llama.cpp
ggml-webgpu: parameterize submission size and add iOS specific limits
#21533
Merged
Merged
ggml-webgpu: parameterize submission size and add iOS specific limits #21533
ggerganov
merged
20 commits into
ggml-org:master
from
reeselevine:browser-params
Work towards removing bitcast
f1eb80ef
Move rest of existing types over
e9af481e
Add timeout back to wait and remove synchronous set_tensor/memset_tensor
b3aa3be8
move to unpackf16 for wider compatibility
67fe0897
cleanup
e85e8bcc
Remove deadlock condition in free_bufs
32ee70a2
Merge remote-tracking branch 'upstream/master' into remove_bitcast
309ef1f7
Start work on removing parameter buffer pools
1fc8b64a
Simplify and optimize further
9592ed56
Merge remote-tracking branch 'upstream/master' into one-buffer
82008f32
simplify profile futures
d307d47c
Fix stride
a2c1d910
Try using a single command buffer per batch
e8ea0441
Merge remote-tracking branch 'upstream/master' into one-buffer
e954542c
formatting
7450dd01
Add parameters for different browsers in-flight submissions
cc6f8089
Update handling of batch size too
d63c7395
Throttle ios as much as possible
f9d63dc5
Merge remote-tracking branch 'upstream/master' into browser-params
29ba9c73
Increase timeout for llvm-pipe testing
5b0fd3d2
nikhilJain17
approved these changes
on 2026-04-07
Reviewers
Assignees
No one assigned
Labels
ggml
merge ready
WebGPU
Milestone
No milestone
Login to write a write a comment.
Login via GitHub