misc : prevent RAM peaking at model loading stage - #27483
Merged
Conversation
Contributor
Author
|
@CISC @ggerganov gentle reminder |
Member
|
I have no particular stake in this, LGTM though, @ggerganov? |
ggerganov
approved these changes
Sep 3, 2026
CISC
approved these changes
Sep 3, 2026
fewtarius
pushed a commit
to fewtarius/CachyLLama
that referenced
this pull request
Sep 5, 2026
thecodacus
pushed a commit
to thecodacus/llama.cpp
that referenced
this pull request
Sep 7, 2026
SteelPh0enix
pushed a commit
to SteelPh0enix/llama.cpp-qwen4exp
that referenced
this pull request
Sep 8, 2026
(cherry picked from commit 5ec4eab)
SteelPh0enix
pushed a commit
to SteelPh0enix/llama.cpp-qwen4exp
that referenced
this pull request
Sep 8, 2026
(cherry picked from commit 5ec4eab)
x1250
pushed a commit
to x1250/llama.cpp
that referenced
this pull request
Sep 9, 2026
zbrad
pushed a commit
to zbrad/llama.cpp
that referenced
this pull request
Sep 10, 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
For models where embeddings have a larger memory footprint than context, peak RAM occurs during model loading when repack is enabled. This makes peak RAM constant across different context lengths.
E.g. here
This PR changes the order of tensors during repack and move the peak to the beginning. I also moved from a single always-growing buffer to per-tensor buffers. There will be more allocations, but no peak.
Here is how it looks after the change
Additional information
The first thing I tried was an in-place repack, and indeed it worked, but the changes were very intrusive and required changes to ggml and backends.
Requirements