Optimizing a Spin-Lock

Automated excerpt

Two threads take 61. 5 ns, twenty times as long. Four take 246 ns. $ . /benchmark --benchmark_filter='V1>' from each other. Four threads gain 8%, 131 ns to 120 ns. $ . /benchmark --benchmark_filter='V3>' L1-d misses fall from 21. 16% to 17. 31%, branches from 7. 43% to 3. 72%.

Selected automatically from source text; not independently written or fact-checked. Read the original for full context.