llama : reduce useless copies when saving session #8916
Merged
+28
−11
Conversation
This file contains 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
compilade
added
performance
slaren
approved these changes
Aug 7, 2024
|
This looks great. Thank you! |
ggerganov
approved these changes
Aug 8, 2024
|
OK, re-ran my profiling so I have real numbers to share. This speeds up This cuts roughly 5-6% off |
arthw
pushed a commit
to arthw/llama.cpp
that referenced
this pull request
Nov 15, 2024
* llama : avoid useless copies in dummy session writer * llama : avoid double tensor copy when saving session to buffer
arthw
pushed a commit
to arthw/llama.cpp
that referenced
this pull request
Nov 18, 2024
* llama : avoid useless copies in dummy session writer * llama : avoid double tensor copy when saving session to buffer
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Outdated suggestions cannot be applied.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.
Should help with #8915.
Some useless copies to immediately discarded temporary buffers were done as of #8699 because session size calculation and session file writing now share mostly the same code.
On CPU, the speed was reasonable, but on CUDA, as reported in #8915, this makes the session size calculation too slow.
To fix this, it's possible to simply avoid calling
ggml_backend_tensor_getwhen the data won't be used (i.e. when calculating the session file size).I've also eliminated the double tensor copies when saving the state to a buffer.
@josharian does this help with your use-case?
TODO