metal: bug-fix when enable ggml-alloc #2757
Merged
Conversation
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
The ggml-alloc should only free tensors at memory barriers.
In certain cases, the allocate_node() function may silently return without performing any memory allocation.
lshzh-ww
force-pushed
the
metal-concur-alloc-fix
branch
from
8496a24 to
0c268a8
Compare
|
@ggerganov |
slaren
approved these changes
Aug 24, 2023
|
This has a seg fault in ggml-alloc.c @ line 508 on 70b parameter models. This is happening right after |
akawrykow
pushed a commit
to akawrykow/llama.cpp
that referenced
this pull request
Aug 29, 2023
* metal: better memory alloc w/ concurrency dispatch The ggml-alloc should only free tensors at memory barriers. * ggml-alloc: avoid return silently In certain cases, the allocate_node() function may silently return without performing any memory allocation.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
The first commit improves memory management when using ggml-alloc: We should only free the tensor at memory barriers. Otherwise, memory released by one tensor might be immediately reused by another tensor that runs concurrently with it.
The second commit fixes a silent return in the
allocate_node()function.This PR also removes workarounds needed by Falcon.