Add IVF sampled ADC start-point router - #1279
Draft
xwj-ox wants to merge 1 commit into
Draft
Conversation
|
@xwj-ox please read the following Contributor License Agreement(CLA). If you agree with the CLA, please reply with the following information.
Contributor License AgreementContribution License AgreementThis Contribution License Agreement (“Agreement”) is agreed to by the party signing below (“You”),
|
Contributor
There was a problem hiding this comment.
Pull request overview
This PR introduces an IVF+PQ posting-list sampled ADC start-point router for DiskANN disk search, including artifact build/load support, query-time routing to seed disk Vamana traversal, and benchmark plumbing to configure and report router metrics. It also adds a bilingual RFC documenting the design and experimental operating points.
Changes:
- Add IVF+PQ router artifact format + builder (centroids + posting IDs + offsets + optional fallback medoid) and query-time routing integration for disk search.
- Extend disk search statistics and benchmark reporting to include router time and sampled-ADC diagnostics.
- Add benchmark configuration schema for
ivf_pq_router_buildandstart_point_router, plus RFC documentation (EN/ZH).
Reviewed changes
Copilot reviewed 10 out of 10 changed files in this pull request and generated 5 comments.
Show a summary per file
| File | Description |
|---|---|
| rfcs/00001-ivf-pq-posting-list-sampled-adc-router.zh.md | Chinese RFC describing sampled ADC router design, metrics, and rollout plan |
| rfcs/00001-ivf-pq-posting-list-sampled-adc-router.en.md | English RFC describing sampled ADC router design, metrics, and rollout plan |
| diskann-disk/src/utils/statistics.rs | Adds router timing and diagnostic counters to QueryStatistics |
| diskann-disk/src/search/start_point_router.rs | Introduces StartPointRouter enum wrapper for query-time routing |
| diskann-disk/src/search/provider/disk_provider.rs | Integrates router into disk search path; records router stats; adds routing helpers + unit tests |
| diskann-disk/src/search/mod.rs | Exposes new ivf_pq_router and start_point_router modules |
| diskann-disk/src/search/ivf_pq_router.rs | Implements IVF router artifact build/serialize/deserialize and query-time router parameters |
| diskann-benchmark/src/inputs/disk.rs | Adds JSON config for router artifact build and start-point router selection |
| diskann-benchmark/src/disk_index/search.rs | Loads router artifacts, passes router into disk searcher, and reports new router metrics |
| diskann-benchmark/src/disk_index/build.rs | Builds IVF+PQ router artifacts during benchmark disk-index build step |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comment on lines
+390
to
+392
| if data.len() != num_points * dim { | ||
| return Err(index_error("data length must equal num_points * dim")); | ||
| } |
Comment on lines
+354
to
+365
| let mut centroids = vec![0.0; num_centroids * dim]; | ||
| for value in &mut centroids { | ||
| *value = reader.read_f32::<LittleEndian>()?; | ||
| } | ||
| let mut offsets = vec![0usize; num_centroids + 1]; | ||
| for offset in &mut offsets { | ||
| *offset = usize::try_from(reader.read_u64::<LittleEndian>()?)?; | ||
| } | ||
| let mut posting_ids = vec![0u32; num_postings]; | ||
| for id in &mut posting_ids { | ||
| *id = reader.read_u32::<LittleEndian>()?; | ||
| } |
Comment on lines
1516
to
1518
| query_stats.cpu_time_us = query_stats.total_execution_time_us | ||
| - query_stats.io_time_us | ||
| - query_stats.query_pq_preprocess_time_us; |
| | Parameter | Value | | ||
| |---|---:| | ||
| | IVF cells | 2048 | | ||
| | load mode | mmap | |
| | 参数 | 值 | | ||
| |---|---:| | ||
| | IVF cells | 2048 | | ||
| | load mode | mmap | |
Codecov Report❌ Patch coverage is Additional details and impacted files@@ Coverage Diff @@
## main #1279 +/- ##
==========================================
- Coverage 91.50% 91.31% -0.20%
==========================================
Files 497 499 +2
Lines 95524 96230 +706
==========================================
+ Hits 87411 87874 +463
- Misses 8113 8356 +243
Flags with carried forward coverage won't be shown. Click here to find out more.
🚀 New features to boost your workflow:
|
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
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
ivf_pq_router_buildandstart_point_routerScope
This intentionally excludes residual PQ, PQ sidecars, block routers, hierarchy, posting-order PQ layout, mmap artifact loading, and experiment result artifacts.
Validation
cargo fmt --all --checkcargo test -p diskann-disk ivf -- --nocapturecargo test -p diskann-benchmark --features disk-index ivf_pq -- --nocapturecargo clippy -p diskann-disk --tests -- -D warningscargo clippy -p diskann-benchmark --features disk-index --bins --tests -- -D warnings