About this question
Medium · low_latency · Quant Developer interview question · restrict, alias-analysis, vectorization, low_latency, simd, cache
In high-frequency trading, risk engines must update thousands of prices per microsecond. Compilers often fail to auto-vectorize simple array operations like dsti = srci factor due to potential pointer aliasing between dst and src. Using the __restrict__ keyword (a C/C++ extension) informs the compiler that the pointers do not overlap, unlocking significant performance gains through SIMD instructions. Task Implement the function update_prices(double __restrict__ dst, const double __restrict__ s
Keep practicing