GitLab

Ensure unique module name for MLIR standalone ops (#2360)
Cherry Pick ASAN build excluding additional bin files -1839 #2370
Use an older version of numpy for compatibility with Python3.6 #2369
Add space after rate message
Fix wrong size check when axes not present for slice (#2270)
Updated Changelog to document latest release work (#2363)
Run simplify_qdq first before optimize_module (#2387)
fix unit tests and verification issues on navi
Set numpy to 1.21.6 and disable py3.6 test
Release notes & changelog updates (#2395)

Cherry Pick ASAN build excluding additional bin files -1839 #2370
Use an older version of numpy for compatibility with Python3.6 #2369
Add space after rate message
Fix wrong size check when axes not present for slice (#2270)
Updated Changelog to document latest release work (#2363)

Adds workarounds to avoid passing capture ops and scalar literals from quantization as arguments to ck_gemm.

* Disable -Wunsafe-buffer-usage when compiling gpu code

This will enable MLIR by default for these cases:

Any convolution fusion
Any int8 gemm fusion
All Navi3 standalone convolutions
With a flag(ie MIGRAPHX_ENABLE_MLIR) to enable MLIR for floating-point gemm fusions
Except:

3x3 winnograd convolutions fusions (except on Navi)
K > 2048 on gemm (as CK)
Also there is MIGRAPHX_DISABLE_MLIR to disable MLIR completely.

Add support for the QLinearMatMul onnx operator

Currently, we see MLIR partition candidates recieving non-standard shape due to not fusing in squeeze, flatten and unsqueeze ops. These ops could be canonicalized to reshape without introducing additional ops as long as MLIR backend is concerned.

* Fix scatter operator for nonstandard shapes

remove standard() shape check for scatter inputs.

* Add nostandard input tests for scatter

---------
Co-authored-by: Chris Austen <causten@users.noreply.github.com>