It uses a sharded architecture (one thread per symbol group) to avoid mutex contention during matching. Recently moved to int64_t fixed-point pricing and added memory compaction to prevent leaks in long-running processes.
It uses a sharded architecture (one thread per symbol group) to avoid mutex contention during matching. Recently moved to int64_t fixed-point pricing and added memory compaction to prevent leaks in long-running processes.