我认为下面是我所知道的最简单的方法。MSB是输入，LSB是“反向”输出:

unsigned char rev(char MSB) {
    unsigned char LSB=0;  // for output
    _FOR(i,0,8) {
        LSB= LSB << 1;
        if(MSB&1) LSB = LSB | 1;
        MSB= MSB >> 1;
    }
    return LSB;
}

//    It works by rotating bytes in opposite directions. 
//    Just repeat for each byte.

高效意味着吞吐量或延迟。

从头到尾，看看安德斯·塞德罗尼厄斯的回答，很好。

为了降低延迟，我推荐以下代码:

uint32_t reverseBits( uint32_t x )
{
#if defined(__arm__) || defined(__aarch64__)
    __asm__( "rbit %0, %1" : "=r" ( x ) : "r" ( x ) );
    return x;
#endif
    // Flip pairwise
    x = ( ( x & 0x55555555 ) << 1 ) | ( ( x & 0xAAAAAAAA ) >> 1 );
    // Flip pairs
    x = ( ( x & 0x33333333 ) << 2 ) | ( ( x & 0xCCCCCCCC ) >> 2 );
    // Flip nibbles
    x = ( ( x & 0x0F0F0F0F ) << 4 ) | ( ( x & 0xF0F0F0F0 ) >> 4 );

    // Flip bytes. CPUs have an instruction for that, pretty fast one.
#ifdef _MSC_VER
    return _byteswap_ulong( x );
#elif defined(__INTEL_COMPILER)
    return (uint32_t)_bswap( (int)x );
#else
    // Assuming gcc or clang
    return __builtin_bswap32( x );
#endif
}

编译器输出:https://godbolt.org/z/5ehd89

当然，玩弄比特的黑客的明显来源是: http://graphics.stanford.edu/~seander/bithacks.html#BitReverseObvious

通用的

C代码。以1字节输入数据num为例。

    unsigned char num = 0xaa;   // 1010 1010 (aa) -> 0101 0101 (55)
    int s = sizeof(num) * 8;    // get number of bits
    int i, x, y, p;
    int var = 0;                // make var data type to be equal or larger than num

    for (i = 0; i < (s / 2); i++) {
        // extract bit on the left, from MSB
        p = s - i - 1;
        x = num & (1 << p);
        x = x >> p;
        printf("x: %d\n", x);

        // extract bit on the right, from LSB
        y = num & (1 << i);
        y = y >> i;
        printf("y: %d\n", y);

        var = var | (x << i);       // apply x
        var = var | (y << p);       // apply y
    }

    printf("new: 0x%x\n", new);

下面这个怎么样:

    uint reverseMSBToLSB32ui(uint input)
    {
        uint output = 0x00000000;
        uint toANDVar = 0;
        int places = 0;

        for (int i = 1; i < 32; i++)
        {
            places = (32 - i);
            toANDVar = (uint)(1 << places);
            output |= (uint)(input & (toANDVar)) >> places;

        }


        return output;
    }

小而简单(不过只有32位)。

这个线程引起了我的注意，因为它处理了一个简单的问题，即使对于现代CPU也需要大量的工作(CPU周期)。有一天我也站在那里，有同样的¤#%“#”问题。我得翻几百万字节。然而，我知道我所有的目标系统都是基于现代英特尔的，所以让我们开始优化到极致!!

所以我使用了Matt J的查找代码作为基础。我正在基准测试的系统是i7 haswell 4700eq。

Matt J的查找位翻转400亿字节:大约0.272秒。

然后我继续尝试，看看英特尔的ISPC编译器是否可以向量化反向的算术。c。

我不打算在这里用我的发现来烦你，因为我尝试了很多来帮助编译器找到东西，无论如何，我最终得到了大约0.15秒的性能来bitflip 400亿字节。这是一个伟大的减少，但对于我的应用程序，这仍然是方式方式太慢。

所以人们让我展示世界上最快的基于英特尔的bitflipper。定时:

时间到bitflip 400000000字节:0.050082秒!!!!!

