20,000
Workload Lines
1.04MB
Fixture Size
7
Repetitions
100%
Byte-Identical
What is measured

Every core runs the identical command hushline mute -- cat <workload>. The workload is a generated 20,000-line (~1.04 MB) log mixing ANSI escapes, sk-… and AKIA… secrets, and doubled whitespace, so each core exercises ANSI stripping, redaction, and the silence rules on the same bytes. The harness records the wall-clock median of seven repetitions per core and verifies all four cores produce byte-identical output before reporting any timing.

# Reproduce locally
bash benchmarks/run_core_benchmarks.sh
# results → benchmarks/results/core_latency.json
Results

11th Gen Intel Core i5-11600K @ 3.90 GHz (12 logical cores), 20,000-line workload, 7 repetitions, median wall-clock time. Cross-core output parity: identical.

Rust
5.07 ms
Go
12.85 ms
Node
52.94 ms
Python
59.85 ms
CoreMedianMinThroughputRelativeParity
Rust5.07 ms4.91 ms204.7 MB/s1.0×Byte-identical
Go12.85 ms11.48 ms80.7 MB/s2.5×Byte-identical
Node52.94 ms46.61 ms19.6 MB/s10.4×Byte-identical
Python59.85 ms57.10 ms17.3 MB/s11.8×Byte-identical

“Relative” is the median time divided by the fastest core's median. These figures are hardware- and workload-specific, not universal performance claims. Reproduce them from the public repository before making deployment decisions.