Mercurial > hg > Game > Cerium

comparison TaskManager/Test/simple_render/spe/CreatePolygonPack.cpp @ 106:c9efdb17e8d2

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
*** empty log message ***
author gongo
date Tue, 04 Mar 2008 15:33:23 +0900
parents 3e331f7576a1
children 028ffc9c0375
comparison
equal deleted inserted replaced
105:3e331f7576a1 106:c9efdb17e8d2
1 #include <stdio.h> 1 #include <stdio.h>
2 #include <stdlib.h> 2 #include <spu_intrinsics.h>
3 #include <malloc.h>
4 #include "CreatePolygonPack.h" 3 #include "CreatePolygonPack.h"
5 #include "polygon_pack.h" 4 #include "polygon_pack.h"
6 #include "scene_graph_pack.h" 5 #include "scene_graph_pack.h"
6 #include "sys.h"
7 7
8 8
9 void 9 void
10 CreatePolygonPack::read(void) 10 CreatePolygonPack::read(void)
11 { 11 {
12 SchedTask::read(); 12 printf("CreatePolgonPack\n");
13
14 SchedTask::read();
15 }
16
17 //create_pp(SceneGraphPack *sgp, PolygonPack *pp)
18 //create_pp(void *read, void *write)
19 //CreatePolygonPack::run(SceneGraphPack *sgp, PolygonPack *pp)
20 inline float
21 CreatePolygonPack::sum_across_float4(vector float v)
22 {
23 vector float c12, c2, c3, c4, c34;
24 vector float result;
25
26 c2 = spu_rlqwbyte(v, 4);
27 c3 = spu_rlqwbyte(v, 8);
28 c4 = spu_rlqwbyte(v, 12);
29 c12 = spu_add(v, c2);
30 c34 = spu_add(c3, c4);
31
32 result = spu_add(c12, c34);
33 return (spu_extract(result, 0));
13 } 34 }
14 35
15 int 36 int
16 CreatePolygonPack::run(void *rbuf, void *wbuf) 37 CreatePolygonPack::run(void *rbuf, void *wbuf)
17 { 38 {
18 SceneGraphPack *sgp = (SceneGraphPack*)rbuf; 39 SceneGraphPack *sgp = (SceneGraphPack*)rbuf;
19 PolygonPack *pp = (PolygonPack*)wbuf; 40 PolygonPack *pp = (PolygonPack*)wbuf;
20 41
21 float xyz1[4],xyz2[4],xyz3[4]; 42 float xyz1[4],xyz2[4],xyz3[4];
22 43
23 for (int i = 0; i < sgp->info.size; i++) { 44 for (int i = 0; i < sgp->info.size; i++) {
24 SceneGraphNodePtr node = &sgp->node[i]; 45 SceneGraphNodePtr node = &sgp->node[i];
25 46
26 int n,nt,pt; 47 int n,nt,pt;
27 for(n=0,nt=0,pt=0; n<node->size*3; n+=9,nt+=6,pt++) { 48 for(n=0,nt=0,pt=0; n<node->size*3; n+=9,nt+=6,pt++) {
28 xyz1[0] = node->vertex[n]; 49 xyz1[0] = node->vertex[n];
29 xyz1[1] = node->vertex[n+1]; 50 xyz1[1] = node->vertex[n+1];
30 xyz1[2] = node->vertex[n+2]*-1; 51 xyz1[2] = node->vertex[n+2]*-1;
31 xyz1[3] = 1; 52 xyz1[3] = 1;
32 xyz2[0] = node->vertex[n+3]; 53 xyz2[0] = node->vertex[n+3];
33 xyz2[1] = node->vertex[n+3+1]; 54 xyz2[1] = node->vertex[n+3+1];
34 xyz2[2] = node->vertex[n+3+2]*-1; 55 xyz2[2] = node->vertex[n+3+2]*-1;
35 xyz2[3] = 1; 56 xyz2[3] = 1;
36 xyz3[0] = node->vertex[n+6]; 57 xyz3[0] = node->vertex[n+6];
37 xyz3[1] = node->vertex[n+6+1]; 58 xyz3[1] = node->vertex[n+6+1];
38 xyz3[2] = node->vertex[n+6+2]*-1; 59 xyz3[2] = node->vertex[n+6+2]*-1;
39 xyz3[3] = 1; 60 xyz3[3] = 1;
40 61
41 rotate(xyz1, node->translation); 62 rotate(xyz1, node->translation);
42 rotate(xyz2, node->translation); 63 rotate(xyz2, node->translation);
43 rotate(xyz3, node->translation); 64 rotate(xyz3, node->translation);
44 65
45 pp->tri[pt].ver1.x = xyz1[0]; 66 pp->tri[pt].ver1.x = xyz1[0];
46 pp->tri[pt].ver1.y = xyz1[1]; 67 pp->tri[pt].ver1.y = xyz1[1];
47 pp->tri[pt].ver1.z = xyz1[2]; 68 pp->tri[pt].ver1.z = xyz1[2];
48 pp->tri[pt].ver1.tex_x = node->texture[nt]; 69 pp->tri[pt].ver1.tex_x = node->texture[nt];
49 pp->tri[pt].ver1.tex_y = node->texture[nt+1]; 70 pp->tri[pt].ver1.tex_y = node->texture[nt+1];
50 71
51 pp->tri[pt].ver2.x = xyz2[0]; 72 pp->tri[pt].ver2.x = xyz2[0];
52 pp->tri[pt].ver2.y = xyz2[1]; 73 pp->tri[pt].ver2.y = xyz2[1];
