描述 ：
我实现了以下类 LabSetInt64（见下面的代码）。
这里的目标是尽可能快地处理大量的大整数（最大为 10M）。我的主要要求集中在：
- !Crucial : 尽可能快地获得集合的大小/基数
- !Important : 能够非常快速地迭代一个集合
所以，从下面的实现开始，还有两点想和大家讨论。
“popcount()”惰性实现
int LabSetInt64::popcount(uint64_t x) {
int count;
for (count=0; x; count++)
x &= x-1;
return count;
}
我知道我选择的实现是市场上最慢的实现之一，但我自己找不到更好的实现。所以，如果你能指点我一个更好的（当然是 64 位）......
我听说过一种非常有效的计算基数的算法：“MIT HAKMEM 169”算法>>
// MIT HAKMEM.
// MIT HAKMEM Count is funky. Consider a 3 bit number as being 4a+2b+c. If we
// shift it right 1 bit, we have 2a+b. Subtracting this from the original gives
// 2a+b+c. If we right-shift the original 3-bit number by two bits, we get a,
// and so with another subtraction we have a+b+c, which is the number of bits
// in the original number. How is this insight employed? The first assignment
// statement in the routine computes tmp. Consider the octal representation of
// tmp. Each digit in the octal representation is simply the number of 1¡äs in
// the corresponding three bit positions in n. The last return statement sums
// these octal digits to produce the final answer. The key idea is to add
// adjacent pairs of octal digits together and then compute the remainder
// modulus 63. This is accomplished by right-shifting tmp by three bits, adding
// it to tmp itself and ANDing with a suitable mask. This yields a number in
// which groups of six adjacent bits (starting from the LSB) contain the number
// of 1¡äs among those six positions in n. This number modulo 63 yields the
// final answer. For 64-bit numbers, we would have to add triples of octal
// digits and use modulus 1023. This is HACKMEM 169, as used in X11 sources.
// Source: MIT AI Lab memo, late 1970¡äs.
// http://gurmeet.net/2008/08/05/fast-bit-counting-routines/
int CheckMIT(unsigned int n)
{
/* works for 32-bit numbers only */
/* fix last line for 64-bit numbers */
register unsigned int tmp;
tmp = n - ((n >> 1) & 033333333333)
- ((n >> 2) & 011111111111);
// the remainder of 63 for i equals the sum of 7 octal numbers which
// consititutes the 32-bit integer.
// For example, 03456 % 63 == 034 + 056
return ((tmp + (tmp >> 3)) & 030707070707) % 63;
}
我对上述算法的问题是我对它的理解不够，无法将其转换为“uint64_t”（64 位）版本（尽管执行此操作的指令很少）。
因此，如果你们中的一个人可以帮助我完成这项任务（或至少帮助我理解），那就太棒了。
这是LabSetInt64.h：
/*
* LabSetInt64.h
*
* Created on: Feb 20, 2013
* Author: golgauth
*/
#ifndef LABSETINT64_H_
#define LABSETINT64_H_
#include <ctype.h>
#include <stdio.h>
#include <string.h>
#include <iostream>
#include <stdint.h>
#include <limits>
#include <algorithm>
#include <LabTimeUtils.h>
using namespace std;
namespace elps {
// Lets assume we need at most 1M distinct indices in our sets
#define DEFAULT_SIZE_IN_BITS 1000000
class LabSetInt64
{
public:
LabSetInt64();
LabSetInt64(int sizeinbits);
LabSetInt64(const LabSetInt64 &);
~LabSetInt64();
void Clear();
void Resize(int sizeinbits);
void Set(int i);
void UnSet(int i);
void Set(int i, bool b);
bool Find(int i);
int Cardinality();
int NextSetBit(int i);
void Print();
/**
* Should have been "=" operator, but this overload is not compatible
* with Cython, so we'll use "<<" instead.
* @param
* @return
*/
const LabSetInt64 & operator << ( const LabSetInt64 & );
LabSetInt64 operator ~ () const;
LabSetInt64 operator + ( const LabSetInt64 & );
LabSetInt64 operator * ( const LabSetInt64 & );
LabSetInt64 operator - ( const LabSetInt64 & );
LabSetInt64 operator ^ ( const LabSetInt64 & );
private:
uint64_t *data;
int data_len;
int bits_size;
int popcount(uint64_t x);
int nb_trailing_zeros(uint64_t v);
};
} /* namespace elps */
#endif /* LABSETINT64_H_ */
这里是LabSetInt64.cpp：
/*
* LabSetInt64.cpp
*
* Created on: Feb 20, 2013
* Author: golgauth
*/
#include "LabSetInt64.h"
namespace elps {
/** PUBLICS **/
LabSetInt64::LabSetInt64() : LabSetInt64(DEFAULT_SIZE_IN_BITS) {
}
LabSetInt64::LabSetInt64(int sizeinbits) {
bits_size = sizeinbits;
data_len = (bits_size + 63) / 64;
data = new uint64_t[data_len];
}
LabSetInt64::LabSetInt64(const LabSetInt64& src) {
bits_size = src.bits_size;
data_len = src.data_len;
data = new uint64_t[data_len];
for( int i = 0; i < data_len; i++ ) // copy into this set
data[i] = src.data[i];
}
LabSetInt64::~LabSetInt64() {
}
void LabSetInt64::Clear() {
std::fill_n(data, data_len, 0);
}
void LabSetInt64::Resize(int sizeinbits) {
bits_size = sizeinbits + 1;
int size = (bits_size + 63) / 64;
uint64_t *new_data = new uint64_t[size];
memcpy( new_data, data, data_len * sizeof(uint64_t) );
data_len = size;
delete [] data;
data = new_data;
}
void LabSetInt64::Set(int i) {
data[i / 64] |= (1l << (i % 64));
}
void LabSetInt64::UnSet(int i) {
data[i / 64] &= ~(1l << (i % 64));
}
void LabSetInt64::Set(int i, bool b) {
if(b) Set(i); else UnSet(i);
}
bool LabSetInt64::Find(int i) {
return ((data[i / 64]) & (1l << (i % 64))); // binary AND;
}
int LabSetInt64::Cardinality() {
int sum = 0;
for(int i=0; i<data_len; i++)
sum += popcount(data[i]);
//sum += __builtin_popcount(data[i]);
return sum;
}
int LabSetInt64::NextSetBit(int i) {
int x = i / 64;
long w = data[x];
w >>= (i % 64);
if (w != 0) {
return i + nb_trailing_zeros(w);
}
++x;
for (; x < data_len; ++x) {
if (data[x] != 0) {
return x * 64 + nb_trailing_zeros(data[x]);
}
}
return -1;
}
void LabSetInt64::Print() {
int cur_id = this->NextSetBit(0);
cout << "\nSet size : " << this->Cardinality() << endl;
cout << "{ ";
while (cur_id != -1)
{
cout << (cur_id) << " ";
cur_id = this->NextSetBit(cur_id+1);
}
cout << "}" << endl;;
}
/** PRIVATES **/
//This is better when most bits in x are 0
//It uses 3 arithmetic operations and one comparison/branch per "1" bit in x.
int LabSetInt64::popcount(uint64_t x) {
int count;
for (count=0; x; count++)
x &= x-1;
return count;
}
// output: c will count v's trailing zero bits,
// so if v is 1101000 (base 2), then c will be 3
int LabSetInt64::nb_trailing_zeros(uint64_t v) {
int c;
if (v)
{
v = (v ^ (v - 1)) >> 1; // Set v's trailing 0s to 1s and zero rest
for (c = 0; v; c++) {
v >>= 1;
}
}
else
c = 8 * sizeof(v);
return c;
}
/** ***************************************************************************
********************************* OPERATORS *******************************
*******************************************************************************/
/** PUBLICS **/
/*******************************************************************************
* TODO >> May be better to use "DEFAULT_SIZE_IN_BITS" instead of "data_len" *
* in all the operators !!! *
* *
* => For now, we assume that all the Sets are created with the default *
* constructor (aka : bit_size = DEFAULT_SIZE_IN_BITS) *
*******************************************************************************/
/**
* "operator = " assigns the right hand side to this set.
* returns: nothing
*/
const LabSetInt64 &LabSetInt64::operator << ( const LabSetInt64 &rhs )
{
if( &rhs != this ) // avoid self assignment
{
this->Resize(rhs.bits_size - 1);
for( int i = 0; i < data_len; i++ ) // copy into this set
data[i] = rhs.data[i];
}
return *this; // enable x << y << z;
}
/**
* "operator ~ " performs set complement operation (not).
* returns: pointer to set
*/
LabSetInt64 LabSetInt64::operator ~ () const
{
LabSetInt64 rv;
for( int i = 0; i < data_len; i++ )
rv.data[i] = ~data[i]; // bitwise complement
return rv;
}
/**
* "operator + " performs set union operation (or).
* returns: pointer to set
*/
LabSetInt64 LabSetInt64::operator + ( const LabSetInt64 &rhs )
{
LabSetInt64 rv;
for( int i = 0; i < data_len; i++ )
rv.data[i] = data[i] | rhs.data[i]; // bitwise OR
return rv;
}
/**
* "operator * " performs set intersection operation (and).
* returns: pointer to set
*/
LabSetInt64 LabSetInt64::operator * ( const LabSetInt64 &rhs )
{
LabSetInt64 rv;
for( int i = 0; i < data_len; i++ )
rv.data[i] = data[i] & rhs.data[i]; // bitwise AND
return rv;
}
/**
* "operator - " performs set difference operation.
* returns: pointer to set
*/
LabSetInt64 LabSetInt64::operator - ( const LabSetInt64 &rhs )
{
LabSetInt64 rv;
for( int i = 0; i < data_len; i++ )
rv.data[i] = data[i] & ( ~rhs.data[i] ); // bitwise a AND ~b
return rv;
}
/**
* "operator ^ " performs set symmetric difference operation (xor).
* returns: pointer to set
*/
LabSetInt64 LabSetInt64::operator ^ ( const LabSetInt64 &rhs )
{
LabSetInt64 rv;
for( int i = 0; i < data_len; i++ )
rv.data[i] = data[i] ^ rhs.data[i]; // bitwise XOR
return rv;
}
/** *****************************************************************************
*********************************************************************************/
} /* namespace elps */
对于其余的实施，我对性能非常满意，但再一次，任何好的评论都将非常非常感谢。
非常感谢您的时间。
编辑： 好吧，毕竟我并不完全相信“稀疏”解决方案，因为我需要联合、交集、差异特征，我想这比我的“按位”版本要贵得多（需要迭代通过潜在的大量项目），所以我仍然有兴趣获得一些关于如何将上述“MIT HAKMEM”算法转换为 64 位的帮助。非常感谢。
编辑：这个版本包含一些小瑕疵。请更好地查看下面给出的源代码。