72aec3a06ad57c9b73c9d8828ab6103d2bc26f4a
[mesa.git] / src / mesa / math / m_xform.c
1 /* $Id: m_xform.c,v 1.5 2000/12/26 05:09:31 keithw Exp $ */
2
3 /*
4 * Mesa 3-D graphics library
5 * Version: 3.5
6 *
7 * Copyright (C) 1999-2000 Brian Paul All Rights Reserved.
8 *
9 * Permission is hereby granted, free of charge, to any person obtaining a
10 * copy of this software and associated documentation files (the "Software"),
11 * to deal in the Software without restriction, including without limitation
12 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
13 * and/or sell copies of the Software, and to permit persons to whom the
14 * Software is furnished to do so, subject to the following conditions:
15 *
16 * The above copyright notice and this permission notice shall be included
17 * in all copies or substantial portions of the Software.
18 *
19 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
20 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
21 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
22 * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
23 * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
24 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
25 */
26
27
28 /*
29 * Matrix/vertex/vector transformation stuff
30 *
31 *
32 * NOTES:
33 * 1. 4x4 transformation matrices are stored in memory in column major order.
34 * 2. Points/vertices are to be thought of as column vectors.
35 * 3. Transformation of a point p by a matrix M is: p' = M * p
36 */
37
38 #include <math.h>
39
40 #include "glheader.h"
41 #include "macros.h"
42 #include "mmath.h"
43
44 #include "m_eval.h"
45 #include "m_matrix.h"
46 #include "m_translate.h"
47 #include "m_xform.h"
48
49
50 #ifdef DEBUG
51 #include "m_debug_xform.h"
52 #endif
53
54 #ifdef USE_X86_ASM
55 #include "X86/common_x86_asm.h"
56 #endif
57
58 clip_func gl_clip_tab[5];
59 dotprod_func gl_dotprod_tab[2][5];
60 vec_copy_func gl_copy_tab[2][0x10];
61 normal_func gl_normal_tab[0xf][0x4];
62 transform_func **(gl_transform_tab[2]);
63 static transform_func *cull_transform_tab[5];
64 static transform_func *raw_transform_tab[5];
65
66
67 /* Raw data format used for:
68 * - Object-to-eye transform prior to culling, although this too
69 * could be culled under some circumstances.
70 * - Eye-to-clip transform (via the function above).
71 * - Cliptesting
72 * - And everything else too, if culling happens to be disabled.
73 */
74 #define TAG(x) x##_raw
75 #define TAG2(x,y) x##y##_raw
76 #define IDX 0
77 #define STRIDE_LOOP for (i=0;i<count;i++, STRIDE_F(from, stride))
78 #define LOOP for (i=0;i<n;i++)
79 #define CULL_CHECK
80 #define CLIP_CHECK
81 #define ARGS
82 #include "m_xform_tmp.h"
83 #include "m_clip_tmp.h"
84 #include "m_norm_tmp.h"
85 #include "m_dotprod_tmp.h"
86 #include "m_copy_tmp.h"
87 #undef TAG
88 #undef TAG2
89 #undef LOOP
90 #undef CULL_CHECK
91 #undef CLIP_CHECK
92 #undef ARGS
93 #undef IDX
94
95 /* Culled data used for:
96 * - texture transformations
97 * - viewport map transformation
98 * - normal transformations prior to lighting
99 * - user cliptests
100 */
101 #define TAG(x) x##_masked
102 #define TAG2(x,y) x##y##_masked
103 #define IDX 1
104 #define STRIDE_LOOP for (i=0;i<count;i++, STRIDE_F(from, stride))
105 #define LOOP for (i=0;i<n;i++)
106 #define CULL_CHECK if (mask[i])
107 #define CLIP_CHECK if ((mask[i] & flag) == 0)
108 #define ARGS , const GLubyte mask[]
109 #include "m_xform_tmp.h"
110 #include "m_norm_tmp.h"
111 #include "m_dotprod_tmp.h"
112 #include "m_copy_tmp.h"
113 #undef TAG
114 #undef TAG2
115 #undef LOOP
116 #undef CULL_CHECK
117 #undef CLIP_CHECK
118 #undef ARGS
119 #undef IDX
120
121
122
123
124
125
126 GLvector4f *gl_project_points( GLvector4f *proj_vec,
127 const GLvector4f *clip_vec )
128 {
129 const GLuint stride = clip_vec->stride;
130 const GLfloat *from = (GLfloat *)clip_vec->start;
131 const GLuint count = clip_vec->count;
132 GLfloat (*vProj)[4] = (GLfloat (*)[4])proj_vec->start;
133 GLuint i;
134
135 for (i = 0 ; i < count ; i++, STRIDE_F(from, stride))
136 {
137 GLfloat oow = 1.0F / from[3];
138 vProj[i][3] = oow;
139 vProj[i][0] = from[0] * oow;
140 vProj[i][1] = from[1] * oow;
141 vProj[i][2] = from[2] * oow;
142 }
143
144 proj_vec->flags |= VEC_SIZE_4;
145 proj_vec->size = 3;
146 proj_vec->count = clip_vec->count;
147 return proj_vec;
148 }
149
150
151
152
153
154
155 /*
156 * Transform a 4-element row vector (1x4 matrix) by a 4x4 matrix. This
157 * function is used for transforming clipping plane equations and spotlight
158 * directions.
