Benchmark for lore storage fragment optimization - #103
Draft
Vazcore wants to merge 10 commits into
Draft
Conversation
write_fragmented called FastCDC::cut once per chunk, shipping the work to the compute pool, awaiting the result, and repeating. For a 1 GiB buffer with ~4 KiB chunks that is ~256k spawn dispatches, oneshot allocations and await yields — overhead that can rival the chunking work itself for small remaining chunks. Compute all FastCDC boundaries in a single compute_pool task by driving the FastCDC Iterator to completion. Fixed-size chunking has trivial per-step math so its boundaries are computed inline. The single-fragment fast path and the storage dispatch loop are unchanged. Verified by a new parity test that compares the batched boundaries against a reference FastCDC iteration on a 256 KiB random buffer, plus edge cases for empty, sub-min-size, all-zero, and non-aligned fixed-size inputs. Signed-off-by: Oleksii Habrusiev <alexgabrusev@gmail.com>
…nt-optimization Signed-off-by: Oleksii Habrusiev <alexgabrusev@gmail.com> # Conflicts: # lore-storage/src/fragment_engine.rs
…ation Signed-off-by: Oleksii Habrusiev <alexgabrusev@gmail.com>
Signed-off-by: Oleksii Habrusiev <alexgabrusev@gmail.com>
…c allocation in FastCDC chunking Signed-off-by: Oleksii Habrusiev <alexgabrusev@gmail.com>
…lore-storage-fragment-optimization
epic-lore-bot Bot
pushed a commit
that referenced
this pull request
Jul 20, 2026
When you store a big file in Lore, write_fragmented cuts it into chunks using FastCDC and ships each cut point to the compute pool separately, waits for the answer, then does it again. For a 1 GiB buffer with 64 KiB average chunks I think that's around 16k spawn dispatches, oneshot allocations and await yields. The actual chunking work is fast, but all that round-tripping added up to a real cost on medium and large writes I believe. ### Why A single chunking pass on the compute pool is enough - there's no reason to wait per boundary. ### How A single compute_pool task drives the FastCDC Iterator to completion and sends back a list of boundaries through one oneshot. The existing storage dispatch loop then iterates them locally. For fixed-size chunking the boundaries are just step arithmetic so those are computed inline. The Arc<FastCDC> wrapper is gone since we only touch the chunker once now. The single-fragment fast path, the JoinSet of storage tasks, hash-only mode and clone_buffer are all preserved, and the public signature of write_fragmented is unchanged. Behaviorally the output is identical to the old code, same cut points for the same input. ### Testing I added a small reference helper that does the standard FastCDC iteration synchronously and compared its output to the batched path on a 256 KiB random buffer. There's also coverage for empty input, sub-min-size input, an all-zero buffer (which defeats the rolling hash), and a non-aligned fixed-size case. cargo test -p lore-storage goes from 154 to 160 passing, clippy is clean with -D warnings, and the whole workspace still compiles. The fastcdc_batch_matches_reference test is the one that would catch a regression in the FastCDC version or a misconfiguration of the chunk sizes. ### Testing using Criterion Draft PR with criterion: #103 #### Micro-benchmark results Comparing the new batched approach against a simulation of the old per-chunk approach. Same FastCDC config as the real code: min=32KB, expected=64KB, threshold=256KB, Level1. #### What's