1
0
mirror of https://github.com/mpv-player/mpv synced 2024-12-15 19:35:49 +00:00
mpv/TOOLS/fastmemcpybench.c

257 lines
6.1 KiB
C
Raw Normal View History

/*
* benchmark tool for fast_memcpy code from libvo
*
* NOTE: This code can not be used on Pentium MMX / II because they contain
* a bug in rdtsc. For Intel processors since P6(PII) rdpmc should be used
* instead. For PIII it's disputable and it seems the bug was fixed but this
* was not confirmed through testing.
*/
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/ioctl.h>
#include <unistd.h>
#include <fcntl.h>
#include <sys/mman.h>
#include <sys/time.h>
#include <inttypes.h>
#include "config.h"
#include "cpudetect.h"
#define BLOCK_SIZE 4096
#define CONFUSION_FACTOR 0
#if HAVE_MMX
#define COMPILE_MMX
#endif
#if HAVE_MMX2
#define COMPILE_MMX2
#endif
#if HAVE_AMD3DNOW
#define COMPILE_AMD3DNOW
#endif
#if HAVE_SSE
#define COMPILE_SSE
#endif
#ifdef COMPILE_MMX
#undef RENAME
#undef HAVE_MMX
#undef HAVE_MMX2
#undef HAVE_AMD3DNOW
#undef HAVE_SSE
#undef HAVE_SSE2
#define HAVE_MMX 1
#define HAVE_MMX2 0
#define HAVE_AMD3DNOW 0
#define HAVE_SSE 0
#define HAVE_SSE2 0
#define RENAME(a) a ## _MMX
#include "libvo/aclib_template.c"
#endif
#ifdef COMPILE_MMX2
#undef RENAME
#undef HAVE_MMX
#undef HAVE_MMX2
#undef HAVE_AMD3DNOW
#undef HAVE_SSE
#undef HAVE_SSE2
#define HAVE_MMX 1
#define HAVE_MMX2 1
#define HAVE_AMD3DNOW 0
#define HAVE_SSE 0
#define HAVE_SSE2 0
#define RENAME(a) a ## _MMX2
#include "libvo/aclib_template.c"
#endif
#ifdef COMPILE_AMD3DNOW
#undef RENAME
#undef HAVE_MMX
#undef HAVE_MMX2
#undef HAVE_AMD3DNOW
#undef HAVE_SSE
#undef HAVE_SSE2
#define HAVE_MMX 1
#define HAVE_MMX2 0
#define HAVE_AMD3DNOW 1
#define HAVE_SSE 0
#define HAVE_SSE2 0
#define RENAME(a) a ## _3DNow
#include "libvo/aclib_template.c"
#endif
#ifdef COMPILE_SSE
#undef RENAME
#undef HAVE_MMX
#undef HAVE_MMX2
#undef HAVE_AMD3DNOW
#undef HAVE_SSE
#undef HAVE_SSE2
#define HAVE_MMX 1
#define HAVE_MMX2 1
#define HAVE_AMD3DNOW 0
#define HAVE_SSE 1
#define HAVE_SSE2 1
#define RENAME(a) a ## _SSE
#include "libvo/aclib_template.c"
#endif
//#define ARR_SIZE 100000
#define ARR_SIZE (1024*768*2)
#ifdef CONFIG_MGA
#include "drivers/mga_vid.h"
static mga_vid_config_t mga_vid_config;
static unsigned char* frame = NULL;
static int f;
static int mga_init(void)
{
f = open("/dev/mga_vid", O_RDWR);
if (f == -1) {
fprintf(stderr, "Couldn't open /dev/mga_vid.\n");
return -1;
}
mga_vid_config.num_frames = 1;
mga_vid_config.frame_size = ARR_SIZE;
mga_vid_config.format = MGA_VID_FORMAT_YUY2;
mga_vid_config.colkey_on = 0;
mga_vid_config.src_width = 640;
mga_vid_config.src_height = 480;
mga_vid_config.dest_width = 320;
mga_vid_config.dest_height = 200;
mga_vid_config.x_org = 0;
mga_vid_config.y_org = 0;
mga_vid_config.version = MGA_VID_VERSION;
if (ioctl(f, MGA_VID_CONFIG, &mga_vid_config)) {
perror("Error in mga_vid_config ioctl()");
