Add logging macros for KleidiAI and fix todos from previous PR #26146
Conversation
Signed-off-by: Orlaith Monahan <[email protected]>
|
@microsoft-github-policy-service agree company="Arm" |
onnxruntime/contrib_ops/cpu/quantization/dynamic_quantize_matmul.cc
Outdated
Show resolved
Hide resolved
…ul.cc Co-authored-by: Edward Chen <[email protected]>
|
Can you please resolve the conflicts ? |
onnxruntime/contrib_ops/cpu/quantization/dynamic_quantize_matmul.cc
Outdated
Show resolved
Hide resolved
Handle cases where NumDimensions for dynamic_quantize_matmul are less than 2
|
/azp run Linux QNN CI Pipeline,Win_TRT_Minimal_CUDA_Test_CI,Windows ARM64 QNN CI Pipeline,Windows GPU Doc Gen CI Pipeline |
|
Azure Pipelines successfully started running 4 pipeline(s). |
Remove redundant checks for SVE
Cover all permutations of 2d tensors for dynamic_quantize_matmul
Lint changes to comment spaces
Lint runner change to if/else formatting
Rename logging macros defines to be clearer Signed-off-by: Orlaith Monahan <[email protected]>
|
/azp run Linux QNN CI Pipeline,Win_TRT_Minimal_CUDA_Test_CI,Windows ARM64 QNN CI Pipeline,Windows GPU Doc Gen CI Pipeline |
|
Azure Pipelines successfully started running 4 pipeline(s). |
Adjust else statement to be in line with lintrunner expectations
|
/azp run Linux QNN CI Pipeline,Win_TRT_Minimal_CUDA_Test_CI,Windows ARM64 QNN CI Pipeline,Windows GPU Doc Gen CI Pipeline |
|
Azure Pipelines successfully started running 4 pipeline(s). |
More lintrunner tweaks
|
/azp run Linux QNN CI Pipeline,Win_TRT_Minimal_CUDA_Test_CI,Windows ARM64 QNN CI Pipeline,Windows GPU Doc Gen CI Pipeline |
|
Azure Pipelines successfully started running 4 pipeline(s). |
|
LGTM. @edgchen1 - Do you have any additional comments ? |
|
The build failures seem unrelated to this PR. Might be something else going on. |
Remove no longer needed comment
|
/azp run Linux QNN CI Pipeline,Win_TRT_Minimal_CUDA_Test_CI,Windows ARM64 QNN CI Pipeline,Windows GPU Doc Gen CI Pipeline |
Co-authored-by: Edward Chen <[email protected]>
|
/azp run Linux QNN CI Pipeline,Win_TRT_Minimal_CUDA_Test_CI,Windows ARM64 QNN CI Pipeline,Windows GPU Doc Gen CI Pipeline |
|
Azure Pipelines successfully started running 4 pipeline(s). |
…runtime_mlas.cmake * Move KLEIDIAI_LOG definition to only be defined when logging is enabled * Refactor how KLEIDIAI_LOG is defined * Added __FILE__ and __LINE__ to the logging outputs Signed-off-by: Orlaith Monahan <[email protected]>
|
/azp run Linux QNN CI Pipeline,Win_TRT_Minimal_CUDA_Test_CI,Windows ARM64 QNN CI Pipeline,Windows GPU Doc Gen CI Pipeline |
|
Azure Pipelines successfully started running 4 pipeline(s). |
* Actually add in the changes to logging to include file and line in the output * Move the logic for enabling logging to be within onnxruntime_USE_KLEIDIAI Signed-off-by: Orlaith Monahan <[email protected]>
|
/azp run Linux QNN CI Pipeline,Win_TRT_Minimal_CUDA_Test_CI,Windows ARM64 QNN CI Pipeline,Windows GPU Doc Gen CI Pipeline |
|
Azure Pipelines successfully started running 4 pipeline(s). |
Signed-off-by: Orlaith Monahan <[email protected]>
|
/azp run Linux QNN CI Pipeline,Win_TRT_Minimal_CUDA_Test_CI,Windows ARM64 QNN CI Pipeline,Windows GPU Doc Gen CI Pipeline |
|
Azure Pipelines successfully started running 4 pipeline(s). |
Signed-off-by: Orlaith Monahan <[email protected]>
|
/azp run Linux QNN CI Pipeline,Win_TRT_Minimal_CUDA_Test_CI,Windows ARM64 QNN CI Pipeline,Windows GPU Doc Gen CI Pipeline |
|
Azure Pipelines successfully started running 4 pipeline(s). |
3 participants
Key changes
This patch contains logging macros for the KleidiAI kernels
It also contains changes / todos from a previous PR:
#25187