[WIP] Grouped GEMM with ck_tile#434

Draft

matthiasdiener wants to merge 21 commits intodevfrom

ck-grouped-gemm

Commits on Dec 9, 2025

GEMM reference HIP implementation
matthiasdiener
committed

Commits on Dec 11, 2025

blockwise amax
matthiasdiener
committed

Commits on Dec 18, 2025

Merge branch 'dev' into compute-ref-offload
matthiasdiener
committed

Commits on Jan 13, 2026

Change to use Tensor arguments, combine mxfp8/non-mxfp8 paths
matthiasdiener
committed

Commits on Jan 15, 2026

Commits on Jan 19, 2026

Merge branch 'dev' into compute-ref-offload
matthiasdiener
committed

Commits on Jan 22, 2026

Commits on Jan 23, 2026

Pass D Tensor into run_reference and move RefD allocation into PerformTest
matthiasdiener
committed

Commits on Jan 27, 2026

[WIP] proof-of-concept: grouped GEMM with ck_tile
matthiasdiener
committed

Commits on Jan 28, 2026

Merge branch 'dev' into ck-grouped-gemm
matthiasdiener
committed

Commits on Jan 29, 2026

Commits on Jan 30, 2026