159 * Mathematically, u = v * m.
160 * Input: v - input vector
161 * m - transformation matrix
162 * Output: u - transformed vector
163 */
164 void gl_transform_vector( GLfloat u[4], const GLfloat v[4], const GLfloat m[16] )
165 {
166 GLfloat v0=v[0], v1=v[1], v2=v[2], v3=v[3];
167 #define M(row,col) m[row + col*4]
168 u[0] = v0 * M(0,0) + v1 * M(1,0) + v2 * M(2,0) + v3 * M(3,0);
169 u[1] = v0 * M(0,1) + v1 * M(1,1) + v2 * M(2,1) + v3 * M(3,1);
170 u[2] = v0 * M(0,2) + v1 * M(1,2) + v2 * M(2,2) + v3 * M(3,2);
171 u[3] = v0 * M(0,3) + v1 * M(1,3) + v2 * M(2,3) + v3 * M(3,3);
172 #undef M
173 }
174
175
176 /* Useful for one-off point transformations, as in clipping.
177 * Note that because the matrix isn't analysed we do too many
178 * multiplies, and that the result is always 4-clean.
179 */
180 void gl_transform_point_sz( GLfloat Q[4], const GLfloat M[16],
181 const GLfloat P[4], GLuint sz )
182 {
183 if (Q == P)
184 return;
185
186 if (sz == 4)
187 {
188 Q[0] = M[0] * P[0] + M[4] * P[1] + M[8] * P[2] + M[12] * P[3];
189 Q[1] = M[1] * P[0] + M[5] * P[1] + M[9] * P[2] + M[13] * P[3];
190 Q[2] = M[2] * P[0] + M[6] * P[1] + M[10] * P[2] + M[14] * P[3];
191 Q[3] = M[3] * P[0] + M[7] * P[1] + M[11] * P[2] + M[15] * P[3];
192 }
193 else if (sz == 3)
194 {
195 Q[0] = M[0] * P[0] + M[4] * P[1] + M[8] * P[2] + M[12];
196 Q[1] = M[1] * P[0] + M[5] * P[1] + M[9] * P[2] + M[13];
197 Q[2] = M[2] * P[0] + M[6] * P[1] + M[10] * P[2] + M[14];
198 Q[3] = M[3] * P[0] + M[7] * P[1] + M[11] * P[2] + M[15];
199 }
200 else if (sz == 2)
201 {
202 Q[0] = M[0] * P[0] + M[4] * P[1] + M[12];
203 Q[1] = M[1] * P[0] + M[5] * P[1] + M[13];
204 Q[2] = M[2] * P[0] + M[6] * P[1] + M[14];
205 Q[3] = M[3] * P[0] + M[7] * P[1] + M[15];
206 }
207 else if (sz == 1)
208 {
209 Q[0] = M[0] * P[0] + M[12];
210 Q[1] = M[1] * P[0] + M[13];
211 Q[2] = M[2] * P[0] + M[14];
212 Q[3] = M[3] * P[0] + M[15];
213 }
214 }
215
216
217 /*
218 * This is called only once. It initializes several tables with pointers
219 * to optimized transformation functions. This is where we can test for
220 * AMD 3Dnow! capability, Intel Katmai, etc. and hook in the right code.
221 */
222 void
223 _math_init_transformation( void )
224 {
225 gl_transform_tab[0] = raw_transform_tab;
226 gl_transform_tab[1] = cull_transform_tab;
227
228 init_c_transformations_raw();
229 init_c_transformations_masked();
230 init_c_norm_transform_raw();
231 init_c_norm_transform_masked();
232 init_c_cliptest_raw();
233 init_copy0_raw();
234 init_copy0_masked();
235 init_dotprod_raw();
236 init_dotprod_masked();
237
238 #ifdef DEBUG
239 gl_test_all_transform_functions( "default" );
240 gl_test_all_normal_transform_functions( "default" );
241 #endif
242
243 #ifdef USE_X86_ASM
244 gl_init_all_x86_transform_asm();
245 #endif
246 }
247
248 void
249 _math_init( void )
250 {
251 _math_init_transformation();
252 _math_init_translate();
253 _math_init_vertices();
254 _math_init_eval();
255 }