vulkan: allow reducing graph submission batches to avoid device timeouts - #24872
Merged
Conversation
github-actions
Bot
added
Vulkan
0cc4m
reviewed
Jun 21, 2026
wbruna
force-pushed
the
ggml_vk_reduce_submit_igpu
branch
from
June 21, 2026 15:05
ff52c05 to
5e37770
Compare
wbruna
changed the title
0cc4m
approved these changes
Jun 23, 2026
wbruna
force-pushed
the
ggml_vk_reduce_submit_igpu
branch
from
June 24, 2026 11:00
5e37770 to
0f0d2af
Compare
0cc4m
approved these changes
Jun 24, 2026
Contributor
|
@ggml-org/maintainers Another approval needed. |
taronaeo
approved these changes
Jun 24, 2026
Geminihaha
pushed a commit
to Geminihaha/llama.cpp
that referenced
this pull request
Jun 25, 2026
papamoose
pushed a commit
to papamoose/llama.cpp
that referenced
this pull request
Jun 27, 2026
LostRuins
added a commit
to LostRuins/koboldcpp
that referenced
this pull request
Jun 27, 2026
adrianhoehne
pushed a commit
to adrianhoehne/llama.cpp
that referenced
this pull request
Jul 5, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Overview
disable graph submission batching on UMA devices, to avoid "device lost" errorsreduce discrete GPU batching from 100 to 64Should fix #21724 .
Edit: removed the original heuristic to reduce the value on iGPUs.
Additional information
Tested with image generation on my own 3400G, which timeouts even with very small batches.
I've reduced the default max batch size because some reports refer to discrete GPUs, but I don't really know how to choose an appropriate safe value here (and 100 seems to work fine on my own 7600 XT, which is also reported as hitting that bug...).Edit: kept the original 100 value.I changed the env var naming because the value is used as a ceiling, instead of a fixed amount; for consistency, added the
max_prefix to the internal variables too.Supersedes #24240 .
Requirements