Ali Hassani eee0cab26c Stamp out 1x1x1 clusters, 128x256 CTA shape (#1665) ...

Adds 128x256 tile shapes to FP16/BF16 and FP8 generators.
Also adds 1x1x1 clusters to all existing FP16/BF16/FP8 generators.

NOTE: it is important to set kernel filter (--kernels /
CUTLASS_LIBRARY_KERNELS) to a non empty string and skip pruning to get
all of the new configurations.

If profiling exhaustively, they can be set to `*`.

Number of CUTLASS 3.X GEMMs before this commit: 2868
Number of CUTLASS 3.X GEMMs after this commit: 4016

Co-authored-by: Ali Hassani <ahassani@nvidia.com>

2024-07-31 20:22:29 -04:00

..

__init__.py

Update license year (#1306)

2024-01-16 14:37:22 -05:00

conv2d_operation.py

CUTLASS 3.5.1 (#1623)

2024-07-29 08:46:24 -04:00

conv3d_operation.py

CUTLASS 3.5.0 (#1411)

2024-03-19 17:51:04 -04:00

conv3x_emitter.py

CUTLASS 3.5.1 (#1623)

2024-07-29 08:46:24 -04:00

gemm_operation.py

CUTLASS 3.5.1 (#1623)

2024-07-29 08:46:24 -04:00

generator.py

Stamp out 1x1x1 clusters, 128x256 CTA shape (#1665)

2024-07-31 20:22:29 -04:00

library.py

Updates for CUTLASS 3.5.0 (#1468)

2024-04-11 21:33:40 -04:00

manifest.py

CUTLASS 3.5.1 (#1623)

2024-07-29 08:46:24 -04:00

rank_2k_operation.py

Update license year (#1306)

2024-01-16 14:37:22 -05:00

rank_k_operation.py

Update license year (#1306)

2024-01-16 14:37:22 -05:00

symm_operation.py

Update license year (#1306)

2024-01-16 14:37:22 -05:00

trmm_operation.py

Update license year (#1306)

2024-01-16 14:37:22 -05:00