* Remove std::regex usage from rocprofiler-sdk and common library
- See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118408
- std::regex usage produces exceptions or segfaults when used when on applications compiled with dual ABI
- Add code restrictions workflow
- simple workflow ensuring code restrictions (such as std::regex) are not used
* Update CHANGELOG
* Explicitly set permissions for restrictions workflow
* Fix handling of /proc/cpuinfo entries with no info
- e.g. "power_management:" (colon is last character)
---------
Co-authored-by: Jonathan R. Madsen <jonathanrmadsen@gmail.com>
[ROCm/rocprofiler-sdk commit: dbb2e52216]
* Remove std::regex usage from rocprofiler-sdk and common library
- See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118408
- std::regex usage produces exceptions or segfaults when used when on applications compiled with dual ABI
- Add code restrictions workflow
- simple workflow ensuring code restrictions (such as std::regex) are not used
* Update CHANGELOG
* Explicitly set permissions for restrictions workflow
* Fix handling of /proc/cpuinfo entries with no info
- e.g. "power_management:" (colon is last character)
---------
Co-authored-by: Jonathan R. Madsen <jonathanrmadsen@gmail.com>
For L2 Cache and above, we report the total amount of cache for the
whole partition, so we add up the L2 Cache entry for each partition.
[ROCm/ROCR-Runtime commit: fc561ff37a]
* CI workflow testing compilation against older ROCm versions
* Run on ubuntu-latest
* Install git
* Fix install requirements
* Set --site to $(hostname)
* Test latest rocm release + fix logic for rocm 6.2
* Report number of processors
* Reduce the build job parallelism
- ubuntu-latest only has 2 CPUs for 8 is probably excessive
* Add explicit permissions to rocm_release_compatibility workflow
---------
Co-authored-by: Jonathan R. Madsen <jonathanrmadsen@gmail.com>
[ROCm/rocprofiler-sdk commit: 7f22b66a9b]
* CI workflow testing compilation against older ROCm versions
* Run on ubuntu-latest
* Install git
* Fix install requirements
* Set --site to $(hostname)
* Test latest rocm release + fix logic for rocm 6.2
* Report number of processors
* Reduce the build job parallelism
- ubuntu-latest only has 2 CPUs for 8 is probably excessive
* Add explicit permissions to rocm_release_compatibility workflow
---------
Co-authored-by: Jonathan R. Madsen <jonathanrmadsen@gmail.com>
1. Move all new gfx12 block enums to aql_profile_v2.h. hsa_ven_amd_aqlprofile.h will be left untouched.
2. Re-org counter info in gfx12_block_info.h to be purely alphabetic-ordered for easy comparison between different IP versions. Also use auto-gen block name from IP header files whenever possible to reduce manual editing.
3. Remove unused counter info from graphics blocks.
4. Added UTCL2 and VML2 support
5. Added all gfx12 blocks to ctrl test
[ROCm/aqlprofile commit: b20803e95d]
1. Move all new gfx12 block enums to aql_profile_v2.h. hsa_ven_amd_aqlprofile.h will be left untouched.
2. Re-org counter info in gfx12_block_info.h to be purely alphabetic-ordered for easy comparison between different IP versions. Also use auto-gen block name from IP header files whenever possible to reduce manual editing.
3. Remove unused counter info from graphics blocks.
4. Added UTCL2 and VML2 support
5. Added all gfx12 blocks to ctrl test
Updated:
- Removed backwards compatibility for jpeg_activity/vcn_activity
- On supported ASICs users can use XCP (partition) stat values:
jpeg_busy and vcn_busy
Change-Id: I78c403f8462668738ec57cac12b107f6a3989b18
Signed-off-by: Charis Poag <Charis.Poag@amd.com>
[ROCm/rocm_smi_lib commit: 1c6b2adae7]
Updated:
- Removed backwards compatibility for jpeg_activity/vcn_activity
- On supported ASICs users can use XCP (partition) stat values:
jpeg_busy and vcn_busy
Change-Id: I78c403f8462668738ec57cac12b107f6a3989b18
Signed-off-by: Charis Poag <Charis.Poag@amd.com>
ROCr initially had a bug where memory allocations that were not 4K
aligned were internally 4K aligned but ROCr would not keep track
of user-requested size. This would cause some pointer_info queries
to fail, but HIP was already aligning the buffer sizes for IPC
requests. For backward compatibility accross 2 minor versions,
we allowed IPC look-ups to be both aligned and un-aligned.
Removing this check as this 4 minor versions have been released
since then.
ROCr initially had a bug where memory allocations that were not 4K
aligned were internally 4K aligned but ROCr would not keep track
of user-requested size. This would cause some pointer_info queries
to fail, but HIP was already aligning the buffer sizes for IPC
requests. For backward compatibility accross 2 minor versions,
we allowed IPC look-ups to be both aligned and un-aligned.
Removing this check as this 4 minor versions have been released
since then.
[ROCm/ROCR-Runtime commit: d52f1d0453]
* GFX950 Stochastic PC sampling
* Use actual type instead void *
* error reporting if the pcs method is inappropriate
[ROCm/rocprofiler-sdk commit: 2bb64e9c9a]
* SWDEV-530921 - Remove the usage of __AMDGCN_WAVEFRONT_SIZE as compile time constant
* wavefrontsize builtin to be used only in device compilation
* SWDEV-530921 - Remove the usage of __AMDGCN_WAVEFRONT_SIZE as compile time constant
[ROCm/clr commit: 54503f0d67]
* SWDEV-530921 - Remove the usage of __AMDGCN_WAVEFRONT_SIZE as compile time constant
* wavefrontsize builtin to be used only in device compilation
* SWDEV-530921 - Remove the usage of __AMDGCN_WAVEFRONT_SIZE as compile time constant
Assiouras, Ioannis
Madsen, Jonathan
David Yat Sin
Ma, Bing
Xie, AlexBin
Baraldi, Giovanni
xuchen-amd
habajpai-amd
Charis Poag
Ben Richard
Belton-Schure, Aidan
Elwazir, Ammar
Indic, Vladimir
Manocha, Rahul
Luzynski, Sebastian
Betigeri, Sourabh
Arandjelovic, Marko
Dittakavi, Satyanvesh