llama.cpp
hexagon: further optimization and tuning of matmul and dot kernels
#19407
Merged
Overview
Commits
7
Changes
View On GitHub

hexagon: further optimization and tuning of matmul and dot kernels #19407

max-krasnyansky merged 7 commits into ggml-org:master from qualcomm:hexagon-vecdot-2x2
max-krasnyansky
max-krasnyansky max-krasnyansky requested a review from lhez lhez 126 days ago
github-actions github-actions added ggml
lhez
lhez approved these changes on 2026-02-07
chraac
trivikram-reddy1 ggml-hexagon: implement 2x2 matmul kernel
82b78c6f
max-krasnyansky hexmm: implement vec_dot_rx2x2 for Q8_0 and MXFP4
0c6e6f35
max-krasnyansky hexagon: fix editor config failures
51a9b96d
max-krasnyansky hexagon: refactor matmul ops to use context struct and remove wrappers
80ddf6f9
max-krasnyansky hexagon: refactor dyn quantizers to use mmctx
6e46aa5b
max-krasnyansky hexagon: remove mm fastdiv from op_ctx
34807a22
max-krasnyansky hexagon: refactor matmul entry point to reduce code duplication
b2ab988d
max-krasnyansky max-krasnyansky force pushed from fcd2f80e to b2ab988d 121 days ago
max-krasnyansky max-krasnyansky merged b1ff83bb into master 121 days ago
max-krasnyansky max-krasnyansky deleted the hexagon-vecdot-2x2 branch 59 days ago

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
ggml
Milestone
No milestone
Terms Of Use Privacy Policy Refund Policy Impressum