String

What this benchmark does

This benchmark generates a deterministic source text of a bit more than 1 MiB by concatenating pseudo-randomly selected UTF-8 or UTF-16 words from a fixed word list. It then measures how fast each implementation can:

• split the source text into words,
• re-wrap those words into lines below 80 characters, and
• hash the resulting wrapped text to verify the output stays correct.

The workload is repeated for the configured iteration count. The benchmark therefore mainly measures string splitting, concatenation, line-wrapping logic, and result verification on a large in-memory text.

String-specific variants

• C (UTF-16): Alternative C implementation using fixed-width UTF-16 code units instead of UTF-8 bytes.
• Rust (UTF-16): Alternative Rust implementation working on UTF-16 code units rather than byte-oriented string handling.

Benchmark system

• Apple M1 Pro (arm64)
• 10 CPU cores

Use ../../run -b string to measure this benchmark. It pre-builds compiled implementations and measures each benchmark command with /usr/bin/time.

Results