shithub: tinygl

Info  •  Files  •  Log  •  Branches

Automatic commit.

--- a/README.md

+++ b/README.md

@@ -48,6 +48,8 @@

 #include <string.h>

 #include <stdarg.h>

 ```

+If your system supports it, the library can also take advantage of `alignas` to get SIMD support,

+which can be disabled in zfeatures. This adds a dependency to `stdalign.h` but greatly increases vertex processing speed.

 If you are unsure if your target platform can support TinyGL, compile it with the buildtime and runtime tests enabled (They are, by default)

--- a/include/zfeatures.h

+++ b/include/zfeatures.h

@@ -70,6 +70,15 @@

 #define TGL_FEATURE_MULTITHREADED_COPY_TEXIMAGE_2D 1

 //Enable multithreading the ZB_CopyBuffer operation.

 #define TGL_FEATURE_MULTITHREADED_ZB_COPYBUFFER 1

+//Enable stdalign

+#define TGL_FEATURE_ALIGNAS 1

+

+#if TGL_FEATURE_ALIGNAS == 1

+#include <stdalign.h>

+#define TGL_ALIGN alignas(16)

+#else

+#define TGL_ALIGN /*a comment*/

+#endif

 //DO NOT TURN THESE ON, I don't maintain them and I actively #error them out.

 #define TGL_FEATURE_8_BITS         0

--- a/src/get.c

+++ b/src/get.c

@@ -145,6 +145,10 @@

 #else

 "TGL_FEATURE_SINGLE_THREADED "

 #endif

+

+#if TGL_FEATURE_ALIGNAS

+"TGL_FEATURE_ALIGNAS "

+#endif

 "TGL_BUFFER_EXT "

 "TGL_FEEDBACK "

 "TGL_SELECT "

--- a/src/zmath.c

+++ b/src/zmath.c

@@ -10,6 +10,7 @@

 void gl_M4_Id(M4* a) {

 	GLint i, j;

+#pragma omp simd collapse(2)

 	for (i = 0; i < 4; i++)

 		for (j = 0; j < 4; j++)

 			if (i == j)

@@ -34,6 +35,7 @@

 void gl_M4_Mul(M4* c, M4* a, M4* b) {

 	GLint i, j, k;

 	GLfloat s;

+#pragma omp simd

 	for (i = 0; i < 4; i++)

 		for (j = 0; j < 4; j++) {

 			s = 0.0;

@@ -52,7 +54,7 @@

 	/*memcpy(&a, c, 16*sizeof(GLfloat));

 	 */

 	a = *c;

-

+#pragma omp simd

 	for (i = 0; i < 4; i++)

 		for (j = 0; j < 4; j++) {

 			s = 0.0;

@@ -79,14 +81,17 @@

 }

 void gl_M4_MulV4(V4* a, M4* b, V4* c) {

-	a->X = b->m[0][0] * c->X + b->m[0][1] * c->Y + b->m[0][2] * c->Z + b->m[0][3] * c->W;

-	a->Y = b->m[1][0] * c->X + b->m[1][1] * c->Y + b->m[1][2] * c->Z + b->m[1][3] * c->W;

-	a->Z = b->m[2][0] * c->X + b->m[2][1] * c->Y + b->m[2][2] * c->Z + b->m[2][3] * c->W;

-	a->W = b->m[3][0] * c->X + b->m[3][1] * c->Y + b->m[3][2] * c->Z + b->m[3][3] * c->W;

+	{

+		a->X = b->m[0][0] * c->X + b->m[0][1] * c->Y + b->m[0][2] * c->Z + b->m[0][3] * c->W;

+		a->Y = b->m[1][0] * c->X + b->m[1][1] * c->Y + b->m[1][2] * c->Z + b->m[1][3] * c->W;

+		a->Z = b->m[2][0] * c->X + b->m[2][1] * c->Y + b->m[2][2] * c->Z + b->m[2][3] * c->W;

+		a->W = b->m[3][0] * c->X + b->m[3][1] * c->Y + b->m[3][2] * c->Z + b->m[3][3] * c->W;

+	}

 }

 /* transposition of a 4x4 matrix */

 void gl_M4_Transpose(M4* a, M4* b) {

+{

 	a->m[0][0] = b->m[0][0];

 	a->m[0][1] = b->m[1][0];

 	a->m[0][2] = b->m[2][0];

