# NexusFIX Benchmarks

> C++23 ultra-low-latency FIX engine performance snapshots.

## Highlights

- ~3× faster than QuickFIX in message parsing throughput
- Execution report parse: ~246ns (illustrative)
- Zero allocations on hot path; arena/pool allocators elsewhere

## Techniques

- AVX2 SIMD tag parsing (32 bytes/cycle)
- Pre-allocated message pools; wait-free order book primitives
- Modern C++ (std::expected, std::span, concepts) for clarity and safety

## Environment

- x86_64 with AVX2; Linux kernel with io_uring (DEFER_TASKRUN) recommended
- Registered buffers; multishot recv for I/O efficiency

---

Updated: 2026-04-10