being measured - **batched** — all chunk boundaries found in one compute-pool dispatch (PR approach) - **per-chunk** — one compute-pool spawn + oneshot channel per boundary (old approach) #### Results | Buffer size | Batched | Per-chunk | Speedup | |------------|---------|-----------|---------| | 64 KB | 19.0 µs | 19.8 µs | **1.04x** | | 1 MB | 222 µs | 327 µs | **1.47x** | | 16 MB | 3.79 ms | 5.53 ms | **1.46x** | | 64 MB | 15.4 ms | 21.5 ms | **1.40x** | #### Why it matters At 64 KB the overhead of a single boundary is too small to notice. But for real-world writes (1 MB and up) the batch approach is consistently **40-47% faster**. A 1 GB file with ~16K chunk boundaries saves ~16K spawn dispatches, oneshot allocations and await yields — replacing them with a single dispatch. Run locally: `cargo bench -p lore-storage --bench chunking` #### Raw log: ``` warning: lore-base@0.8.5-nightly: Failed to execute Lore to get revision information, no data extracted Finished `bench` profile [optimized + debuginfo] target(s) in 0.35s Running benches\chunking.rs (target\release\deps\chunking-b21acb5ed9e17d59.exe) Benchmarking fastcdc_chunking/batched_64KiB Benchmarking fastcdc_chunking/batched_64KiB: Warming up for 3.0000 s Benchmarking fastcdc_chunking/batched_64KiB: Collecting 100 samples in estimated 5.0093 s (263k iterations) Benchmarking fastcdc_chunking/batched_64KiB: Analyzing fastcdc_chunking/batched_64KiB time: [18.946 ┬╡s 19.007 ┬╡s 19.072 ┬╡s] change: [ΓêÆ5.5576% ΓêÆ4.0130% ΓêÆ2.6139%] (p = 0.00 < 0.05) Performance has improved. Found 2 outliers among 100 measurements (2.00%) 1 (1.00%) high mild 1 (1.00%) high severe Benchmarking fastcdc_chunking/per_chunk_64KiB Benchmarking fastcdc_chunking/per_chunk_64KiB: Warming up for 3.0000 s Benchmarking fastcdc_chunking/per_chunk_64KiB: Collecting 100 samples in estimated 5.0627 s (258k iterations) Benchmarking fastcdc_chunking/per_chunk_64KiB: Analyzing fastcdc_chunking/per_chunk_64KiB time: [19.718 ┬╡s 19.759 ┬╡s 19.802 ┬╡s] change: [ΓêÆ1.6628% ΓêÆ0.6054% +0.3810%] (p = 0.26 > 0.05) No change in performance detected. Found 9 outliers among 100 measurements (9.00%) 1 (1.00%) low severe 3 (3.00%) low mild 4 (4.00%) high mild 1 (1.00%) high severe Benchmarking fastcdc_chunking/batched_1MiB Benchmarking fastcdc_chunking/batched_1MiB: Warming up for 3.0000 s Benchmarking fastcdc_chunking/batched_1MiB: Collecting 100 samples in estimated 5.6211 s (25k iterations) Benchmarking fastcdc_chunking/batched_1MiB: Analyzing fastcdc_chunking/batched_1MiB time: [221.66 ┬╡s 222.25 ┬╡s 222.95 ┬╡s] change: [ΓêÆ17.166% ΓêÆ16.724% ΓêÆ16.093%] (p = 0.00 < 0.05) Performance has improved. Found 6 outliers among 100 measurements (6.00%) 5 (5.00%) high mild 1 (1.00%) high severe Benchmarking fastcdc_chunking/per_chunk_1MiB Benchmarking fastcdc_chunking/per_chunk_1MiB: Warming up for 3.0000 s Benchmarking fastcdc_chunking/per_chunk_1MiB: Collecting 100 samples in estimated 6.6377 s (20k iterations) Benchmarking fastcdc_chunking/per_chunk_1MiB: Analyzing fastcdc_chunking/per_chunk_1MiB time: [324.99 ┬╡s 326.69 ┬╡s 329.20 ┬╡s] change: [ΓêÆ3.5149% ΓêÆ2.9853% ΓêÆ2.2318%] (p = 0.00 < 0.05) Performance has improved. Found 7 outliers among 100 measurements (7.00%) 2 (2.00%) high mild 5 (5.00%) high severe Benchmarking fastcdc_chunking/batched_16MiB Benchmarking fastcdc_chunking/batched_16MiB: Warming up for 3.0000 s Benchmarking fastcdc_chunking/batched_16MiB: Collecting 100 samples in