Skip to content

Instantly share code, notes, and snippets.

@jzakiya jzakiya/gcdbenchmarks.cpp
Last active Sep 27, 2018

Embed
What would you like to do?
Learn more about clone URLs
Download ZIP
gcd implementation comparison benchmarks in C++
Raw
gcdbenchmarks.cpp
// Daniel Lemire in this article
// https://lemire.me/blog/2013/12/26/fastest-way-to-compute-the-greatest-common-divisor/
// presented benchmark comparisons of different implementations of Stein's (binary) gcd algorithm.
// The wikipedia (iterative) implementation of the algorithm was noted to be very inefficient.
// Lemire presented benchmarked comparisons of various implementations, original code below.
// https://github.com/lemire/Code-used-on-Daniel-Lemire-s-blog/blob/master/2013/12/26/gcd.cpp
// I have modified the output to make it explicit and clear, modified the code in some functions
// to make them easier to follow and standardized variable names, and added the Ruby implementation.
// I also ordered the ouput to show fastest to slowest.
// The results: gcdwikipedia7fast32 is fastest by far; the implementation shown in widipedia is slowest.
// To compile: g++ -O3 gcdbenchmarks.cpp -o gcdbenchmarks; and runs as ./gcdbenchmarks
#include <stdio.h>
#include <stdlib.h>
#include <sys/stat.h>
#include <sys/time.h>
#include <sys/types.h>
#include <iostream>
#include <cassert>
#include <algorithm>
#include <utility>
//#define ctz_shift_bits 8;
//#define ctz_mask 255; //2**ctz_shift - 1;
using namespace std;
class WallClockTimer
{
public:
struct timeval t1, t2;
public:
WallClockTimer() : t1(), t2() {
gettimeofday(&t1,0);
t2 = t1;
}
void reset() {
gettimeofday(&t1,0);
t2 = t1;
}
int elapsed() {
return (t2.tv_sec * 1000 + t2.tv_usec / 1000) - (t1.tv_sec * 1000 + t1.tv_usec / 1000);
}
int split() {
gettimeofday(&t2,0);
return elapsed();
}
};
// Create ctz (count trailing zeros) function to mimic __builtin_ctz compiler directive
// Create to do 6-bit rights shifts at once. Can increase to hearts content.
//int ctz_shift_bits = 8;
//int ctz_mask = 255; // 2**ctz_shift_bits - 1;
int ctz_bits[256] = {0, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
4, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
5, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
4, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
6, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
4, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
5, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
4, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
7, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
4, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
5, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
4, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
6, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
4, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
5, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
4, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0};
// from http://en.wikipedia.org/wiki/Binary_GCD_algorithm
unsigned int gcdwikipedia2(unsigned int u, unsigned int v)
{
int shift;
if (u == 0) return v;
if (v == 0) return u;
for (shift = 0; ((u | v) & 1) == 0; ++shift) {
u >>= 1;
v >>= 1;
}
while ((u & 1) == 0) u >>= 1;
do {
