phc-discussions - Re: [PHC] wider integer multiply on 32-bit x86

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date: Tue, 04 Mar 2014 22:51:00 +0000
From: Samuel Neves <sneves@....uc.pt>
To: discussions@...sword-hashing.net
Subject: Re: [PHC] wider integer multiply on 32-bit x86

On 04-03-2014 18:54, Solar Designer wrote:
>
> Thanks.  If you send me a complete test program that I can compile and
> run on Linux, I'll test on P2, P3, P4.  (No renaming on P1.)
>
> Alexander

This should be good enough:

#include <stdio.h>
#include <stdint.h>

#if   defined(__i386__)
static unsigned long long cpucycles( void )
{
  unsigned long long result;
  __asm__ __volatile__
  (
    ".byte 15;.byte 49"
    : "=A" ( result )
  );
  return result;
}
#elif defined(__x86_64__)
static unsigned long long cpucycles( void )
{
  unsigned long long result;
  __asm__ __volatile__
  (
    ".byte 15;.byte 49\n"
    "shlq $32,%%rdx\n"
    "orq %%rdx,%%rax"
    : "=a" ( result ) ::  "%rdx"
  );
  return result;
}
#endif

void code1(uint32_t iters)
{
  __asm__ __volatile__
  (
    "mov %0, %%ecx\n\t"
    "sub $1, %%ecx\n\t"
    "1:\n\t"
    "mul %%ebx\n\t"
    "mov %%eax, %%esi\n\t"
    "mov %%ebx, %%eax\n\t"
    "mul %%ebx\n\t"
    "mov %%eax, %%edi\n\t"
    "mov %%ebx, %%eax\n\t"
    "sub $1, %%ecx\n\t"
    "jnz 1b\n\t"
    :
    : "r"(iters)
    : "eax", "ebx", "ecx", "esi", "edi"
  );
}


void code2(uint32_t iters)
{
  __asm__ __volatile__
  (
    "mov %0, %%ecx\n\t"
    "sub $1, %%ecx\n\t"
    "1:\n\t"
    "mul %%ebx\n\t"
    "mov %%eax, %%esi\n\t"
    "mul %%ebx\n\t"
    "mov %%eax, %%edi\n\t"
    "sub $1, %%ecx\n\t"
    "jnz 1b\n\t"
    :
    : "r"(iters)
    : "eax", "ebx", "ecx", "esi", "edi"
  );
}


int main(int argc, char **argv)
{
  double n = 1UL << 28;
  if(argc == 2) n = 1ULL << strtoul(argv[1], 0, 10);
  uint64_t t;

  t = cpucycles();
  code1(n);
  t = cpucycles() - t;
  printf("%f\n", t / n);
 
  t = cpucycles();
  code2(n);
  t = cpucycles() - t;
  printf("%f\n", t / n);
 
  return 0;
}