gistfile1.txt

Comparison to rc4_amd64.s on Core 2 Duo

benchmark           old ns/op    new ns/op    delta
BenchmarkRC4_128          820          604  -26.34%
BenchmarkRC4_1K          6420         4444  -30.78%
BenchmarkRC4_8K         50373        35335  -29.85%

benchmark            old MB/s     new MB/s  speedup
BenchmarkRC4_128       156.00       211.80    1.36x
BenchmarkRC4_1K        159.49       230.40    1.44x
BenchmarkRC4_8K        160.72       229.12    1.43x

rc4_unsafe.go

// Copyright 2013 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.

// +build amd64

package rc4

 
import "unsafe"
 
// XORKeyStream sets dst to the result of XORing src with the key stream.
// Dst and src may be the same slice but otherwise should not overlap.
func (c *Cipher) XORKeyStream(dst, src []byte) {
  // Go unsafe version.
	if len(dst) < len(src) {
		panic("rc4: dst is shorter than src")
	}
	sp := uintptr(unsafe.Pointer(&src[0]))
	dp := uintptr(unsafe.Pointer(&dst[0]))
	send := sp + uintptr(len(src))

	i, j := c.i + 1, c.j
	s := &c.s
	a := s[i]

	for sp < send {
		j += a
		b := s[j]
		s[i], s[j] = b, a
		*(*byte)(unsafe.Pointer(dp)) = s[a+b] ^ *(*byte)(unsafe.Pointer(sp))
		sp++
		dp++
		i += 1
		a = s[i]
	}
	c.i, c.j = i-1, j
}