while ((v & 1) == 0) v >>= 1;
if (u > v) {
unsigned int t = v;
v = u;
u = t;
}
v = v - u;
} while (v != 0);
return u << shift;
}
// based on wikipedia's article,
// fixed by D. Lemire and R. Corderoy
unsigned int gcdwikipedia2fast(unsigned int u, unsigned int v)
{
int shift;
if (u == 0) return v;
if (v == 0) return u;
shift = __builtin_ctz(u | v);
u >>= __builtin_ctz(u);
do {
v >>= __builtin_ctz(v);
if (u > v) {
unsigned int t = v;
v = u;
u = t;
}
v = v - u;
} while (v != 0);
return u << shift;
}
// based on wikipedia's article,
// fixed by D. Lemire and R. Corderoy
unsigned int gcdwikipedia2fastswap(unsigned int u, unsigned int v)
{
int shift;
if (u == 0) return v;
if (v == 0) return u;
shift = __builtin_ctz(u | v);
u >>= __builtin_ctz(u);
do {
v >>= __builtin_ctz(v);
if(u>v) std::swap(u,v);
v = v - u;
} while (v != 0);
return u << shift;
}
// based on wikipedia's article,
// fixed by D. Lemire and R. Corderoy
unsigned int gcdwikipedia2fastxchg(unsigned int u, unsigned int v)
{
int shift;
if (u == 0) return v;
if (v == 0) return u;
shift = __builtin_ctz(u | v);
u >>= __builtin_ctz(u);
do {
v >>= __builtin_ctz(v);
if (u > v) asm volatile("xchg %0, %1\n": "+r"(u), "+r" (v):);
v = v - u;
} while (v != 0);
return u << shift;
}
// based on wikipedia's article,
// fixed by D. Lemire, R. Corderoy, K. Willets
unsigned int gcdwikipedia3fast(unsigned int u, unsigned int v)
{
int shift;
if (u == 0) return v;
if (v == 0) return u;
shift = __builtin_ctz(u | v);
u >>= __builtin_ctz(u);
do {
v >>= __builtin_ctz(v);
if (u == v)
break;
else if (u > v) {
unsigned int t = v;
v = u;
u = t;
}
v = v - u;
} while (true);
return u << shift;
}
// based on wikipedia's article,
// fixed by D. Lemire and R. Corderoy (twice)
unsigned int gcdwikipedia4fast(unsigned int u, unsigned int v)
{
int shift;
if (u == 0) return v;
if (v == 0) return u;
shift = __builtin_ctz(u | v);
u >>= __builtin_ctz(u);
do {
unsigned m;
v >>= __builtin_ctz(v);
m = (v ^ u) & -(v < u);
u ^= m;
v ^= m;
v -= u;
} while (v != 0);
return u << shift;
}
// based on wikipedia's article,
// fixed by D. Lemire and R. Corderoy (twice)
unsigned int gcdwikipedia4fast_a(unsigned int u, unsigned int v)
{
int shift;
unsigned int t;
if (u == 0) return v;
if (v == 0) return u;
t = (u | v); shift = 0;
while ((t & 255) == 0) {t >>= 8; shift += 8;}
shift += ctz_bits[t & 255];
//while ((u & ctz_mask) == 0) u >>= ctz_shift_bits;
//u >>= ctz_bits[u & ctz_mask];
while ((u & 255) == 0) u >>= 8;
u >>= ctz_bits[u & 255];
do {
unsigned m;
//while ((v & ctz_mask) == 0) v >>= ctz_shift_bits;
//v >>= ctz_bits[v & ctz_mask];
while ((v & 255) == 0) v >>= 8;
v >>= ctz_bits[v & 255];
m = (v ^ u) & -(v < u);
u ^= m;
v ^= m;
v -= u;
} while (v != 0);
return u << shift;
}
// based on wikipedia's article,
// fixed by D. Lemire, K. Willets
unsigned int gcdwikipedia5fast(unsigned int u, unsigned int v)
{
int shift, uz, vz;
uz = __builtin_ctz(u);
if ( u == 0) return v;
vz = __builtin_ctz(v);
if ( v == 0) return u;
shift = uz > vz ? vz : uz;
u >>= uz;
do {
v >>= vz;
if (u > v) {
unsigned int t = v;
v = u;
u = t;
}
v = v - u;
vz = __builtin_ctz(v);
} while( v != 0 );
return u << shift;
}
// based on wikipedia's article,
// fixed by D. Lemire, K. Willets
unsigned int gcdwikipedia6fastxchg(unsigned int u, unsigned int v)
