ggml-webgpu: Improve performance of mat-vec and mat-mat for MUL_MAT_ID (#22464)

Browse Source

* Add mat-vec fast path of MUL_MAT_ID.

* Add shared accumulation vec logic and the other types supports.

* Add i-quant mat-mat for MUL_MAT_ID and fix some parts

* Remove n_experts from shader_lib_context.

...

This commit is contained in:

Masashi Yoshimura

2026-04-30 14:19:10 -07:00

committed by GitHub

parent 5cbfb18075

commit a95a11e5b8

5 changed files with 1780 additions and 1295 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-webgpu/wgsl-shaders/mul_mat_vec.wgsl

+3 -1281

File diff suppressed because it is too large Load Diff