⚡ Bolt: Optimize decompression for offset 24#228
Conversation
Optimized `decompress_bmi2` for `offset == 24` by removing the `alignr` dependency chain. The pattern for offset 24 repeats every 48 bytes (3 vectors). The implementation now precomputes these 3 vectors (`v1`, `v2`, `v0`) using `unpacklo` and `alignr` and stores them in an unrolled loop. This breaks the loop-carried dependency found in the previous implementation, allowing better pipelining and ILP. Benchmarks show a ~32% improvement in throughput for offset 24 decompression. Co-authored-by: 404Setup <[email protected]>
|
👋 Jules, reporting for duty! I'm here to lend a hand with this pull request. When you start a review, I'll add a 👀 emoji to each comment to let you know I've read it. I'll focus on feedback directed at me and will do my best to stay out of conversations between you and other bots or reviewers to keep the noise down. I'll push a commit with your requested changes shortly after. Please note there might be a delay between these steps, but rest assured I'm on the job! For more direct control, you can switch me to Reactive Mode. When this mode is on, I will only act on comments where you specifically mention me with New to Jules? Learn more at jules.google/docs. For security, I will only act on instructions from the user who triggered this task. |
1 participant
💡 What: Optimized the decompression hot path for
offset == 24insrc/decompress/x86.rs.🎯 Why: The previous implementation used an
alignrchain with loop-carried dependencies, limiting throughput.📊 Impact: Increases throughput for offset 24 by ~32% (7.73 GiB/s -> 10.20 GiB/s).
🔬 Measurement: Run
cargo bench --bench bench_main "Decompress offset24"to verify.PR created automatically by Jules for task 3244999817098573056 started by @404Setup