m_clip_tmp.h revision 7117f1b4
1/*
2 * Mesa 3-D graphics library
3 * Version:  6.2
4 *
5 * Copyright (C) 1999-2004  Brian Paul   All Rights Reserved.
6 *
7 * Permission is hereby granted, free of charge, to any person obtaining a
8 * copy of this software and associated documentation files (the "Software"),
9 * to deal in the Software without restriction, including without limitation
10 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
11 * and/or sell copies of the Software, and to permit persons to whom the
12 * Software is furnished to do so, subject to the following conditions:
13 *
14 * The above copyright notice and this permission notice shall be included
15 * in all copies or substantial portions of the Software.
16 *
17 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
18 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
19 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
20 * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
21 * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
22 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
23 */
24
25/*
26 * New (3.1) transformation code written by Keith Whitwell.
27 */
28
29
30/* KW: a clever asm implementation would nestle integer versions
31 * of the outcode calculation underneath the division.  Gcc won't
32 * do this, strangely enough, so I only do the divide in
33 * the case where the cliptest passes.  This isn't essential,
34 * and an asm implementation needn't replicate that behaviour.
35 *
36 * \param clip_vec vector of incoming clip-space coords
37 * \param proj_vec vector of resultant NDC-space projected coords
38 * \param clipMask resulting array of clip flags
39 * \param orMask bitwise-OR of clipMask values
40 * \param andMask bitwise-AND of clipMask values
41 * \return proj_vec pointer
42 */
43static GLvector4f * _XFORMAPI TAG(cliptest_points4)( GLvector4f *clip_vec,
44                                                     GLvector4f *proj_vec,
45                                                     GLubyte clipMask[],
46                                                     GLubyte *orMask,
47                                                     GLubyte *andMask )
48{
49   const GLuint stride = clip_vec->stride;
50   const GLfloat *from = (GLfloat *)clip_vec->start;
51   const GLuint count = clip_vec->count;
52   GLuint c = 0;
53   GLfloat (*vProj)[4] = (GLfloat (*)[4])proj_vec->start;
54   GLubyte tmpAndMask = *andMask;
55   GLubyte tmpOrMask = *orMask;
56   GLuint i;
57   STRIDE_LOOP {
58      const GLfloat cx = from[0];
59      const GLfloat cy = from[1];
60      const GLfloat cz = from[2];
61      const GLfloat cw = from[3];
62#if defined(macintosh) || defined(__powerpc__)
63      /* on powerpc cliptest is 17% faster in this way. */
64      GLuint mask;
65      mask = (((cw < cx) << CLIP_RIGHT_SHIFT));
66      mask |= (((cw < -cx) << CLIP_LEFT_SHIFT));
67      mask |= (((cw < cy) << CLIP_TOP_SHIFT));
68      mask |= (((cw < -cy) << CLIP_BOTTOM_SHIFT));
69      mask |= (((cw < cz) << CLIP_FAR_SHIFT));
70      mask |= (((cw < -cz) << CLIP_NEAR_SHIFT));
71#else /* !defined(macintosh)) */
72      GLubyte mask = 0;
73      if (-cx + cw < 0) mask |= CLIP_RIGHT_BIT;
74      if ( cx + cw < 0) mask |= CLIP_LEFT_BIT;
75      if (-cy + cw < 0) mask |= CLIP_TOP_BIT;
76      if ( cy + cw < 0) mask |= CLIP_BOTTOM_BIT;
77      if (-cz + cw < 0) mask |= CLIP_FAR_BIT;
78      if ( cz + cw < 0) mask |= CLIP_NEAR_BIT;
79#endif /* defined(macintosh) */
80
81      clipMask[i] = mask;
82      if (mask) {
83	 c++;
84	 tmpAndMask &= mask;
85	 tmpOrMask |= mask;
86	 vProj[i][0] = 0;
87	 vProj[i][1] = 0;
88	 vProj[i][2] = 0;
89	 vProj[i][3] = 1;
90      } else {
91	 GLfloat oow = 1.0F / cw;
92	 vProj[i][0] = cx * oow;
93	 vProj[i][1] = cy * oow;
94	 vProj[i][2] = cz * oow;
95	 vProj[i][3] = oow;
96      }
97   }
98
99   *orMask = tmpOrMask;
100   *andMask = (GLubyte) (c < count ? 0 : tmpAndMask);
101
102   proj_vec->flags |= VEC_SIZE_4;
103   proj_vec->size = 4;
104   proj_vec->count = clip_vec->count;
105   return proj_vec;
106}
107
108
109
110/*
111 * \param clip_vec vector of incoming clip-space coords
112 * \param proj_vec vector of resultant NDC-space projected coords
113 * \param clipMask resulting array of clip flags
114 * \param orMask bitwise-OR of clipMask values
115 * \param andMask bitwise-AND of clipMask values
116 * \return clip_vec pointer
117 */
118static GLvector4f * _XFORMAPI TAG(cliptest_np_points4)( GLvector4f *clip_vec,
