SIMDxorshift_cpp

A C++ implementation of Daniel Lemire's SIMD (single instruction multiple data) xorshift128+ implementation.

Getting Started

A version of Lemire's benchmarking function is included to allow testing of each implementation.

All you should need to do is

make
./simd_xor

And you should get something like

Testing function : xor128
 2.28 cycles per operation
Testing function : xor128_simd
 0.70 cycles per operation
Testing function : xor128_simd_two
 0.62 cycles per operation
Testing function : xor128_simd_four
 0.75 cycles per operation
Testing function : aes_dragontamer
 0.58 cycles per operation

Requirements

A processor that supports AVX2. An AVX-512 version is available for processors supporting these extensions.

References

Daniel Lemire - for the original C code - https://github.com/lemire/SIMDxorshift

Sebastiano Vigna - xorshift128+ implementation - http://xorshift.di.unimi.it/xorshift128plus.c

Melissa O'Neil - for the functions allowing seeding of the PRNGs - http://www.pcg-random.org/posts/developing-a-seed_seq-alternative.html