llama.cpp
hexagon: optimization for HMX mat_mul
#21554
Merged

hexagon: optimization for HMX mat_mul #21554

njsyw1997
njsyw1997 njsyw1997 requested a review 72 days ago
github-actions github-actions added ggml
github-actions github-actions added Hexagon
max-krasnyansky
njsyw1997 njsyw1997 force pushed from f4b1704d to 0aedac28 68 days ago
njsyw1997 njsyw1997 force pushed from 0aedac28 to a51d65b2 68 days ago
njsyw1997
max-krasnyansky
njsyw1997 njsyw1997 force pushed from 0895350b to 0d799771 68 days ago
njsyw1997
max-krasnyansky
max-krasnyansky approved these changes on 2026-04-12
max-krasnyansky
max-krasnyansky
njsyw1997 hexagon: add async HMX worker
de152555
njsyw1997 hexagon: cost-based VTCM chunk search for out-stationary matmul
b2ec80a6
njsyw1997 hexagon: fix futex race in hmx_worker_drain
f14e9c6f
hex-mm: hmx optimize scatter/transpose and use HMX intrinsics
f8737609
max-krasnyansky hex-vmem: drop vmem limit a touch under 3GB on v73
189a2d2f
njsyw1997 hexagon: add fwd declaration of htp_context
6d82f7fb
max-krasnyansky hex-hmx: replace hmx-worker with hmx-queue that mimics dma-queue inte…
2af9a1c0
max-krasnyansky hex-mm: add debug log to hmx work func called from hmx-queue
c2b48b8f
njsyw1997
max-krasnyansky
max-krasnyansky
njsyw1997 njsyw1997 force pushed from 0d799771 to c2b48b8f 65 days ago
njsyw1997
max-krasnyansky
max-krasnyansky approved these changes on 2026-04-14
max-krasnyansky
max-krasnyansky
max-krasnyansky
max-krasnyansky commented on 2026-04-14
njsyw1997 Update hmx-queue.h
b122c95d
CISC
CISC approved these changes on 2026-04-14
max-krasnyansky max-krasnyansky merged 5d14e5d1 into master 65 days ago

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
Milestone