Alexander Efimov 605a90c58e [MFMA] Support tile size 4x4 version 1 (#413) ...

This PR enables 4x4 tile size in MFMA based dot operations.

Supported tiled dot is (4x64) x (64x4) -> (4x4) in MFMA layout.
However, actual dot operation should have at least 64 output elements, this is a limitation of other layouts appearing during result processing (i.e. blocked layout can not handle tensors smaller than wavesize).

For example, following dots are supported: (4x64) x (64x16) -> (4x16), (16x64) x (64x4) -> (16x4) or (8x64) x (64x8) -> (8x8)
Following dots are not supporter: (4x128) x (128x4) -> (4x4), (4x64) x (64x8) -> (4x8)

This is a first version of dot using mfma 4x4 instructions, with redundancy and reductions.

2023-12-12 18:23:55 +01:00

..

backend

[FRONTEND] Enable triton to support register thirdparty backend at runtime (#1643)

2023-06-09 09:09:59 -07:00

kernel_comparison

[CI] Update kernels names (#2093)

2023-08-14 19:41:41 -07:00

regression

Revert "[OPTIMIZER] Tweak warpsPerCTA based on the shape of MMA output (#2485)" (#2497)

2023-10-13 23:32:59 -07:00

tools

[CI] upload only test/unit/operators cache to artifacts and rely on kernel names in cache to compare artifacts (#2111)

2023-08-16 20:34:40 -07:00

unit

[MFMA] Support tile size 4x4 version 1 (#413)

2023-12-12 18:23:55 +01:00