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

Browse Source

* ggml-hexagon: implement 2x2 matmul kernel

* hexmm: implement vec_dot_rx2x2 for Q8_0 and MXFP4

* hexagon: fix editor config failures

* hexagon: refactor matmul ops to use context struct and remove wrappers

Also implement vec_dot_f16 2x2

* hexagon: refactor dyn quantizers to use mmctx

* hexagon: remove mm fastdiv from op_ctx

* hexagon: refactor matmul entry point to reduce code duplication

---------

Co-authored-by: Trivikram Reddy <tamarnat@qti.qualcomm.com>

...

This commit is contained in:

Max Krasnyansky

2026-02-11 23:04:27 -08:00

committed by GitHub

co-authored by Trivikram Reddy

parent 4ae1b7517a

commit b1ff83bbb0

2 changed files with 854 additions and 678 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download Patch File Download Diff File Expand all files Collapse all files

ggml/src/ggml-hexagon/htp/matmul-ops.c

+854 -665

File diff suppressed because it is too large Load Diff