LLM: update split tensor conditions. #10872
Merged
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
hkvision
reviewed
Apr 28, 2024
| # split tensor for memory block limitation | ||
| # support fp16 and set input length threshold at 5000 for now | ||
| return True | ||
| if query_layer.element_size()*bsz*n_head*seq_len*seq_len >= 4*1024*1024*1024: |
There was a problem hiding this comment.
will this check hurt performance? query_layer.element_size(), should be no?
There was a problem hiding this comment.
Seems not:
use condition with element_size():
0,meta-llama/Llama-2-7b-chat-hf,4839.14,27.14,0.0,8192-512,1,8192-512,1,fp8,False,4.49,11.966796875,N/A,N/A
use condition with shape:
0,meta-llama/Llama-2-7b-chat-hf,4837.71,27.13,0.0,8192-512,1,8192-512,1,fp8,False,4.48,11.966796875,N/A,N/A
hkvision
approved these changes
Apr 30, 2024
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Outdated suggestions cannot be applied.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.
Description
This PR is to update split tensor conditions:
attn_weightssize larger than block memory limitation.