Benchmarks

ChadScript compiles to native binaries via LLVM IR — no runtime, no JIT warmup, no cold start penalty.

These benchmarks compare ChadScript against C (clang -O2 -march=native), Go, and Node.js across compute, I/O, and real-world workloads.

Methodology

• Machine: Apple Silicon (M-series, arm64 native), dedicated hardware, all runtimes tested sequentially
• Optimization: ChadScript and C compiled with -O2 -march=native
• Sampling: Each compute benchmark runs N=10 times. The reported value is the median of the 10 samples, annotated with a 95% bootstrap confidence interval (2000 resamples, random.seed(0xC4AD) for reproducibility).
• Ties: Two languages are treated as tied when their 95% CIs overlap — i.e. the difference between them is not statistically distinguishable at the 95% level. This prevents spurious medal-flipping when jitter produces a 2-3% gap that's inside both intervals.
• Cold start: Average of 50 launches (aggregate measurement, not per-sample).
• Timing: Wall-clock time via clock_gettime / Date.now() / time.Now().

The CI whiskers on the bars show the 95% confidence interval. A narrow whisker means the samples were very consistent; a wide one means the runtime is variable and the point estimate should be trusted less. Bars whose CIs visibly overlap are statistically tied — ranking between them is meaningless.

Reproduce locally:

BENCH_RUNS=10 ./benchmarks/run.sh

Source code for all benchmarks: github.com/cs01/ChadScript/tree/main/benchmarks