Google's portable Quicksort hits 1 GB/s, 19x faster than std::sort
We've created the first vectorized Quicksort
Google open-sourced a vectorized Quicksort that sorts 1 million numbers at up to 1.1 GB/s on a single Skylake core, 9–19x faster than C++ std::sort. Using Highway's portable SIMD, it runs on six instruction sets across three architectures—including AVX2, AVX-512, and Arm NEON—and outperforms architecture-specific implementations while supporting 16- to 128-bit inputs.
Previously, sorting has been considered expensive. We are interested to see what new applications and capabilities will be unlocked by being able to sort at 1 GB/s on a single CPU core.