printf("Your mga_vid driver version is incompatible with this MPlayer version!\n");
exit(1);
}
ioctl(f, MGA_VID_ON, 0);
frame = (char*)mmap(0, mga_vid_config.frame_size*mga_vid_config.num_frames,
PROT_WRITE,MAP_SHARED, f, 0);
if (!frame) {
printf("Can't mmap MGA frame.\n");
exit(1);
}
//clear the buffer
//memset(frames[0], 0x80, mga_vid_config.frame_size*mga_vid_config.num_frames);
return 0;
}
#endif
// Returns current time in microseconds
static unsigned int GetTimer(void)
{
struct timeval tv;
struct timezone tz;
//float s;
gettimeofday(&tv, &tz);
//s = tv.tv_usec; s *= 0.000001; s += tv.tv_sec;
return tv.tv_sec * 1000000 + tv.tv_usec;
}
static inline unsigned long long int read_tsc(void)
{
unsigned long long int retval;
__asm__ volatile ("rdtsc":"=A" (retval)::"memory");
return retval;
}
unsigned char __attribute__((aligned(4096)))arr1[ARR_SIZE], arr2[ARR_SIZE];
int main(void)
{
unsigned long long int v1, v2;
unsigned char *marr1, *marr2;
int i;
unsigned int t;
#ifdef CONFIG_MGA
mga_init();
marr1 = &frame[3];
#else
marr1 = &arr1[0];
#endif
marr2 = &arr2[0];
for (i = 0; i < ARR_SIZE - 16; i++)
marr1[i] = marr2[i] = i;
t = GetTimer();
v1 = read_tsc();
for (i = 0; i < 100; i++)
memcpy(marr1, marr2, ARR_SIZE - 16);
v2 = read_tsc();
t = GetTimer() - t;
// ARR_SIZE*100 / (1024*1024) / (t/1000000) = ARR_SIZE*95.36743 / t
printf("libc: CPU clocks=%llu = %dus (%5.3ffps) %5.1fMB/s\n", v2-v1, t,
100000000.0f/(float)t, (float)ARR_SIZE*95.36743f/(float)t);
#if HAVE_MMX
t = GetTimer();
v1 = read_tsc();
for (i = 0; i < 100; i++)
fast_memcpy_MMX(marr1, marr2, ARR_SIZE - 16);
v2 = read_tsc();
t = GetTimer() - t;
// ARR_SIZE*100 / (1024*1024) / (t/1000000) = ARR_SIZE*95.36743 / t
printf("MMX: CPU clocks=%llu = %dus (%5.3ffps) %5.1fMB/s\n", v2-v1, t,
100000000.0f/(float)t, (float)ARR_SIZE*95.36743f/(float)t);
#endif
#if HAVE_AMD3DNOW
t = GetTimer();
v1 = read_tsc();
for (i = 0; i < 100; i++)
fast_memcpy_3DNow(marr1, marr2, ARR_SIZE - 16);
v2 = read_tsc();
t = GetTimer() - t;
// ARR_SIZE*100 / (1024*1024) / (t/1000000) = ARR_SIZE*95.36743 / t
printf("3DNow!: CPU clocks=%llu = %dus (%5.3ffps) %5.1fMB/s\n", v2-v1, t,
100000000.0f/(float)t, (float)ARR_SIZE*95.36743f/(float)t);
#endif
#if HAVE_MMX2
t = GetTimer();
v1 = read_tsc();
for (i = 0; i < 100; i++)
fast_memcpy_MMX2(marr1, marr2, ARR_SIZE - 16);
v2 = read_tsc();
t = GetTimer() - t;
// ARR_SIZE*100 / (1024*1024) / (t/1000000) = ARR_SIZE*95.36743 / t
printf("MMX2: CPU clocks=%llu = %dus (%5.3ffps) %5.1fMB/s\n", v2-v1, t,
100000000.0f/(float)t, (float)ARR_SIZE*95.36743f/(float)t);
#endif
#if HAVE_SSE
t = GetTimer();
v1 = read_tsc();
for (i = 0; i < 100; i++)
fast_memcpy_SSE(marr1, marr2, ARR_SIZE - 16);
v2 = read_tsc();
t = GetTimer() - t;
// ARR_SIZE*100 / (1024*1024) / (t/1000000) = ARR_SIZE*95.36743 / t
printf("SSE: CPU clocks=%llu = %dus (%5.3ffps) %5.1fMB/s\n", v2-v1, t,
100000000.0f/(float)t, (float)ARR_SIZE*95.36743f/(float)t);
#endif
return 0;
}