Some more work on interal debugging, timing routines for things that
[mesa.git] / src / mesa / math / m_xform.c
1 /* $Id: m_xform.c,v 1.8 2001/02/03 08:41:04 gareth 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 #include "mathmod.h"
49
50
51 #ifdef DEBUG
52 #include "m_debug.h"
53 #endif
54
55 #ifdef USE_X86_ASM
56 #include "X86/common_x86_asm.h"
57 #endif
58
59 clip_func gl_clip_tab[5];
60 clip_func gl_clip_np_tab[5];
61 dotprod_func gl_dotprod_tab[2][5];
62 vec_copy_func gl_copy_tab[2][0x10];
63 normal_func gl_normal_tab[0xf][0x4];
64 transform_func **(gl_transform_tab[2]);
65 static transform_func *cull_transform_tab[5];
66 static transform_func *raw_transform_tab[5];
67
68
69 /* Raw data format used for:
70 * - Object-to-eye transform prior to culling, although this too
71 * could be culled under some circumstances.
72 * - Eye-to-clip transform (via the function above).
73 * - Cliptesting
74 * - And everything else too, if culling happens to be disabled.
75 */
76 #define TAG(x) x##_raw
77 #define TAG2(x,y) x##y##_raw
78 #define IDX 0
79 #define STRIDE_LOOP for (i=0;i<count;i++, STRIDE_F(from, stride))
80 #define LOOP for (i=0;i<n;i++)
81 #define CULL_CHECK
82 #define CLIP_CHECK
83 #define ARGS
84 #include "m_xform_tmp.h"
85 #include "m_clip_tmp.h"
86 #include "m_norm_tmp.h"
87 #include "m_dotprod_tmp.h"
88 #include "m_copy_tmp.h"
89 #undef TAG
90 #undef TAG2
91 #undef LOOP
92 #undef CULL_CHECK
93 #undef CLIP_CHECK
94 #undef ARGS
95 #undef IDX
96
97 /* Culled data used for:
98 * - texture transformations
99 * - viewport map transformation
100 * - normal transformations prior to lighting
101 * - user cliptests
102 */
103 #define TAG(x) x##_masked
104 #define TAG2(x,y) x##y##_masked
105 #define IDX 1
106 #define STRIDE_LOOP for (i=0;i<count;i++, STRIDE_F(from, stride))
107 #define LOOP for (i=0;i<n;i++)
108 #define CULL_CHECK if (mask[i])
109 #define CLIP_CHECK if ((mask[i] & flag) == 0)
110 #define ARGS , const GLubyte mask[]
111 #include "m_xform_tmp.h"
112 #include "m_norm_tmp.h"
113 #include "m_dotprod_tmp.h"
114 #include "m_copy_tmp.h"
115 #undef TAG
116 #undef TAG2
117 #undef LOOP
118 #undef CULL_CHECK
119 #undef CLIP_CHECK
120 #undef ARGS
121 #undef IDX
122
123
124
125
126
127
128 GLvector4f *gl_project_points( GLvector4f *proj_vec,
129 const GLvector4f *clip_vec )
130 {
131 const GLuint stride = clip_vec->stride;
132 const GLfloat *from = (GLfloat *)clip_vec->start;
133 const GLuint count = clip_vec->count;
134 GLfloat (*vProj)[4] = (GLfloat (*)[4])proj_vec->start;
135 GLuint i;
136
137 for (i = 0 ; i < count ; i++, STRIDE_F(from, stride))
138 {
139 GLfloat oow = 1.0F / from[3];
140 vProj[i][3] = oow;
141 vProj[i][0] = from[0] * oow;
142 vProj[i][1] = from[1] * oow;
143 vProj[i][2] = from[2] * oow;
144 }
145
146 proj_vec->flags |= VEC_SIZE_4;
147 proj_vec->size = 3;
148 proj_vec->count = clip_vec->count;
149 return proj_vec;
150 }
151
152
153
154
155
156
157 /*
158 * Transform a 4-element row vector (1x4 matrix) by a 4x4 matrix. This
159 * function is used for transforming clipping plane equations and spotlight
160 * directions.
161 * Mathematically, u = v * m.
