golang-github-minio-blake2b-simd-dev 0.0~git20160723.3f5f724-2 (i386 binary) in ubuntu kinetic

 BLAKE2b-SIMD Pure Go implementation of BLAKE2b using SIMD
 optimizations. Introduction This package was initially based
 on the pure go BLAKE2b (https://github.com/dchest/blake2b)
 implementation of Dmitry Chestnykh and merged with the (cgo
 dependent) AVX optimized BLAKE2 (https://github.com/codahale/blake2)
 implementation (which in turn is based on the official implementation
 (https://github.com/BLAKE2/BLAKE2). It does so by using Go's Assembler
 (https://golang.org/doc/asm) for amd64 architectures with a golang only
 fallback for other architectures.
 .
 In addition to AVX there is also support for AVX2 as well as SSE. Best
 performance is obtained with AVX2 which gives roughly a 4X performance
 increase approaching hashing speeds of 1GB/sec on a single core.
 .
 BLAKE2b is a hashing algorithm that operates on 64-bit integer values. The
 AVX2 version uses the 256-bit wide YMM registers in order to essentially
 process four operations in parallel. AVX and SSE operate on 128-bit
 values simultaneously (two operations in parallel). Below are excerpts
 from compressAvx2_amd64.s, compressAvx_amd64.s, and compress_generic.go
 respectively.
 .
     VPADDQ YMM0,YMM0,YMM1 /* v0 += v4, v1 += v5, v2 += v6, v3 += v7 */
 .
     VPADDQ XMM0,XMM0,XMM2 /* v0 += v4, v1 += v5 */ VPADDQ
     XMM1,XMM1,XMM3 /* v2 += v6, v3 += v7 */
 .
     v0 += v4 v1 += v5 v2 += v6 v3 += v7