* Added buffer tracing support for rocdecode and updated tests to work with buffer tracing
* Updated perfetto to output args individually rather than as a string list
* Updated docstrings and operation type, changed OTF2 code to remove warning due to change in operation type
* Updated tests for review comments
* Test args exist and return value
* Updated to use string entry
* Change function name
* Updated PR to reflect review comments
* Updated for PR review comments
* Change function name
[ROCm/rocprofiler-sdk commit: 077723337a]
* Added buffer tracing support for rocdecode and updated tests to work with buffer tracing
* Updated perfetto to output args individually rather than as a string list
* Updated docstrings and operation type, changed OTF2 code to remove warning due to change in operation type
* Updated tests for review comments
* Test args exist and return value
* Updated to use string entry
* Change function name
* Updated PR to reflect review comments
* Updated for PR review comments
* Change function name
-add check for rhel10 (platform:el10), force use rhel roof binary
-update changelog in 'unreleased- added' section
Signed-off-by: Carrie Fallows <Carrie.Fallows@amd.com>
Invalid rpath on roofline binaries reported during build testing for new RHEL10 addition, removed rpaths to prevent rpath check failures.
Signed-off-by: Carrie Fallows <Carrie.Fallows@amd.com>
* SWDEV-517078 - Maintain the trap handler ABI version in CLR
The trap handler ABI version is communicated to the debugger using
the r_version field in the r_debug structure. This structure is
an external dependency, which makes it complicated to keep the trap
handler source (in CRL) and the ABI version number (external dependency)
in sync.
This patch proposes to patch the trap handler ABI version number in
_amdgpu_r_debug before communicating it to the debugger.
We can't directly include sc's executable.hpp file in CRL as it relies
on conflicting definition of ELF related types, so instead we need to
rely on a-priori knowledge on the r_debug structure. Fortunately, this
structure is part of a stable ABI, so its layout is guaranteed to be
kept stable.
Update the 2nd level trap handler to follow updates from the
ROCr-runtime. The trap handlers are stripped from parts dedicated to
architectures unsupported by CLR.
Bump the r_debug.r_version to track the ABI changes in the trap handler.
[ROCm/clr commit: 7b72c1b786]
* SWDEV-517078 - Maintain the trap handler ABI version in CLR
The trap handler ABI version is communicated to the debugger using
the r_version field in the r_debug structure. This structure is
an external dependency, which makes it complicated to keep the trap
handler source (in CRL) and the ABI version number (external dependency)
in sync.
This patch proposes to patch the trap handler ABI version number in
_amdgpu_r_debug before communicating it to the debugger.
We can't directly include sc's executable.hpp file in CRL as it relies
on conflicting definition of ELF related types, so instead we need to
rely on a-priori knowledge on the r_debug structure. Fortunately, this
structure is part of a stable ABI, so its layout is guaranteed to be
kept stable.
Update the 2nd level trap handler to follow updates from the
ROCr-runtime. The trap handlers are stripped from parts dedicated to
architectures unsupported by CLR.
Bump the r_debug.r_version to track the ABI changes in the trap handler.
This patch will adds recommended sdma supports with
limited XGMI SDMA engine. It will use one PCIe SDMA
to do gpu <-> gpu copies which will help improve all
to all copy performance.
Signed-off-by: Shane Xiao <shane.xiao@amd.com>
This patch will adds recommended sdma supports with
limited XGMI SDMA engine. It will use one PCIe SDMA
to do gpu <-> gpu copies which will help improve all
to all copy performance.
Signed-off-by: Shane Xiao <shane.xiao@amd.com>
[ROCm/ROCR-Runtime commit: 6a63170b38]
Fix pytorch 2.5 issues, by defining reduce sync operations for type __half in amd_hip_fp16.h and not in
amd_warp_sync_functions.h which is problematic in case __half does not get included before that header.
Only define types not supported by cuda if HIP_ENABLE_EXTRA_WARP_SYNC_TYPES is defined, to avoid portability issues
[ROCm/clr commit: 66496258b4]
Fix pytorch 2.5 issues, by defining reduce sync operations for type __half in amd_hip_fp16.h and not in
amd_warp_sync_functions.h which is problematic in case __half does not get included before that header.
Only define types not supported by cuda if HIP_ENABLE_EXTRA_WARP_SYNC_TYPES is defined, to avoid portability issues
* Remove unused forward_list
* Remove unused __read_clock function
* Replace wallClk code with hip function
* Remove unused unit test for ipc
* Remove slab heap
* Remove unused EBO spinlock
[ROCm/rocshmem commit: 0fd628458c]
* Remove unused forward_list
* Remove unused __read_clock function
* Replace wallClk code with hip function
* Remove unused unit test for ipc
* Remove slab heap
* Remove unused EBO spinlock
Explicitly nulling the pointer causes us to report the error below
instead of keeping a dangling pointer around that will most likely lead
to a subsequent segfault.
[ROCm/clr commit: 199b0f1086]
Explicitly nulling the pointer causes us to report the error below
instead of keeping a dangling pointer around that will most likely lead
to a subsequent segfault.
* Update the naming convention for collective APIs to ensure consistency across the interface.
* Move all collective API declarations to rocshmem_COLL.hpp
* The following APIs were updated as part of this change:
- `barrier`
- `barrier_all`
- `sync`
- `sync_all`
- `all_to_all`
- `broadcast`
- `fcollect`
- `all_reduce`
* Update header file generation code for collective APIs
[ROCm/rocshmem commit: 68421895d6]
* Update the naming convention for collective APIs to ensure consistency across the interface.
* Move all collective API declarations to rocshmem_COLL.hpp
* The following APIs were updated as part of this change:
- `barrier`
- `barrier_all`
- `sync`
- `sync_all`
- `all_to_all`
- `broadcast`
- `fcollect`
- `all_reduce`
* Update header file generation code for collective APIs
This now has host conversions too, which is directly from Christopher's
work on fcbx.
Signed-off-by: Christopher M. Riedl
* add const to func parameter
* do not depend on builtins, use gfx950 detection
[ROCm/clr commit: 628777b73d]
This now has host conversions too, which is directly from Christopher's
work on fcbx.
Signed-off-by: Christopher M. Riedl
* add const to func parameter
* do not depend on builtins, use gfx950 detection
Currently, we check if there's enough system RAM even if we don't allocate on host device. This is incorrect logic.
We should not check for this size on windows because PAL checks for memory allocation. See SWDEV-467263.
Co-authored-by: Jimbo Xie <jiabaxie@amd.com>
[ROCm/clr commit: 0d6e554d92]
Currently, we check if there's enough system RAM even if we don't allocate on host device. This is incorrect logic.
We should not check for this size on windows because PAL checks for memory allocation. See SWDEV-467263.
Co-authored-by: Jimbo Xie <jiabaxie@amd.com>
Galantsev, Dmitrii
Trowbridge, Ian
cfallows-amd
anujshuk-amd
Daniel Su
Six, Lancelot
Milanov, Aleksandar
Shane Xiao
Hernandez, Gerardo
Jiang, Julia
Yao, Longlong
Pryor, Adam
Kiriti Gowda
Brandon Potter
Jonathan Kim
Haehnle, Nicolai
Avinash Kethineedi
Nilesh M Negi
Sang, Tao
Chaudhary, Jatin Jaikishan
Andryeyev, German
Xie, Jiabao(Jimbo)