svb 0.2.0

Pure-Rust StreamVByte: integer compression for u16/u32/u64 with SIMD decode (AVX2, SSSE3, NEON)
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103

name: Benchmarks

# Manual trigger only — run via Actions → Benchmarks → Run workflow.
# Benchmark jobs run in parallel; each writes its own GitHub Step Summary.
on:
  workflow_dispatch:

# Cancel any in-progress run for the same ref so we don't queue up stale results.
concurrency:
  group: bench-${{ github.ref }}
  cancel-in-progress: true

env:
  CARGO_TERM_COLOR: always
  # Reduced sample count / timing keeps each feature-flag run to ~5 min.
  # Increase locally for publication-quality numbers.
  CRITERION_ARGS: --output-format bencher --sample-size 20 --warm-up-time 1 --measurement-time 3

jobs:
  # ── x86-64: scalar → SSSE3 → AVX2 ────────────────────────────────────────────
  bench-x86_64:
    name: bench / x86-64
    runs-on: ubuntu-latest

    steps:
      - uses: actions/checkout@v4

      - uses: dtolnay/rust-toolchain@stable

      - uses: Swatinem/rust-cache@v2
        with:
          key: bench-x86_64

      # Each cargo bench invocation compiles a separate bench binary (different
      # feature set) and immediately runs it.  rust-cache keeps incremental
      # artifacts so subsequent runs with overlapping features are faster.

      - name: Bench — scalar (no SIMD)
        run: |
          cargo bench --bench decode -- ${{ env.CRITERION_ARGS }} 2>&1 | tee results-scalar.txt

      - name: Bench — SSSE3
        run: |
          cargo bench --bench decode --features simd-ssse3 -- ${{ env.CRITERION_ARGS }} 2>&1 | tee results-ssse3.txt

      - name: Bench — AVX2
        run: |
          cargo bench --bench decode --features simd-avx2 -- ${{ env.CRITERION_ARGS }} 2>&1 | tee results-avx2.txt

      - name: Generate step summary
        run: |
          python3 scripts/bench_summary.py \
            --arch x86-64 \
            scalar:results-scalar.txt \
            ssse3:results-ssse3.txt \
            avx2:results-avx2.txt \
            >> "$GITHUB_STEP_SUMMARY"

      - name: Upload raw results
        uses: actions/upload-artifact@v4
        if: always()
        with:
          name: bench-results-x86_64
          path: results-*.txt
          retention-days: 30

  # ── AArch64: scalar → NEON ────────────────────────────────────────────────────
  bench-aarch64:
    name: bench / aarch64
    runs-on: ubuntu-24.04-arm

    steps:
      - uses: actions/checkout@v4

      - uses: dtolnay/rust-toolchain@stable

      - uses: Swatinem/rust-cache@v2
        with:
          key: bench-aarch64

      - name: Bench — scalar (no SIMD)
        run: |
          cargo bench --bench decode -- ${{ env.CRITERION_ARGS }} 2>&1 | tee results-scalar.txt

      - name: Bench — NEON
        run: |
          cargo bench --bench decode --features simd-neon -- ${{ env.CRITERION_ARGS }} 2>&1 | tee results-neon.txt

      - name: Generate step summary
        run: |
          python3 scripts/bench_summary.py \
            --arch aarch64 \
            scalar:results-scalar.txt \
            neon:results-neon.txt \
            >> "$GITHUB_STEP_SUMMARY"

      - name: Upload raw results
        uses: actions/upload-artifact@v4
        if: always()
        with:
          name: bench-results-aarch64
          path: results-*.txt
          retention-days: 30