SNMG ANN #231
Merged
SNMG ANN #231
+6,800
−46
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
divyegala
reviewed
Jul 18, 2024
divyegala
reviewed
Jul 19, 2024
tfeher
requested changes
Jul 24, 2024
tfeher
requested changes
Jul 28, 2024
tfeher
requested changes
Jul 29, 2024
There was a problem hiding this comment.
We are almost there, but the matrix extent type has to be fixed: we should distinguish the neighbor index type (IdxT, usually int64), and the mdspan extent type which is always int64. In a previous review I have marked such changes in ann_mp.cuh, but actually the public API in ann_mg.hpp has also a few places where this need to be fixed
tfeher
added
feature request
achirkin
reviewed
Jul 30, 2024
|
/ok to test |
|
/ok to test |
|
/ok to test |
|
/ok to test |
|
/ok to test |
|
/ok to test |
|
/ok to test |
|
/ok to test |
|
CI error here seemed to have been a connection error: |
|
/ok to test |
|
/ok to test |
|
/ok to test |
|
/ok to test |
|
/merge |
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Outdated suggestions cannot be applied.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.
This PR implements a distributed (single-node-multiple-GPUs) implementation of ANN indexes. It allows to build, extend and search an index on multiple GPUs.
Before building the index, the user has to choose between two modes :
Sharding mode : The index dataset is split, each GPU trains its own index with its respective share of the dataset. This is intended to both increase the search throughput and the maximal size of the index.
Index duplication mode : The index is built once on a GPU and then copied over to others. Alternatively, the index dataset is sent to each GPU to be built there. This intended to increase the search throughput.
SNMG indexes can be serialized and de-serialized. Local models can also be deserialized and deployed in index duplication mode.
Migrated from rapidsai/raft#1993