@@ -107,11 +112,13 @@

 	a->m[3][2] = b->m[2][3];

 	a->m[3][3] = b->m[3][3];

 }

+}

 /* inversion of an orthogonal matrix of type Y=M.X+P */

 void gl_M4_InvOrtho(M4* a, M4 b) {

 	GLint i, j;

 	GLfloat s;

+#pragma omp simd

 	for (i = 0; i < 3; i++)

 		for (j = 0; j < 3; j++)

 			a->m[i][j] = b.m[j][i];

@@ -119,8 +126,10 @@

 	a->m[3][1] = 0.0;

 	a->m[3][2] = 0.0;

 	a->m[3][3] = 1.0;

+

 	for (i = 0; i < 3; i++) {

 		s = 0;

+#pragma omp simd

 		for (j = 0; j < 3; j++)

 			s -= b.m[j][i] * b.m[j][3];

 		a->m[i][3] = s;

@@ -134,12 +143,12 @@

 	GLint i, j, k, l;

 	GLfloat max, tmp, t;

-	/* identit�e dans r */

+	/*  */

+#pragma omp simd

 	for (i = 0; i < n * n; i++)

 		r[i] = 0;

 	for (i = 0; i < n; i++)

 		r[i * n + i] = 1;

-

 	for (j = 0; j < n; j++) {

 		/* recherche du nombre de plus grand module sur la colonne j */

@@ -157,6 +166,7 @@

 		/* permutation des lignes j et k */

 		if (k != j) {

+#pragma omp simd

 			for (i = 0; i < n; i++) {

 				tmp = m[j * n + i];

 				m[j * n + i] = m[k * n + i];

@@ -170,11 +180,11 @@

 		/* multiplication de la ligne j par 1/max */

 		max = 1 / max;

+#pragma omp simd

 		for (i = 0; i < n; i++) {

 			m[j * n + i] *= max;

 			r[j * n + i] *= max;

 		}

-

 		for (l = 0; l < n; l++)

 			if (l != j) {

 				t = m[l * n + j];

@@ -219,7 +229,6 @@

 	det = m->m[0][0] * m->m[1][1] * m->m[2][2] - m->m[0][0] * m->m[1][2] * m->m[2][1] - m->m[1][0] * m->m[0][1] * m->m[2][2] +

 		  m->m[1][0] * m->m[0][2] * m->m[2][1] + m->m[2][0] * m->m[0][1] * m->m[1][2] - m->m[2][0] * m->m[0][2] * m->m[1][1];

-

 	a->m[0][0] = (m->m[1][1] * m->m[2][2] - m->m[1][2] * m->m[2][1]) / det;

 	a->m[0][1] = -(m->m[0][1] * m->m[2][2] - m->m[0][2] * m->m[2][1]) / det;

 	a->m[0][2] = -(-m->m[0][1] * m->m[1][2] + m->m[0][2] * m->m[1][1]) / det;

@@ -231,6 +240,7 @@

 	a->m[2][0] = (m->m[1][0] * m->m[2][1] - m->m[1][1] * m->m[2][0]) / det;

 	a->m[2][1] = -(m->m[0][0] * m->m[2][1] - m->m[0][1] * m->m[2][0]) / det;

 	a->m[2][2] = (m->m[0][0] * m->m[1][1] - m->m[0][1] * m->m[1][0]) / det;

+

 }

 /* vector arithmetic */

--- a/src/zmath.h

+++ b/src/zmath.h

@@ -1,6 +1,7 @@

 #ifndef __ZMATH__

 #define __ZMATH__

 #include "../include/GL/gl.h"

+#include "../include/zfeatures.h"

 #include <stdlib.h>

 #include <string.h> //For memcpy

 #include <math.h>

@@ -7,15 +8,15 @@

 /* Matrix & Vertex */

 typedef struct {

-	GLfloat m[4][4];

+	TGL_ALIGN GLfloat m[4][4];

 } M4;

 typedef struct {

-	GLfloat m[3][3];

+	TGL_ALIGN GLfloat m[3][3];

 } M3;

 typedef struct {

-	GLfloat m[3][4];

+	TGL_ALIGN GLfloat m[3][4];

 } M34;

 #define X v[0]

@@ -24,11 +25,11 @@

 #define W v[3]

 typedef struct {

-	GLfloat v[3];

+	TGL_ALIGN GLfloat v[3];

 } V3;

 typedef struct {

-	GLfloat v[4];

+	TGL_ALIGN GLfloat v[4];

 } V4;

 void gl_M4_Id(M4* a);

-- 

⑨