// Bitflip using AVX2 - The fastest Intel based bitflip in the world!!
// Made by Anders Cedronius 2014 (anders.cedronius (you know what) gmail.com)

#include <stdio.h>
#include <stdlib.h>
#include <math.h>
#include <omp.h>

using namespace std;

#define DISPLAY_HEIGHT  4
#define DISPLAY_WIDTH   32
#define NUM_DATA_BYTES  400000000

// Constants (first we got the mask, then the high order nibble look up table and last we got the low order nibble lookup table)
__attribute__ ((aligned(32))) static unsigned char k1[32*3]={
        0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,0x0f,
        0x00,0x08,0x04,0x0c,0x02,0x0a,0x06,0x0e,0x01,0x09,0x05,0x0d,0x03,0x0b,0x07,0x0f,0x00,0x08,0x04,0x0c,0x02,0x0a,0x06,0x0e,0x01,0x09,0x05,0x0d,0x03,0x0b,0x07,0x0f,
        0x00,0x80,0x40,0xc0,0x20,0xa0,0x60,0xe0,0x10,0x90,0x50,0xd0,0x30,0xb0,0x70,0xf0,0x00,0x80,0x40,0xc0,0x20,0xa0,0x60,0xe0,0x10,0x90,0x50,0xd0,0x30,0xb0,0x70,0xf0
};

// The data to be bitflipped (+32 to avoid the quantization out of memory problem)
__attribute__ ((aligned(32))) static unsigned char data[NUM_DATA_BYTES+32]={};

extern "C" {
void bitflipbyte(unsigned char[],unsigned int,unsigned char[]);
}

int main()
{

    for(unsigned int i = 0; i < NUM_DATA_BYTES; i++)
    {
        data[i] = rand();
    }

    printf ("\r\nData in(start):\r\n");
    for (unsigned int j = 0; j < 4; j++)
    {
        for (unsigned int i = 0; i < DISPLAY_WIDTH; i++)
        {
            printf ("0x%02x,",data[i+(j*DISPLAY_WIDTH)]);
        }
        printf ("\r\n");
    }

    printf ("\r\nNumber of 32-byte chunks to convert: %d\r\n",(unsigned int)ceil(NUM_DATA_BYTES/32.0));

    double start_time = omp_get_wtime();
    bitflipbyte(data,(unsigned int)ceil(NUM_DATA_BYTES/32.0),k1);
    double end_time = omp_get_wtime();

    printf ("\r\nData out:\r\n");
    for (unsigned int j = 0; j < 4; j++)
    {
        for (unsigned int i = 0; i < DISPLAY_WIDTH; i++)
        {
            printf ("0x%02x,",data[i+(j*DISPLAY_WIDTH)]);
        }
        printf ("\r\n");
    }
    printf("\r\n\r\nTime to bitflip %d bytes: %f seconds\r\n\r\n",NUM_DATA_BYTES, end_time-start_time);

    // return with no errors
    return 0;
}

printf是用来调试的。

这是主要的工作:

bits 64
global bitflipbyte

bitflipbyte:    
        vmovdqa     ymm2, [rdx]
        add         rdx, 20h
        vmovdqa     ymm3, [rdx]
        add         rdx, 20h
        vmovdqa     ymm4, [rdx]
bitflipp_loop:
        vmovdqa     ymm0, [rdi] 
        vpand       ymm1, ymm2, ymm0 
        vpandn      ymm0, ymm2, ymm0 
        vpsrld      ymm0, ymm0, 4h 
        vpshufb     ymm1, ymm4, ymm1 
        vpshufb     ymm0, ymm3, ymm0         
        vpor        ymm0, ymm0, ymm1
        vmovdqa     [rdi], ymm0
        add     rdi, 20h
        dec     rsi
        jnz     bitflipp_loop
        ret

代码占用32个字节，然后屏蔽掉蚕食。高啃角右移了4。然后使用vpshufb和ymm4 / ymm3作为查找表。我可以使用一个单独的查找表，但我将不得不在ORing再次一起啃啃之前向左移动。

还有更快的翻转比特的方法。但我被绑定到单线程和CPU，所以这是我能实现的最快速度。你能做一个快一点的版本吗?

关于使用Intel C/ c++编译器内在等效命令，请不要发表任何评论…