162 * Input: v - input vector
163 * m - transformation matrix
164 * Output: u - transformed vector
165 */
166 void gl_transform_vector( GLfloat u[4], const GLfloat v[4], const GLfloat m[16] )
167 {
168 GLfloat v0=v[0], v1=v[1], v2=v[2], v3=v[3];
169 #define M(row,col) m[row + col*4]
170 u[0] = v0 * M(0,0) + v1 * M(1,0) + v2 * M(2,0) + v3 * M(3,0);
171 u[1] = v0 * M(0,1) + v1 * M(1,1) + v2 * M(2,1) + v3 * M(3,1);
172 u[2] = v0 * M(0,2) + v1 * M(1,2) + v2 * M(2,2) + v3 * M(3,2);
173 u[3] = v0 * M(0,3) + v1 * M(1,3) + v2 * M(2,3) + v3 * M(3,3);
174 #undef M
175 }
176
177
178 /* Useful for one-off point transformations, as in clipping.
179 * Note that because the matrix isn't analysed we do too many
180 * multiplies, and that the result is always 4-clean.
181 */
182 void gl_transform_point_sz( GLfloat Q[4], const GLfloat M[16],
183 const GLfloat P[4], GLuint sz )
184 {
185 if (Q == P)
186 return;
187
188 if (sz == 4)
189 {
190 Q[0] = M[0] * P[0] + M[4] * P[1] + M[8] * P[2] + M[12] * P[3];
191 Q[1] = M[1] * P[0] + M[5] * P[1] + M[9] * P[2] + M[13] * P[3];
192 Q[2] = M[2] * P[0] + M[6] * P[1] + M[10] * P[2] + M[14] * P[3];
193 Q[3] = M[3] * P[0] + M[7] * P[1] + M[11] * P[2] + M[15] * P[3];
194 }
195 else if (sz == 3)
196 {
197 Q[0] = M[0] * P[0] + M[4] * P[1] + M[8] * P[2] + M[12];
198 Q[1] = M[1] * P[0] + M[5] * P[1] + M[9] * P[2] + M[13];
199 Q[2] = M[2] * P[0] + M[6] * P[1] + M[10] * P[2] + M[14];
200 Q[3] = M[3] * P[0] + M[7] * P[1] + M[11] * P[2] + M[15];
201 }
202 else if (sz == 2)
203 {
204 Q[0] = M[0] * P[0] + M[4] * P[1] + M[12];
205 Q[1] = M[1] * P[0] + M[5] * P[1] + M[13];
206 Q[2] = M[2] * P[0] + M[6] * P[1] + M[14];
207 Q[3] = M[3] * P[0] + M[7] * P[1] + M[15];
208 }
209 else if (sz == 1)
210 {
211 Q[0] = M[0] * P[0] + M[12];
212 Q[1] = M[1] * P[0] + M[13];
213 Q[2] = M[2] * P[0] + M[14];
214 Q[3] = M[3] * P[0] + M[15];
215 }
216 }
217
218
219 /*
220 * This is called only once. It initializes several tables with pointers
221 * to optimized transformation functions. This is where we can test for
222 * AMD 3Dnow! capability, Intel Katmai, etc. and hook in the right code.
223 */
224 void
225 _math_init_transformation( void )
226 {
227 gl_transform_tab[0] = raw_transform_tab;
228 gl_transform_tab[1] = cull_transform_tab;
229
230 init_c_transformations_raw();
231 init_c_transformations_masked();
232 init_c_norm_transform_raw();
233 init_c_norm_transform_masked();
234 init_c_cliptest_raw();
235 init_copy0_raw();
236 init_copy0_masked();
237 init_dotprod_raw();
238 init_dotprod_masked();
239
240 #ifdef DEBUG
241 _math_test_all_transform_functions( "default" );
242 _math_test_all_normal_transform_functions( "default" );
243 #endif
244
245 #ifdef USE_X86_ASM
246 gl_init_all_x86_transform_asm();
247 #endif
248 }
249
250 void
251 _math_init( void )
252 {
253 _math_init_transformation();
254 _math_init_translate();
255 _math_init_vertices();
256 _math_init_eval();
257 }