Forward-merge branch-24.02 to branch-24.04 #2131
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
This PR address #1901 by subsampling the input dataset for PQ codebook training to reduce the runtime. Currently, a similar strategy is applied to `per_cluster` method, but not to the default `per_subset` method. This PR fixes this gap. Similar to the subsampling mechanism of the `per_cluster` method, we pick at minimum `256*max(pq_book_size, pq_dim)` number of input rows for training each code book. https://github.com/rapidsai/raft/blob/cf4e03d0b952c1baac73f695f94d6482d8c391d8/cpp/include/raft/neighbors/detail/ivf_pq_build.cuh#L408 The following performance numbers are generated using Deep-100M dataset. After subsampling, the search time and accuracy are not impacted (within +-5%) except one case where I saw 9% performance drop on search (using 10K batch for search). More extensive benchmarking across datasets seems to be needed for justification. Dataset | n_iter | n_list | pq_bits | pq_dim | ratio | Original time (s) | Subsampling (s) | Speedup [subsampling] -- | -- | -- | -- | -- | -- | -- | -- | -- Deep-100M | 25 | 50000 | 4 | 96 | 10 | 129 | 89.5 | 1.44 Deep-100M | 25 | 50000 | 5 | 96 | 10 | 128 | 89.4 | 1.43 Deep-100M | 25 | 50000 | 6 | 96 | 10 | 131 | 90 | 1.46 Deep-100M | 25 | 50000 | 7 | 96 | 10 | 129 | 91.1 | 1.42 Deep-100M | 25 | 50000 | 8 | 96 | 10 | 149 | 93.4 | 1.60 Note, after subsampling, the PQ codebook generation is no longer a bottleneck in the IVF-PQ index building. More optimizations on PQ codebook generation seem unnecessary. Although we could in theory apply the custom kernel approach (#2050) with subsampling, my early tests show the current GEMM approach performs better than the custom kernel after subsampling. Using multiple stream could improve the performance further by overlapping kernels for different `pq_dim`, given kernels are small after subsampling and may not fully utilize GPU. However, as mention above, since the entire PQ codebook is fast, this optimization may not be worthwhile. TODO - [x] Benchmark the performance/accuracy impacts on multiple datasets Authors: - Rui Lan (https://github.com/abc99lr) - Ray Douglass (https://github.com/raydouglass) - gpuCI (https://github.com/GPUtester) Approvers: - Tamas Bela Feher (https://github.com/tfeher) - Corey J. Nolet (https://github.com/cjnolet) URL: #2052
|
SUCCESS - forward-merge complete. |
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.
Forward-merge triggered by push to
branch-24.02that creates a PR to keepbranch-24.04up-to-date. If this PR is unable to be immediately merged due to conflicts, it will remain open for the team to manually merge.