First steps in SIMD. Implementing STL algorithms with AVX2 vector extensions

RU / День 1 / 12:30 / Зал 1

Denis saw some good talk about SIMD but they didn't talk about how to use SIMD to write algorithms. For strlen works, for example? What about vectorized sorting from std::unseq?

During this session, the speaker will briefly talk about the implementation of some algorithms using AVX2 — x86 SIMD extensions. All algorithms are carefully implemented and measured. The audience will get a general idea of how SIMD algorithms work, what is possible, and what kind of acceleration can be achieved.

Go to presentation