{
int shift, uz, vz;
uz = __builtin_ctz(u);
if ( u == 0) return v;
vz = __builtin_ctz(v);
if ( v == 0) return u;
shift = uz > vz ? vz : uz;
u >>= uz;
do {
v >>= vz;
if (u > v) asm volatile("xchg %0, %1\n": "+r"(u), "+r" (v):);
v = v - u;
vz = __builtin_ctz(v);
} while( v != 0 );
return u << shift;
}
// based on wikipedia's article,
// fixed by D. Lemire, K. Willets
unsigned int gcdwikipedia7fast(unsigned int u, unsigned int v)
{
int shift, uz, vz;
if ( u == 0) return v;
if ( v == 0) return u;
uz = __builtin_ctz(u);
vz = __builtin_ctz(v);
shift = uz > vz ? vz : uz;
u >>= uz;
do {
v >>= vz;
long long int diff = v ;
diff -= u;
vz = __builtin_ctz(diff);
if ( diff == 0 ) break;
if ( diff < 0 ) u = v;
v = abs(diff);
} while( 1 );
return u << shift;
}
// based on wikipedia's article,
// fixed by D. Lemire, K. Willets
unsigned int gcdwikipedia7fast32(unsigned int u, unsigned int v)
{
int shift, uz, vz;
if ( u == 0) return v;
if ( v == 0) return u;
uz = __builtin_ctz(u);
vz = __builtin_ctz(v);
shift = uz > vz ? vz : uz;
u >>= uz;
do {
v >>= vz;
int diff = v;
diff -= u;
if ( diff == 0 ) break;
vz = __builtin_ctz(diff);
if ( v < u ) u = v;
v = abs(diff);
} while( 1 );
return u << shift;
}
unsigned int gcdwikipedia7fast32_a(unsigned int u, unsigned int v)
{
int shift, uz, vz, tmp;
if ( u == 0) return v;
if ( v == 0) return u;
uz = 0;
while ((u & 255) == 0) {u >>= 8; uz += 8;}
uz += ctz_bits[u & 255]; u >>= ctz_bits[u & 255];
vz = 0;
while ((v & 255) == 0) {v >>= 8; vz += 8;}
vz += ctz_bits[v & 255]; v >>= ctz_bits[v & 255];
shift = uz > vz ? vz : uz;
do {
int diff = v;
diff -= u;
if ( diff == 0 ) break;
tmp = diff; vz = 0;
while ((tmp & 255) == 0) {tmp >>= 8; vz += 8;}
vz += ctz_bits[tmp & 255];
if ( v < u ) u = v;
v = abs(diff);
v >>= vz;
} while( 1 );
return u << shift;
}
// best from http://hbfs.wordpress.com/2013/12/10/the-speed-of-gcd/
unsigned gcd_recursive(unsigned a, unsigned b)
{
if (b)
return gcd_recursive(b, a % b);
else
return a;
}
// from http://hbfs.wordpress.com/2013/12/10/the-speed-of-gcd/
unsigned gcd_iterative_mod(unsigned a, unsigned b)
{
while (b)
{
unsigned t=b;
b=a % b;
a=t;
}
return a;
}
unsigned basicgcd(unsigned u, unsigned v) {
return (u % v) == 0 ? v : basicgcd(v, u % v);
}
// attributed to Jens
// Franke by Samuel Neves
unsigned gcdFranke (unsigned u, unsigned v)
{
unsigned shift, s0, s1;
if(0 == u) return v;
if(0 == v) return u;
s0 = __builtin_ctz(u);
s1 = __builtin_ctz(v);
shift = s0 < s1 ? s0 : s1;
u >>= s0;
v >>= s1;
while (u != v)
{
if (u < v)
{ v -= u; v >>= __builtin_ctz(v); }
else
{ u -= v; u >>= __builtin_ctz(u); }
}
return u << shift;
}
unsigned rubygcd(long x, long y)
{
unsigned long u, v, t;
int shift;
if (x == 0) return y;
if (y == 0) return x;
if (x < 0) x = -x;
if (y < 0) y = -y;
u = (unsigned long)x;
v = (unsigned long)y;
for (shift = 0; ((u | v) & 1) == 0; ++shift) {
u >>= 1;
v >>= 1;
}
while ((u & 1) == 0) u >>= 1;
do {
while ((v & 1) == 0) v >>= 1;
if (u > v) {
t = v;
v = u;
u = t;
}
v = v - u;
} while (v != 0);
return (long)(u << shift);
}
unsigned rubygcd_a(long x, long y)
{
unsigned long u, v, t;
int shift;
if (x == 0) return y;
if (y == 0) return x;
if (x < 0) x = -x;
if (y < 0) y = -y;
u = (unsigned long)x;
v = (unsigned long)y;
for (shift = 0; ((u | v) & 1) == 0; ++shift) {
