Draft
Conversation
kathirgounder
force-pushed
the
BatchedGPURetriever
branch
from
4ebf898 to
16fa3b8
Compare
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Main Idea: Doing the similarity matrix multiplication in one shot on the GPU is about 2 seconds for the Dubrovnik Descriptors with dimension (6044, 8448). Trying to see how far I could push my 2080 TI (11GB VRAM) the max dataset I could fit was (40,000 , 8448).
So I batched the matrix multiply up and was able to scale upto 100,000 descriptors in 20 seconds. The current Similarity Retriever takes 321 seconds on a 30k image dataset. The FAISS implementation I had earlier took about 200 seconds while it missed a lot of pairs, the Index Building times were also just not worth it.
Will update this PR with more timing info and experiments