Add fast math functions

This commit is contained in:
iabdalkader 2014-03-03 16:18:48 +02:00
parent ffb3dac484
commit ccc88548f7
2 changed files with 84 additions and 10 deletions

View File

@ -1,6 +1,5 @@
#include <stdlib.h>
#include <string.h>
#include <math.h>
#include <arm_math.h>
#include <stm32f4xx.h>
#include "array.h"
@ -63,7 +62,7 @@ const uint8_t xyz_table[256]= {
93.868573, 94.730654, 95.597335, 96.468625, 97.344529, 98.225055, 99.110210, 100.000000,
};
float vsqrt(float x)
float fast_sqrtf(float x)
{
asm volatile (
"vsqrt.f32 %[r], %[x]\n"
@ -72,8 +71,10 @@ float vsqrt(float x)
return x;
}
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-Wstrict-aliasing"
/* This is slightly faster the VSQRT */
uint32_t fsqrt(float x)
uint32_t fast_sqrt(float x)
{
uint32_t i = *(uint32_t*) &x;
// adjust bias
@ -82,6 +83,61 @@ uint32_t fsqrt(float x)
i >>= 1;
return (uint32_t) *(float*) &i;
}
#pragma GCC diagnostic pop
int fast_floorf(float x)
{
int i;
asm volatile (
"vcvt.S32.f32 %[r], %[x]\n"
: [r] "=t" (i)
: [x] "t" (x));
return i;
}
int fast_roundf(float x)
{
int i;
asm volatile (
"vcvtr.s32.f32 %[r], %[x]\n"
: [r] "=t" (i)
: [x] "t" (x));
return i;
// return (int) floor(x+0.5f);
}
float fast_atanf(float x)
{
return (0.97239f * x) - (0.19195f * x*x*x);
}
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-Wstrict-aliasing"
float fast_expf(float x)
{
double d;
*((int*)(&d) + 0) = 0;
*((int*)(&d) + 1) = (int)(1512775 * x + 1072632447);
return (float) d;
}
#pragma GCC diagnostic pop
float fast_cbrtf(float f)
{
unsigned int* p = (unsigned int *) &f;
*p = *p/3 + 709921077;
return f;
}
float fast_fabsf(float x)
{
asm volatile (
"vabs.f32 %[r], %[x]\n"
: [r] "=t" (x)
: [x] "t" (x));
return x;
}
uint32_t imlib_lab_distance(struct color *c0, struct color *c1)
{
@ -89,7 +145,7 @@ uint32_t imlib_lab_distance(struct color *c0, struct color *c1)
sum += (c0->L - c1->L) * (c0->L - c1->L);
sum += (c0->A - c1->A) * (c0->A - c1->A);
sum += (c0->B - c1->B) * (c0->B - c1->B);
return fsqrt(sum);
return fast_sqrtf(sum);
}
uint32_t imlib_rgb_distance(struct color *c0, struct color *c1)
@ -98,7 +154,7 @@ uint32_t imlib_rgb_distance(struct color *c0, struct color *c1)
sum += (c0->r - c1->r) * (c0->r - c1->r);
sum += (c0->g - c1->g) * (c0->g - c1->g);
sum += (c0->b - c1->b) * (c0->b - c1->b);
return fsqrt(sum);
return fast_sqrtf(sum);
}
uint32_t imlib_hsv_distance(struct color *c0, struct color *c1)
@ -107,7 +163,7 @@ uint32_t imlib_hsv_distance(struct color *c0, struct color *c1)
sum += (c0->h - c1->h) * (c0->h - c1->h);
sum += (c0->s - c1->s) * (c0->s - c1->s);
sum += (c0->v - c1->v) * (c0->v - c1->v);
return fsqrt(sum);
return fast_sqrtf(sum);
}
void imlib_rgb_to_lab(struct color *rgb, struct color *lab)
@ -135,7 +191,7 @@ void imlib_rgb_to_lab(struct color *rgb, struct color *lab)
for (int i=0; i<3; i++) {
t = xyz[i];
if (t > 0.008856f) {
t = cbrtf(t);
t = fast_cbrtf(t);
} else {
t = (7.787f * t) + c1;
}
@ -228,13 +284,24 @@ void imlib_erosion_filter(struct image *src, uint8_t *kernel, int k_size)
xfree(dst);
}
#define SWAP(x)\
({ uint16_t _x = (x); \
(((_x & 0xff)<<8 |(_x & 0xff00) >> 8));})
void imlib_threshold(image_t *image, struct color *color, int threshold)
{
color_t lab1,lab2;
uint16_t *pixels = (uint16_t*) image->pixels;
/* Convert reference RGB to LAB */
imlib_rgb_to_lab(color, &lab1);
uint16_t r = color->r*31/255;
uint16_t g = color->g*63/255;
uint16_t b = color->b*31/255;
r = SWAP((r << 11) | (g << 5) | b);
lab1.L = lab_table[r*3];
lab1.A = lab_table[r*3+1];
lab1.B = lab_table[r*3+2];
for (int y=0; y<image->h; y++) {
for (int x=0; x<image->w; x++) {

View File

@ -152,8 +152,15 @@ typedef struct cascade {
int8_t *rectangles_array;
} cascade_t;
float vsqrt(float x);
uint32_t fsqrt(float x);
/* Basic math functions */
float fast_sqrtf(float x);
uint32_t fast_sqrt(float x);
int fast_floor(float x);
int fast_roundf(float x);
float fast_atanf(float x);
float fast_expf(float x);
float fast_cbrtf(float d);
float fast_fabsf(float d);
/* Point functions */
point_t *point_alloc(int x, int y);