u >>= 1;
v >>= 1;
}
//while ((u & ctz_mask) == 0) u >>= ctz_shift_bits;
//u >>= ctz_bits[u & ctz_mask];
while ((u & 255) == 0) u >>= 8;
u >>= ctz_bits[u & 255];
do {
//while ((v & ctz_mask) == 0) v >>= ctz_shift_bits;
//v >>= ctz_bits[v & ctz_mask];
while ((v & 255) == 0) v >>= 8;
v >>= ctz_bits[v & 255];
if (u > v) {
t = v;
v = u;
u = t;
}
v = v - u;
} while (v != 0);
return (long)(u << shift);
}
unsigned rubygcd_b(long x, long y)
{
unsigned long u, v, t;
int shift;
if (x == 0) return y;
if (y == 0) return x;
if (x < 0) x = -x;
if (y < 0) y = -y;
u = (unsigned long)x;
v = (unsigned long)y;
for (shift = 0; ((u | v) & 1) == 0; ++shift) {
u >>= 1;
v >>= 1;
}
//while ((u & ctz_mask) == 0) u >>= ctz_shift_bits;
//u >>= ctz_bits[u & ctz_mask];
//while ((u & 255) == 0) u >>= 8;
//u >>= ctz_bits[u & 255];
u >>= __builtin_ctz(u);
do {
//while ((v & ctz_mask) == 0) v >>= ctz_shift_bits;
//v >>= ctz_bits[v & ctz_mask];
//while ((v & 255) == 0) v >>= 8;
//v >>= ctz_bits[v & 255];
v >>= __builtin_ctz(v);
if (u > v) {
t = v;
v = u;
u = t;
}
v = v - u;
} while (v != 0);
return (long)(u << shift);
}
unsigned int test(unsigned int offset) {
const bool XCHG = true;
WallClockTimer timer;
int N = 2000;
cout << "gcd between numbers in [" << offset+1 << " and "
<< offset+N << "]" << endl;
int ti1 = 0;
int ti2 = 0;
int ti3 = 0;
int ti4 = 0;
int ti5 = 0;
int ti6 = 0;
int ti7 = 0;
int ti8 = 0;
int ti9 = 0;
int ti10 = 0;
int ti11 = 0;
int ti12 = 0;
int ti13 = 0;
int ti14 = 0;
int ti15 = 0;
int ti16 = 0;
int ti17 = 0;
int ti18 = 0;
int ti19 = 0;
int bogus = 0;
timer.reset();
for(unsigned int x = 1+offset; x<=N+offset; ++x)
for(unsigned int y = 1+offset; y<=N+offset; ++y) {
assert(gcdwikipedia2(x,y)==gcdwikipedia2fast(x,y));
assert(gcdwikipedia2(x,y)==gcdwikipedia2fastswap(x,y));
if(XCHG) assert(gcdwikipedia2(x,y)==gcdwikipedia2fastxchg(x,y));
assert(gcdwikipedia2(x,y)==gcdwikipedia3fast(x,y));
assert(gcdwikipedia2(x,y)==gcdwikipedia4fast(x,y));
assert(gcdwikipedia2(x,y)==gcdwikipedia5fast(x,y));
if(XCHG) assert(gcdwikipedia2(x,y)==gcdwikipedia6fastxchg(x,y));
assert(gcdwikipedia2(x,y)==gcdwikipedia7fast(x,y));
assert(gcdwikipedia2(x,y)==gcdwikipedia7fast32(x,y));
assert(gcdwikipedia2(x,y)==gcd_recursive(x,y));
assert(gcdwikipedia2(x,y)==gcd_iterative_mod(x,y));
assert(gcdwikipedia2(x,y)==basicgcd(x,y));
assert(gcdwikipedia2(x,y)==gcdFranke(x,y));
assert(gcdwikipedia2(x,y)==rubygcd(x,y));
assert(gcdwikipedia2(x,y)==gcdwikipedia7fast32_a(x,y));
assert(gcdwikipedia2(x,y)==gcdwikipedia4fast_a(x,y));
assert(gcdwikipedia2(x,y)==rubygcd_a(x,y));
assert(gcdwikipedia2(x,y)==rubygcd_b(x,y));
}
timer.reset();
for(unsigned int x = 1; x<=N; ++x)
for(unsigned int y = 1; y<=N; ++y)
bogus += basicgcd(x,y);
ti1 += timer.split();
timer.reset();
for(unsigned int x = 1; x<=N; ++x)
for(unsigned int y = 1; y<=N; ++y)
bogus += gcdwikipedia2(x,y);
ti2 += timer.split();
timer.reset();
for(unsigned int x = 1; x<=N; ++x)
for(unsigned int y = 1; y<=N; ++y)
bogus += gcdwikipedia2fast(x,y);
ti3 += timer.split();
timer.reset();
for(unsigned int x = 1; x<=N; ++x)
for(unsigned int y = 1; y<=N; ++y)
bogus += gcd_recursive(x,y);
ti4 += timer.split();
timer.reset();
for(unsigned int x = 1; x<=N; ++x)
for(unsigned int y = 1; y<=N; ++y)
bogus += gcd_iterative_mod(x,y);
