ggml-webgpu: compute pass batching and removing profiling overhead (llama/21873)

Browse Source

* Update register tiling matmul to use f32 accumulation

* fix profiling code

* Fix register tiling matmul for chrome, i'm blaming dawn

* Update batch tuning value for iOS

* compile fix

* Fix use of new load function

* Move to a single query set for GPU profiling

* Move to batching compute passes when not profiling

* Refactor build_multi

* remove iOS throttling now that we're batching compute passes

...

This commit is contained in:

Reese Levine 2026-04-16 01:12:19 -07:00 committed by Georgi Gerganov

parent f62bb13320

commit 092330b474

No known key found for this signature in database

GPG Key ID: 449E073F9DC10735

1 changed files with 349 additions and 452 deletions

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

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

801

ggml/src/ggml-webgpu/ggml-webgpu.cpp

View File

File diff suppressed because it is too large Load Diff