53 pp->tri[pt].ver2.z = xyz2[2]; 74 pp->tri[pt].ver2.z = xyz2[2];
54 pp->tri[pt].ver2.tex_x = node->texture[nt+2]; 75 pp->tri[pt].ver2.tex_x = node->texture[nt+2];
55 pp->tri[pt].ver2.tex_y = node->texture[nt+2+1]; 76 pp->tri[pt].ver2.tex_y = node->texture[nt+2+1];
56 77
57 pp->tri[pt].ver3.x = xyz3[0]; 78 pp->tri[pt].ver3.x = xyz3[0];
58 pp->tri[pt].ver3.y = xyz3[1]; 79 pp->tri[pt].ver3.y = xyz3[1];
59 pp->tri[pt].ver3.z = xyz3[2]; 80 pp->tri[pt].ver3.z = xyz3[2];
60 pp->tri[pt].ver3.tex_x = node->texture[nt+4]; 81 pp->tri[pt].ver3.tex_x = node->texture[nt+4];
61 pp->tri[pt].ver3.tex_y = node->texture[nt+4+1]; 82 pp->tri[pt].ver3.tex_y = node->texture[nt+4+1];
62 83
63 pp->tri[pt].tex_width = node->tex_width; 84 pp->tri[pt].tex_width = node->tex_width;
64 pp->tri[pt].tex_height = node->tex_height; 85 pp->tri[pt].tex_height = node->tex_height;
65 } 86 }
66 pp->info.size = pt; 87 pp->info.size = pt;
67 pp->ssl = sgp->ssl; 88 pp->ssl = sgp->ssl;
68 } 89 }
69 90
70 return sizeof(PolygonPack); 91 return sizeof(PolygonPack);
71 } 92 }
72 93
73 void
74 CreatePolygonPack::write(void)
75 {
76 SchedTask::write();
77
78 free(readbuf);
79 free(writebuf);
80 }
81 94
82 void 95 void
83 CreatePolygonPack::rotate(float *xyz, float *matrix) 96 CreatePolygonPack::rotate(float *xyz, float *matrix)
84 { 97 {
85 float abc[4]; 98 #if 1
99 float abc[4];
86 100
87 abc[0] = xyz[0]; 101 abc[0] = xyz[0];
88 abc[1] = xyz[1]; 102 abc[1] = xyz[1];
89 abc[2] = xyz[2]; 103 abc[2] = xyz[2];
90 abc[3] = xyz[3]; 104 abc[3] = xyz[3];
91 105
92 // SIMD 使えるよね 106 // SIMD 使えるよね
93 for (int i=0; i<4; i++) 107 for (int i=0; i<4; i++)
94 { 108 {
95 xyz[i] = abc[0]*matrix[i] + abc[1]*matrix[i+4] 109 xyz[i] = abc[0]*matrix[i] + abc[1]*matrix[i+4] + abc[2]*matrix[i+8] + abc[3]*matrix[i+12];
96 + abc[2]*matrix[i+8] + abc[3]*matrix[i+12]; 110 }
97 } 111 #else
112 vector float *abc = (vector float *)xyz;
113 float tmp[4];
114
115 vector float matrixT0 = (vector float){matrix[0], matrix[4], matrix[8], matrix[12]};
116 vector float matrixT1 = (vector float){matrix[1], matrix[5], matrix[9], matrix[13]};
117 vector float matrixT2 = (vector float){matrix[2], matrix[6], matrix[10], matrix[14]};
118 vector float matrixT3 = (vector float){matrix[3], matrix[7], matrix[11], matrix[15]};
119
120 #if 1
121 vector float *v_tmp = (vector float *)tmp;
122
123 *v_tmp = spu_mul(*abc, matrixT0);
124 xyz[0] = tmp[0] + tmp[1] + tmp[2] + tmp[3];
125 *v_tmp = spu_mul(*abc, matrixT1);
126 xyz[1] = tmp[0] + tmp[1] + tmp[2] + tmp[3];
127 *v_tmp = spu_mul(*abc, matrixT2);
128 xyz[2] = tmp[0] + tmp[1] + tmp[2] + tmp[3];
129 *v_tmp = spu_mul(*abc, matrixT3);
130 xyz[3] = tmp[0] + tmp[1] + tmp[2] + tmp[3];
131 #else
132 vector float v_tmp;
133
134 v_tmp = spu_mul(*abc, matrixT0);
135 xyz[0] = sum_across_float4(v_tmp);
136 v_tmp = spu_mul(*abc, matrixT1);
137 xyz[1] = sum_across_float4(v_tmp);
138 v_tmp = spu_mul(*abc, matrixT2);
139 xyz[2] = sum_across_float4(v_tmp);
140 v_tmp = spu_mul(*abc, matrixT3);
141 xyz[3] = sum_across_float4(v_tmp);
142 #endif
143 #endif
144
98 } 145 }
99 146
100 SchedTask* 147 SchedTask*
101 createTask_createPolygonPack(TaskListPtr _taskList, TaskPtr _task, 148 createTask_createPolygonPack(TaskListPtr _taskList, TaskPtr _task,
102 void *rbuff, void *wbuff, DmaManager *dma) 149 void *rbuff, void *wbuff, DmaManager *dma)
103 { 150 {
104 rbuff = memalign(16, sizeof(SceneGraphPack)); 151 return new CreatePolygonPack(_taskList, _task, rbuff, wbuff, dma);
105 wbuff = memalign(16, sizeof(PolygonPack));
106
107 return new CreatePolygonPack(_taskList, _task, rbuff, wbuff, dma);
108 } 152 }