119							GLvector4f *proj_vec,
120							GLubyte clipMask[],
121							GLubyte *orMask,
122							GLubyte *andMask )
123{
124   const GLuint stride = clip_vec->stride;
125   const GLuint count = clip_vec->count;
126   const GLfloat *from = (GLfloat *)clip_vec->start;
127   GLuint c = 0;
128   GLubyte tmpAndMask = *andMask;
129   GLubyte tmpOrMask = *orMask;
130   GLuint i;
131   (void) proj_vec;
132   STRIDE_LOOP {
133      const GLfloat cx = from[0];
134      const GLfloat cy = from[1];
135      const GLfloat cz = from[2];
136      const GLfloat cw = from[3];
137#if defined(macintosh) || defined(__powerpc__)
138      /* on powerpc cliptest is 17% faster in this way. */
139      GLuint mask;
140      mask = (((cw < cx) << CLIP_RIGHT_SHIFT));
141      mask |= (((cw < -cx) << CLIP_LEFT_SHIFT));
142      mask |= (((cw < cy) << CLIP_TOP_SHIFT));
143      mask |= (((cw < -cy) << CLIP_BOTTOM_SHIFT));
144      mask |= (((cw < cz) << CLIP_FAR_SHIFT));
145      mask |= (((cw < -cz) << CLIP_NEAR_SHIFT));
146#else /* !defined(macintosh)) */
147      GLubyte mask = 0;
148      if (-cx + cw < 0) mask |= CLIP_RIGHT_BIT;
149      if ( cx + cw < 0) mask |= CLIP_LEFT_BIT;
150      if (-cy + cw < 0) mask |= CLIP_TOP_BIT;
151      if ( cy + cw < 0) mask |= CLIP_BOTTOM_BIT;
152      if (-cz + cw < 0) mask |= CLIP_FAR_BIT;
153      if ( cz + cw < 0) mask |= CLIP_NEAR_BIT;
154#endif /* defined(macintosh) */
155
156      clipMask[i] = mask;
157      if (mask) {
158	 c++;
159	 tmpAndMask &= mask;
160	 tmpOrMask |= mask;
161      }
162   }
163
164   *orMask = tmpOrMask;
165   *andMask = (GLubyte) (c < count ? 0 : tmpAndMask);
166   return clip_vec;
167}
168
169
170static GLvector4f * _XFORMAPI TAG(cliptest_points3)( GLvector4f *clip_vec,
171                                                     GLvector4f *proj_vec,
172                                                     GLubyte clipMask[],
173                                                     GLubyte *orMask,
174                                                     GLubyte *andMask )
175{
176   const GLuint stride = clip_vec->stride;
177   const GLuint count = clip_vec->count;
178   const GLfloat *from = (GLfloat *)clip_vec->start;
179   GLubyte tmpOrMask = *orMask;
180   GLubyte tmpAndMask = *andMask;
181   GLuint i;
182   (void) proj_vec;
183   STRIDE_LOOP {
184      const GLfloat cx = from[0], cy = from[1], cz = from[2];
185      GLubyte mask = 0;
186      if (cx >  1.0)       mask |= CLIP_RIGHT_BIT;
187      else if (cx < -1.0)  mask |= CLIP_LEFT_BIT;
188      if (cy >  1.0)       mask |= CLIP_TOP_BIT;
189      else if (cy < -1.0)  mask |= CLIP_BOTTOM_BIT;
190      if (cz >  1.0)       mask |= CLIP_FAR_BIT;
191      else if (cz < -1.0)  mask |= CLIP_NEAR_BIT;
192      clipMask[i] = mask;
193      tmpOrMask |= mask;
194      tmpAndMask &= mask;
195   }
196
197   *orMask = tmpOrMask;
198   *andMask = tmpAndMask;
199   return clip_vec;
200}
201
202
203static GLvector4f * _XFORMAPI TAG(cliptest_points2)( GLvector4f *clip_vec,
204                                                     GLvector4f *proj_vec,
205                                                     GLubyte clipMask[],
206                                                     GLubyte *orMask,
207                                                     GLubyte *andMask )
208{
209   const GLuint stride = clip_vec->stride;
210   const GLuint count = clip_vec->count;
211   const GLfloat *from = (GLfloat *)clip_vec->start;
212   GLubyte tmpOrMask = *orMask;
213   GLubyte tmpAndMask = *andMask;
214   GLuint i;
215   (void) proj_vec;
216   STRIDE_LOOP {
217      const GLfloat cx = from[0], cy = from[1];
218      GLubyte mask = 0;
219      if (cx >  1.0)       mask |= CLIP_RIGHT_BIT;
220      else if (cx < -1.0)  mask |= CLIP_LEFT_BIT;
221      if (cy >  1.0)       mask |= CLIP_TOP_BIT;
222      else if (cy < -1.0)  mask |= CLIP_BOTTOM_BIT;
223      clipMask[i] = mask;
224      tmpOrMask |= mask;
225      tmpAndMask &= mask;
226   }
227
228   *orMask = tmpOrMask;
229   *andMask = tmpAndMask;
230   return clip_vec;
231}
232
233
234static void TAG(init_c_cliptest)( void )
235{
236   _mesa_clip_tab[4] = TAG(cliptest_points4);
237   _mesa_clip_tab[3] = TAG(cliptest_points3);
238   _mesa_clip_tab[2] = TAG(cliptest_points2);
239
240   _mesa_clip_np_tab[4] = TAG(cliptest_np_points4);
241   _mesa_clip_np_tab[3] = TAG(cliptest_points3);
242   _mesa_clip_np_tab[2] = TAG(cliptest_points2);
243}
244