We benchmarked every common open-source FIFO matching-engine implementation we could find, deduplicated, and driven through a single C-ABI harness on one identical workload - 246 distinct engines across more than twenty programming languages. The workload doubles as a byte-identical correctness oracle, and it is unforgiving: only 47 are correct as shipped; it surfaced 267+ distinct defects across the other 199, and we filed 172+ GitHub issues upstream. Our engine leads the 160 that reproduce the consensus by +25 M/s (roughly 4x from the second best) on worst-case throughput, and outruns every engine in the survey written by a professional trading-industry engineer. A single CPU core sustains 33.2~million order messages per second at sub-microsecond median end-to-end host-path latency; a 96-core commodity server (~$1,630/month) sustains ~640 million messages per second across 10,000 symbols - over 20x the provisioned capacity of the U.S. consolidated quote feed.

翻译：暂无翻译