estimated 5.3552 s (1400 iterations) Benchmarking fastcdc_chunking/batched_16MiB: Analyzing fastcdc_chunking/batched_16MiB time: [3.7744 ms 3.7851 ms 3.7964 ms] change: [+3.0532% +3.4713% +3.8820%] (p = 0.00 < 0.05) Performance has regressed. Found 2 outliers among 100 measurements (2.00%) 2 (2.00%) high mild Benchmarking fastcdc_chunking/per_chunk_16MiB Benchmarking fastcdc_chunking/per_chunk_16MiB: Warming up for 3.0000 s Benchmarking fastcdc_chunking/per_chunk_16MiB: Collecting 100 samples in estimated 5.4254 s (1000 iterations) Benchmarking fastcdc_chunking/per_chunk_16MiB: Analyzing fastcdc_chunking/per_chunk_16MiB time: [5.4582 ms 5.5349 ms 5.6246 ms] change: [+5.0870% +6.5725% +8.0610%] (p = 0.00 < 0.05) Performance has regressed. Found 15 outliers among 100 measurements (15.00%) 7 (7.00%) high mild 8 (8.00%) high severe Benchmarking fastcdc_chunking/batched_64MiB Benchmarking fastcdc_chunking/batched_64MiB: Warming up for 3.0000 s Benchmarking fastcdc_chunking/batched_64MiB: Collecting 100 samples in estimated 6.2283 s (400 iterations) Benchmarking fastcdc_chunking/batched_64MiB: Analyzing fastcdc_chunking/batched_64MiB time: [15.341 ms 15.424 ms 15.513 ms] change: [+1.2826% +1.9061% +2.6672%] (p = 0.00 < 0.05) Performance has regressed. Found 2 outliers among 100 measurements (2.00%) 1 (1.00%) high mild 1 (1.00%) high severe Benchmarking fastcdc_chunking/per_chunk_64MiB Benchmarking fastcdc_chunking/per_chunk_64MiB: Warming up for 3.0000 s Benchmarking fastcdc_chunking/per_chunk_64MiB: Collecting 100 samples in estimated 6.5061 s (300 iterations) Benchmarking fastcdc_chunking/per_chunk_64MiB: Analyzing fastcdc_chunking/per_chunk_64MiB time: [21.428 ms 21.534 ms 21.675 ms] change: [+2.4232% +2.9622% +3.6167%] (p = 0.00 < 0.05) Performance has regressed. Found 6 outliers among 100 measurements (6.00%) 3 (3.00%) high mild 3 (3.00%) high severe Benchmarking fixed_size/16MiB Benchmarking fixed_size/16MiB: Warming up for 3.0000 s Benchmarking fixed_size/16MiB: Collecting 100 samples in estimated 5.0009 s (20M iterations) Benchmarking fixed_size/16MiB: Analyzing fixed_size/16MiB time: [250.77 ns 251.66 ns 252.62 ns] change: [+0.3309% +0.9096% +1.4554%] (p = 0.00 < 0.05) Change within noise threshold. Found 4 outliers among 100 measurements (4.00%) 4 (4.00%) high mild ``` ``` Imported-PR: #61 Imported-From: 8d7aab9 Imported-Base: a00b1c4 Imported-Merge: 30a8d7b Imported-Author: Oleksii Habrusiev (Vazcore) Signed-off-by: Oleksii Habrusiev <alexgabrusev@gmail.com> GH-URL: #61 ``` Lore-RevId: 323 Lore-Signature: b9e7f3161fda891797a950e7e04b1de54e06b664fabd8e179666b5927175cc3f
ajcarberry
added
the
area:core
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
2 participants
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.
Micro-benchmark results
Comparing the new batched approach against a simulation of the old per-chunk approach.
Same FastCDC config as the real code: min=32KB, expected=64KB, threshold=256KB, Level1.
What's being measured
Results
Fixed-size chunking (for reference): 252 ns at 16 MB (just arithmetic, no hashing).
Why it matters
At 64 KB the overhead of a single boundary is too small to notice. But for real-world
writes (1 MB and up) the batch approach is consistently 40-47% faster. A 1 GB file
with ~16K chunk boundaries saves ~16K spawn dispatches, oneshot allocations and await
yields — replacing them with a single dispatch.
Run locally:
cargo bench -p lore-storage --bench chunkingRaw log: