test(vector/bench): land the deletion-aware search benchmark from the #625 work (#848)#850
Merged
Conversation
…625 work (#848) Add deletion_search_bench, developed during #625's investigation and archived with its NO-GO diff: VectorStore::search (HNSW graph path) on a committed in-memory index (n=10k, eager int8, top-10), measuring the pristine traversal (del0pct) and the deletion-bookkeeping traversal with 10% soft-deleted docs (del10pct). This is the first benchmark covering the per-neighbour is_deleted bookkeeping path. It measures very stably (same-code spread was +/-0.06-3.5% during #625, far tighter than raw HNSW-searcher benches); baseline on current main: del0pct 83.0 us, del10pct 62.3 us. Closes #848
This was referenced Jul 9, 2026
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Lands
deletion_search_bench, developed during #625's investigation and archived with its NO-GO diff (the abandoned COW source changes stay abandoned — this restores only the benchmark + itsCargo.tomlregistration).VectorStore::search(HNSW graph path) on a committed in-memory index: n = 10 000, eager int8, top-10.del0pct: the pristine, bookkeeping-free traversal (regression guard).del10pct: 10% soft-deleted docs — the first benchmark covering the per-neighbouris_deletedbookkeeping path.Measures very stably (same-code spread ±0.06–3.5% during #625, far tighter than raw HNSW-searcher benches). Baseline on current
main:del0pct83.0 µs,del10pct62.3 µs — consistent with the #625-era measurements.Testing
cargo bench --bench deletion_search_benchruns both cases (baseline above); fmt/clippy clean.Closes #848