ti5 += timer.split();
timer.reset();
for(unsigned int x = 1; x<=N; ++x)
for(unsigned int y = 1; y<=N; ++y)
bogus += gcdFranke(x,y);
ti6 += timer.split();
timer.reset();
for(unsigned int x = 1; x<=N; ++x)
for(unsigned int y = 1; y<=N; ++y)
bogus += gcdwikipedia3fast(x,y);
ti7 += timer.split();
timer.reset();
for(unsigned int x = 1; x<=N; ++x)
for(unsigned int y = 1; y<=N; ++y)
bogus += gcdwikipedia4fast(x,y);
ti8 += timer.split();
timer.reset();
for(unsigned int x = 1; x<=N; ++x)
for(unsigned int y = 1; y<=N; ++y)
bogus += gcdwikipedia5fast(x,y);
ti9 += timer.split();
timer.reset();
for(unsigned int x = 1; x<=N; ++x)
for(unsigned int y = 1; y<=N; ++y)
bogus += gcdwikipedia2fastswap(x,y);
ti10 += timer.split();
timer.reset();
if(XCHG) for(unsigned int x = 1; x<=N; ++x)
for(unsigned int y = 1; y<=N; ++y)
bogus += gcdwikipedia6fastxchg(x,y);
ti11 += timer.split();
timer.reset();
if(XCHG) for(unsigned int x = 1; x<=N; ++x)
for(unsigned int y = 1; y<=N; ++y)
bogus += gcdwikipedia2fastxchg(x,y);
ti12 += timer.split();
timer.reset();
for(unsigned int x = 1; x<=N; ++x)
for(unsigned int y = 1; y<=N; ++y)
bogus += gcdwikipedia7fast(x,y);
ti13 += timer.split();
timer.reset();
for(unsigned int x = 1; x<=N; ++x)
for(unsigned int y = 1; y<=N; ++y)
bogus += gcdwikipedia7fast32(x,y);
ti14 += timer.split();
timer.reset();
for(unsigned int x = 1; x<=N; ++x)
for(unsigned int y = 1; y<=N; ++y)
bogus += rubygcd(x,y);
ti15 += timer.split();
timer.reset();
for(unsigned int x = 1; x<=N; ++x)
for(unsigned int y = 1; y<=N; ++y)
bogus += gcdwikipedia7fast32_a(x,y);
ti16 += timer.split();
timer.reset();
for(unsigned int x = 1; x<=N; ++x)
for(unsigned int y = 1; y<=N; ++y)
bogus += gcdwikipedia4fast_a(x,y);
ti17 += timer.split();
timer.reset();
for(unsigned int x = 1; x<=N; ++x)
for(unsigned int y = 1; y<=N; ++y)
bogus += rubygcd_a(x,y);
ti18 += timer.split();
timer.reset();
for(unsigned int x = 1; x<=N; ++x)
for(unsigned int y = 1; y<=N; ++y)
bogus += rubygcd_b(x,y);
ti19 += timer.split();
cout
<< "\n" << "gcdwikipedia7fast32 : " << " time = " << ti14
<< "\n" << "gcdwikipedia4fast : " << " time = " << ti8
<< "\n" << "gcdFranke : " << " time = " << ti6
<< "\n" << "gcdwikipedia3fast : " << " time = " << ti7
<< "\n" << "gcdwikipedia2fastswap : " << " time = " << ti10
<< "\n" << "gcdwikipedia5fast : " << " time = " << ti9
<< "\n" << "gcdwikipedia7fast : " << " time = " << ti13
<< "\n" << "gcdwikipedia2fast : " << " time = " << ti3
<< "\n" << "gcdwikipedia6fastxchg : " << " time = " << ti11
<< "\n" << "gcdwikipedia2fastxchg : " << " time = " << ti12
<< "\n" << "gcd_iterative_mod : " << " time = " << ti5
<< "\n" << "gcd_recursive : " << " time = " << ti4
<< "\n" << "basicgcd : " << " time = " << ti1
<< "\n" << "rubygcd : " << " time = " << ti15
<< "\n" << "gcdwikipedia2 : " << " time = " << ti2
<< "\n" << "gcdwikipedia7fast32_a : " << " time = " << ti16
<< "\n" << "gcdwikipedia4fast_a : " << " time = " << ti17
<< "\n" << "rubygcd_a : " << " time = " << ti18
<< "\n" << "rubygcd_b : " << " time = " << ti19
<< endl;
return bogus;
}
int main() {
assert(sizeof(long)==8);
assert(sizeof(int)==4);
unsigned int bogus = 0;
bogus += test(0);
bogus += test(1000*1000*1000);
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
You can’t perform that action at this time.