llama.cpp
Clean up per-thread parameter buffer pool and job submission logic
#19772
Merged
Merged
Clean up per-thread parameter buffer pool and job submission logic #19772
reeselevine
merged
18 commits into
ggml-org:master
from
nikhilJain17:nikhilJain17/webgpu_buf_pool
Merge
df604979
Merge
5ae7583c
merge
70238e02
Allow webgpu_buf_pool to resize if needed, remove inflight_threads, a…
2806b7d0
Run clang-format
f4279953
Keep track of num batched kernels that have not been submitted yet
7cecc8a2
Run clang-format
58d2a91e
Increase buf pool max size
059e21b1
Increase param buf pool init size
4fa05f7a
Remove webgpu buf pool resizing
4b2af8ea
Merge with master
9f9d0640
Merge with master
a3e6d6fa
Merge with master
a87f3db9
Add buffer pool growth
a8694c07
Move buffer pool growth outside of lock
596e258b
Reduce max pool size to 32
a487db99
Run clang-format
8512456d
Only resize param buf pool
60ae42e3
reeselevine
commented
on 2026-03-02
reeselevine
approved these changes
on 2026-03-02
Login to write a write a comment.
Login via GitHub