Read this first. These are synthetic microbenchmarks of local harness operations, run on shared CI runners with no warm-up. They quantify runtime overhead in the recorded environment only. They do not measure developer productivity, code quality or statistical significance.
5 repetition(s) of harness benchmark run; values are the median across repetitions of each run's median (p95 likewise).
Latency per operation
Median and p95 latency per operation (log scale, milliseconds)
medianp95
Latency per operation across repetitions
Operation
Iterations/run
Median ms
p95 ms
Ops/s
Min–max of medians
Std. dev.
State transition
1000
0.00125
0.00128
777,704.2
0.00124 – 0.00133
3.68e-05
Gate evaluation
1000
0.002
0.00204
489,138.2
0.00197 – 0.00215
7.31e-05
Canonical hash (digest)
1000
0.0112
0.0124
85,834.4
0.0107 – 0.0116
0.000321
Artifact put (deduplicated)
1000
0.0729
0.118
12,320.9
0.0726 – 0.077
0.00187
Event append (hash chain)
1000
0.257
0.519
3,195.4
0.248 – 0.268
0.00803
Process launch, direct
20
12.394
13.416
80.507
12.117 – 12.809
0.27
Process launch, governed
20
13.122
13.889
75.940
13.033 – 13.445
0.199
Direct versus governed process launch
Process launch median: governed overhead 0.639 ms (5.16 %)
median
Governance overhead per repetition. Local process-launch overhead only; excludes validators, agents and human wait.
Repetition
Absolute median overhead (ms)
Relative median overhead (%)
1
0.614
4.79
2
1.328
10.96
3
0.639
5.16
4
0.86
7.04
5
0.624
4.99
Throughput
Operations per second (log scale; higher is faster)
ops/s
Dispersion across repetitions
Median of each repetition (one dot per repetition, log scale)
Per-repetition medians
Operation
Run 1 median ms
Run 2 median ms
Run 3 median ms
Run 4 median ms
Run 5 median ms
State transition
0.00133
0.00125
0.00127
0.00124
0.00125
Gate evaluation
0.00215
0.00197
0.00202
0.002
0.00198
Canonical hash (digest)
0.0116
0.0107
0.0112
0.011
0.0114
Artifact put (deduplicated)
0.077
0.0726
0.0729
0.0726
0.0741
Event append (hash chain)
0.255
0.248
0.268
0.266
0.257
Process launch, direct
12.809
12.117
12.394
12.213
12.499
Process launch, governed
13.423
13.445
13.033
13.072
13.122
Thesis reference point
Transcribed from the thesis, chapter 6 (implementation and demonstration), for the v0.8.0 cut: 1,000 iterations per operation, no warm-up, 20 samples for process launches. The raw evidence file is not part of this repository.
Reference values recorded in the thesis; environments differ, compare orders of magnitude only.