link failures for mixed shared
and static link invocations using hipcc.
Change-Id: I51c33438106af066a13e5ff9533721cdc326790d
Current packaging assumes that HIP runtime will always be installed in
/opt/rocm/lib. This is false to assume, because some distros like CentOS
will use the lib64 directory instead of lib. Relying on CMake to choose
the library directory for us will default in that case to lib64. Hence
there will be a mismatch between where HIP is installed and where CMake
thinks it is.
Change-Id: I46e405b25bda987e74b095fcb41319f8a504ae41
[ROCm/clr commit: 52ff66f4fc]
Current packaging assumes that HIP runtime will always be installed in
/opt/rocm/lib. This is false to assume, because some distros like CentOS
will use the lib64 directory instead of lib. Relying on CMake to choose
the library directory for us will default in that case to lib64. Hence
there will be a mismatch between where HIP is installed and where CMake
thinks it is.
Change-Id: I46e405b25bda987e74b095fcb41319f8a504ae41
Check GPU signal status for event before falling into CPU command
status validation
Change-Id: I66f15752d7dca550c0fa1a2252ec5a63817391c3
[ROCm/clr commit: dd8265fc87]
Currently build is only passing because amdhip64 is being built with
headers from /opt/rocm/include. It should be using the headers from
the hipamd project.
Change-Id: I27cb64b460547ef281d368322afac376fa49db51
[ROCm/clr commit: 37be0e449a]
Currently build is only passing because amdhip64 is being built with
headers from /opt/rocm/include. It should be using the headers from
the hipamd project.
Change-Id: I27cb64b460547ef281d368322afac376fa49db51
CoherentHostAccess flag member moved from HSA_MEMORYPROPERTY
to HSA_CAPABILITY struct. Now this is reported to the
topology as a capability of the device instead of a device
memory property.
Change-Id: I48e43e4b4a0635b711b62933734587facdfbf88b
Signed-off-by: Alex Sierra <alex.sierra@amd.com>
CoherentHostAccess flag member moved from HSA_MEMORYPROPERTY
to HSA_CAPABILITY struct. Now this is reported to the
topology as a capability of the device instead of a device
memory property.
Change-Id: I48e43e4b4a0635b711b62933734587facdfbf88b
Signed-off-by: Alex Sierra <alex.sierra@amd.com>
[ROCm/ROCR-Runtime commit: f85b428265]
Enables the fragment allocator to handle >2MB allocations, maintaining
good TLB alignment. Prior code contained a bug that caused the effective
API granule for vram allocations >2MB to be bumped to 2MB.
Also adjusts the block cache's block retention heuristic to not
count discarded blocks as in use. This will reduce block retention
when a significant amount of large blocks or IPC is in use.
Change-Id: I30bd85eb87951df822211f799d9cfe579ab109c6
Enables the fragment allocator to handle >2MB allocations, maintaining
good TLB alignment. Prior code contained a bug that caused the effective
API granule for vram allocations >2MB to be bumped to 2MB.
Also adjusts the block cache's block retention heuristic to not
count discarded blocks as in use. This will reduce block retention
when a significant amount of large blocks or IPC is in use.
Change-Id: I30bd85eb87951df822211f799d9cfe579ab109c6
[ROCm/ROCR-Runtime commit: 8adbda1c18]
Marker events inside hcc_ops_trace.txt are from barriers so they are not meant to be stored in ops_patch_data map.
Added support for hipMemset events which are a kind of memory copy.
Change-Id: I213fe959bcd35ff0371613ba5bffd95bc53e06b5
Marker events inside hcc_ops_trace.txt are from barriers so they are not meant to be stored in ops_patch_data map.
Added support for hipMemset events which are a kind of memory copy.
Change-Id: I213fe959bcd35ff0371613ba5bffd95bc53e06b5
[ROCm/rocprofiler commit: caa5f32300]
This reverts commit 261b237f17.
Reason for revert: Reverting this change to enable split project flow working. Once that is enabled this change will be submitted in hipamd project.
Change-Id: Iad2e044a173787a57a80c5a3d04795e7c2dc5ad5
[ROCm/hip commit: 93b53dc6fa]
This reverts commit ed9e4ee7df.
Reason for revert: Reverting this change to enable split project flow working. Once that is enabled this change will be submitted in hipamd project.
Change-Id: Iad2e044a173787a57a80c5a3d04795e7c2dc5ad5
RDC can optimize by bulk fetching multiple metrics using a single
rocm_smi call. However, currently this is not completely supported in
all ASIC generations. By default disable this for now.
Set environment variable RDC_BULK_FETCH_ENABLED=TRUE to enable
RDC bulk fetch.
BUG: SWDEV-289316
Change-Id: Ibb55514f198356dccf5f47bb0fd2d53c17acb251
RDC can optimize by bulk fetching multiple metrics using a single
rocm_smi call. However, currently this is not completely supported in
all ASIC generations. By default disable this for now.
Set environment variable RDC_BULK_FETCH_ENABLED=TRUE to enable
RDC bulk fetch.
BUG: SWDEV-289316
Change-Id: Ibb55514f198356dccf5f47bb0fd2d53c17acb251
[ROCm/rdc commit: 673f5a4ee1]
recordid cannot be just a counter. The code removed was doing
just that i.e. incrementing a counter. Recordid has to come
from recvals data structure. That code was left there since
a while when Evgeny and Rachida were trying to prototype this feature.
I am not sure why it was not spotted before.
Change-Id: Ia867066dcfca083fcd4111f2aefc2fec88c26314
recordid cannot be just a counter. The code removed was doing
just that i.e. incrementing a counter. Recordid has to come
from recvals data structure. That code was left there since
a while when Evgeny and Rachida were trying to prototype this feature.
I am not sure why it was not spotted before.
Change-Id: Ia867066dcfca083fcd4111f2aefc2fec88c26314
[ROCm/rocprofiler commit: 4ba91a972c]
Current packaging assumes that HIP runtime will always be installed in
/opt/rocm/lib. This is false to assume, because some distros like CentOS
will use the lib64 directory instead of lib. Relying on CMake to choose
the library directory for us will default in that case to lib64. Hence
there will be a mismatch between where HIP is installed and where CMake
thinks it is.
Change-Id: If6f881eab027bdbe8435d3de5295b00277e96a6a
[ROCm/hip commit: ed9e4ee7df]
Current packaging assumes that HIP runtime will always be installed in
/opt/rocm/lib. This is false to assume, because some distros like CentOS
will use the lib64 directory instead of lib. Relying on CMake to choose
the library directory for us will default in that case to lib64. Hence
there will be a mismatch between where HIP is installed and where CMake
thinks it is.
Change-Id: If6f881eab027bdbe8435d3de5295b00277e96a6a
pghafari
Philip Yang
Nick Curtis
Vladislav Sytchenko
German Andryeyev
Scott Linder
Sarbojit Sarkar
Saleel Kudchadker
Alex Sierra
Sean Keely
Julia Jiang
rachida
Arya.Rafii
Rahul Garg
Bill(Shuzhou) Liu