- Refactor deviceLocalAlloc arguments
- Refactor hostAlloc code, have cleaner interface
- Kern args buffer need to have execute flag set as CP enforces this on
certain newer HW.
* Remove L2 channels from --list-metrics
--list-metrics moved to general options
List metrics for the current architecture
Filter blocks for metrics
Removed test for --list-metrics in profile mode
Test the options don't throw error
Fixed --config-dir error
Test stdout for command line options
Provide path list for loading panel configs
Show L2 Cache (per) channel metrics
Changed command line option names
Can show two levels only
Removed filtering blocks
Moved blocks to original position
Removed filter block tests
Removed filtering
Formaating fix
Readability enhancement
Test formatting
Filter L2 channels without sysinfo
Show avilable metrics for current arch
Intermediate commit
Fixed tests
Added argument sanitization
Added list_metrics to ctest
merge iconflict resolution
Updated test marker
Updated changelog
Fixed formatting
* Updated docs
* libhsakmt: Update ioctl version to 1.18
Sync with kernel ioctl version.
Also explicitly set the ioctl flag to KFD_PROC_FLAG_MFMA_HIGH_PRECISION
Signed-off-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
* libhsakmt: Sync ioctl header by adding kfd_ioctl_profiler
Sync with kernel ioctl version. Add kfd_ioctl_profiler.
Signed-off-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
---------
Signed-off-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Move check for kernel filtering to add to roofline pdfs- was originally only labelling pdfs with the filtered kernel names from --kernel when --kernel-names was called, we want it at all times when kernels are filtered.
Signed-off-by: Carrie Fallows <Carrie.Fallows@amd.com>
* SWDEV-552613 - Disable Direct Dispatch on Windows
* SWDEV-552613 - Use Direct Dispatch on HSA backend only
---------
Co-authored-by: GunaShekar <agunashe@amd.com>
Co-authored-by: Christophe Paquot <35546540+chrispaquot@users.noreply.github.com>
This PR intends to cover the edge case seen in https://github.com/ROCm/rocm-systems/issues/694.
`hip-config-amd.cmake` uses rocm_agent_enumerator to determine which GPU architecture to target when no target is specified.
https://github.com/ROCm/rocm-systems/blob/9a02dae75f8df9d8f08923d34d06d76e96ced7b4/projects/clr/hipamd/hip-config-amd.cmake.in#L86-L95
On WSL, both `readFromKFD` and `readFromLSPCI` are skipped. If `readFromTargetLstFile()` isn't in use, `readFromROCMINFO()` is called on. If rocminfo times out, it prints the following message to stdout.
```
"Timeout querying rocminfo. Are you compiling with more than 254 threads?"
```
Because this is output and not an explicit error message, `execute_command` in the previous code blocks treats the output as `OUTPUT_VARIABLE` and passes it on as a valid gfx arch which causes these errors in CMake,
```
lang++: error: invalid target ID 'Timeout'; format is a processor name followed by an optional colon-delimited list of features followed by an enable/disable sign (e.g., 'gfx908:sramecc+:xnack-')
clang++: error: invalid target ID 'querying'; format is a processor name followed by an optional colon-delimited list of features followed by an enable/disable sign (e.g., 'gfx908:sramecc+:xnack-')
clang++: error: invalid target ID 'rocminfo.'; format is a processor name followed by an optional colon-delimited list of features followed by an enable/disable sign (e.g., 'gfx908:sramecc+:xnack-')
clang++: error: invalid target ID 'Are'; format is a processor name followed by an optional colon-delimited list of features followed by an enable/disable sign (e.g., 'gfx908:sramecc+:xnack-')
clang++: error: invalid target ID 'you'; format is a processor name followed by an optional colon-delimited list of features followed by an enable/disable sign (e.g., 'gfx908:sramecc+:xnack-')
clang++: error: invalid target ID 'compiling'; format is a processor name followed by an optional colon-delimited list of features followed by an enable/disable sign (e.g., 'gfx908:sramecc+:xnack-')
```
The output can be properly pushed to `ERROR_VARIABLE` if rocm_agent_enumerator pushes the output to stderr instead of stdout. This can be done with the changes to the print statement in this PR or using the `logging` module.
Benjamin Welton
David Yat Sin
amd-hsivasun
systems-assistant[bot]
Joseph Macaranas
vedithal-amd
ggottipa-amd
Satyanvesh Dittakavi
Ioannis Assiouras
Flora Cui
SaleelK
abchoudh-amd
vstojilj
ywang103-amd
hkasivis
cfallows-amd
Todd tiantuo Li
Julia Jiang
David Galiffi
jamessiddeley-amd
Jimbo
Sam Ruscica
Ajay GunaShekar
habajpai-amd
harkgill-amd
marandje
German Andryeyev
RahulC
estewart08
Ammar ELWazir
Saurabh Verma
Pengda Xie