quickwit-oss/tantivy

Optimize Segment Readers Alive Doc Iterator

Open

#296 opened on May 8, 2018

View on GitHub
 (3 comments) (0 reactions) (0 assignees)Rust (8,354 stars) (499 forks)batch import
enhancementgood first issueperformance

Description

Noted in PR #293

Use the same trick as what is used in the [BitSetDocSet] (https://github.com/tantivy-search/tantivy/blob/78673172d001a0c4c7c73cdd3d9923fc43fc0312/src/query/bitset/mod.rs) which relies on TinySet

In a nutshell : instead of looping on the bitset to find the next thing alive, it uses a special cpu instruction to get the next bit set to 1 in a u64. The instruction is not free, but except if your bitset is EXTREMELY saturated it is generally faster.

https://github.com/tantivy-search/tantivy/pull/293#discussion-diff-186488306R373

Contributor guide

Tech stack
rust
Domain
backendperformance
Issue type
performance
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
3
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
1-3 hours
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
stale
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
mostly clear
Prerequisites
Rust proficiencyunderstanding of bitsets and low level optimizationfamiliarity with tantivy's query engine
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
30
Research direction
The issue suggests optimizing the segment readers alive doc iterator by adopting the same technique used in BitSetDocSet (https://github.com/tantivy search/tantivy/blob/78673172d001a0c4c7c73cdd3d9923fc43fc0312/src/query/bitset/mod.rs), which relies on TinySet. The key improvement is to use a CPU instruction (likely x86 BSF or TZCNT) to find the next set bit in a u64, instead of looping. The contributor should examine PR #293 (https://github.com/tantivy search/tantivy/pull/293) and the referenced discussion diff. The implementation should replace the current iteration logic in the segment reader's alive doc iterator with a TinySet based approach, ensuring correctness and performance gains.