Savvy vs Bazel format benchmark results — services-pilot (2026-05-22)

benchmark-results.txt

Savvy vs Bazel Format Benchmark — Combined Results
===================================================

Run on M4 Max MacBook Pro, 128GB RAM
Savvy 0.10.1 (--no-progress build), Bazel 9.0.2
Date: 2026-05-22

3 runs per scenario, median reported.
Two separate runs combined below (SMALL + LARGE_BAZEL from one run,
LARGE from another).


=== Run 1: SMALL + LARGE_BAZEL ===

  Savvy vs Bazel Format Benchmark
  ================================

Started: 2026-05-22 15:50:45 EDT
Commit:  b51eb2734a1d0 (mattbrown/savvy-benchmark)
Runs per scenario: 3 (median reported)

Setting up benchmark worktree at /var/folders/m1/5018nxw5655fd55ql3xqqpjm0000gn/T//savvy-bench-66347...

Sanity check: bazel run //:format in worktree...
  bazel: OK
  savvy: OK

━━━ SMALL set (4 files) ━━━

Warming caches...
[1/8] warm
  bazel (warm)                    4.7s  4.7s  5.0s  → 4.7s median
  savvy (warm)                    6.1s  1.5s  1.6s  → 1.6s median
[2/8] cold server — after bazel shutdown
  bazel (cold server)             10.0s  8.7s  10.7s  → 10.0s median
  savvy (cold server)             2.0s  1.6s  1.5s  → 1.6s median
[3/8] stale analysis — checkout 2-week-old commit then back
  bazel (stale analysis)          11.1s  10.9s  9.5s  → 10.9s median
  savvy (stale analysis)          2.6s  2.2s  2.1s  → 2.2s median
[4/8] prior state expunged — all caches wiped
  bazel (prior state expunged)    64.7s  65.7s  84.6s  → 65.7s median
  savvy (prior state expunged)    19.6s  19.0s  20.4s  → 19.6s median

━━━ LARGE_BAZEL set (18 files) ━━━

Warming caches...
[5/8] warm
  bazel (warm)                    4.6s  4.4s  4.2s  → 4.4s median
  savvy (warm)                    3.7s  3.7s  3.6s  → 3.7s median
[6/8] cold server — after bazel shutdown
  bazel (cold server)             10.8s  10.2s  9.8s  → 10.2s median
  savvy (cold server)             9.0s  8.5s  9.5s  → 9.0s median
[7/8] stale analysis — checkout 2-week-old commit then back
  bazel (stale analysis)          10.8s  11.8s  10.7s  → 10.8s median
  savvy (stale analysis)          11.9s  9.5s  10.4s  → 10.4s median
[8/8] prior state expunged — all caches wiped
  bazel (prior state expunged)    72.3s  102.7s  85.7s  → 85.7s median
  savvy (prior state expunged)    58.4s  39.7s  59.9s  → 58.4s median

Finished: 2026-05-22 16:15:44 EDT
Done.

Cleaning up worktree...

=== Run 2: LARGE (no bazel deps) ===

  Savvy vs Bazel Format Benchmark
  ================================

Started: 2026-05-22 14:57:48 EDT
Commit:  865c0e1083880 (mattbrown/savvy-benchmark)
Runs per scenario: 3 (median reported)

Setting up benchmark worktree at /var/folders/m1/5018nxw5655fd55ql3xqqpjm0000gn/T//savvy-bench-42079...

Sanity check: bazel run //:format in worktree...
  bazel: OK
  savvy: OK

━━━ LARGE set (17 files) ━━━

Warming caches...
[1/4] warm
  bazel (warm)                    5.9s  4.9s  4.2s  → 4.9s median
  savvy (warm)                    7.8s  3.1s  2.9s  → 3.1s median
[2/4] cold server — after bazel shutdown
  bazel (cold server)             11.1s  9.9s  9.9s  → 9.9s median
  savvy (cold server)             4.2s  2.8s  3.0s  → 3.0s median
[3/4] stale analysis — checkout 2-week-old commit then back
  bazel (stale analysis)          10.9s  11.6s  11.1s  → 11.1s median
  savvy (stale analysis)          4.3s  3.3s  3.2s  → 3.3s median
[4/4] prior state expunged — all caches wiped
  bazel (prior state expunged)    112.4s  76.5s  63.7s  → 76.5s median
  savvy (prior state expunged)    69.3s  43.4s  33.3s  → 43.4s median

Finished: 2026-05-22 15:11:42 EDT
Done.

Cleaning up worktree...