ROCm / AMDMIGraphX

AMD's graph optimization engine.
https://rocm.docs.amd.com/projects/AMDMIGraphX/en/latest/
MIT License
185 stars 86 forks source link

add flag to trace quantization passes #3571

Closed shivadbhavsar closed 2 weeks ago

shivadbhavsar commented 2 weeks ago

Currently if we do something like: driver compile model.onnx --fp16

The quantize_fp16 pass will be run before compilation (meaning MIGRAPHX_TRACE_COMPILE will not be able to trace passes run here). Needed to use this trace for debugging so might be useful to merge in

codecov[bot] commented 2 weeks ago

Codecov Report

Attention: Patch coverage is 80.00000% with 2 lines in your changes missing coverage. Please review.

Project coverage is 92.16%. Comparing base (6faf926) to head (8ff7567). Report is 2 commits behind head on develop.

Files with missing lines Patch % Lines
src/quantization.cpp 80.00% 2 Missing :warning:
Additional details and impacted files ```diff @@ Coverage Diff @@ ## develop #3571 +/- ## =========================================== - Coverage 92.16% 92.16% -0.01% =========================================== Files 512 512 Lines 21398 21403 +5 =========================================== + Hits 19722 19726 +4 - Misses 1676 1677 +1 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.