head 3.10; access; symbols merge-1:3.8.2.3 autoconf:3.8.0.4 experimental-1:3.8.0.2 mesa-3-1-with-kw3:3.1; locks; strict; comment @ * @; 3.10 date 99.07.12.19.36.05; author miklos; state Exp; branches; next 3.9; 3.9 date 99.07.12.12.05.23; author keithw; state Exp; branches; next 3.8; 3.8 date 99.04.21.11.28.45; author joukj; state Exp; branches 3.8.2.1; next 3.7; 3.7 date 99.04.07.22.19.04; author brianp; state Exp; branches; next 3.6; 3.6 date 99.04.06.01.11.49; author keithw; state Exp; branches; next 3.5; 3.5 date 99.03.31.20.18.38; author keithw; state Exp; branches; next 3.4; 3.4 date 99.03.20.18.56.47; author brianp; state Exp; branches; next 3.3; 3.3 date 99.03.17.12.08.22; author keithw; state Exp; branches; next 3.2; 3.2 date 99.03.16.09.29.20; author joukj; state Exp; branches; next 3.1; 3.1 date 99.02.25.14.12.30; author keithw; state Exp; branches; next ; 3.8.2.1 date 99.05.21.21.29.25; author keithw; state Exp; branches; next 3.8.2.2; 3.8.2.2 date 99.06.19.15.04.13; author keithw; state Exp; branches; next 3.8.2.3; 3.8.2.3 date 99.07.06.09.28.06; author miklos; state Exp; branches; next 3.8.2.4; 3.8.2.4 date 99.07.09.19.24.39; author miklos; state Exp; branches; next ; desc @@ 3.10 log @Typo fixed in powerpc code. @ text @/* $Id: clip_tmp.h,v 3.9 1999/07/12 12:05:23 keithw Exp $ */ /* * Mesa 3-D graphics library * Version: 3.1 * * Copyright (C) 1999 Brian Paul All Rights Reserved. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), * to deal in the Software without restriction, including without limitation * the rights to use, copy, modify, merge, publish, distribute, sublicense, * and/or sell copies of the Software, and to permit persons to whom the * Software is furnished to do so, subject to the following conditions: * * The above copyright notice and this permission notice shall be included * in all copies or substantial portions of the Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ /* * New (3.1) transformation code written by Keith Whitwell. */ /* KW: a clever asm implementation would nestle integer versions * of the outcode calculation underneath the division. Gcc won't * do this, strangely enough, so I only do the divide in * the case where the cliptest passes. This isn't essential, * and an asm implementation needn't replicate that behaviour. */ static GLvector4f * TAG(cliptest_points4)( GLvector4f *clip_vec, GLvector4f *proj_vec, GLubyte clipMask[], GLubyte *orMask, GLubyte *andMask ) { const GLuint stride = clip_vec->stride; const GLfloat *from = (GLfloat *)clip_vec->start; const GLuint count = clip_vec->count; GLuint c = 0; GLfloat (*vProj)[4] = (GLfloat (*)[4])proj_vec->start; GLubyte tmpAndMask = *andMask; GLubyte tmpOrMask = *orMask; GLuint i; STRIDE_LOOP { const GLfloat cx = from[0]; const GLfloat cy = from[1]; const GLfloat cz = from[2]; const GLfloat cw = from[3]; #if defined(macintosh) /* on powerpc cliptest is 17% faster in this way. */ GLuint mask; mask = (((cw < cx) << CLIP_RIGHT_SHIFT)); mask |= (((cw < -cx) << CLIP_LEFT_SHIFT)); mask |= (((cw < cy) << CLIP_TOP_SHIFT)); mask |= (((cw < -cy) << CLIP_BOTTOM_SHIFT)); mask |= (((cw < cz) << CLIP_FAR_SHIFT)); mask |= (((cw < -cz) << CLIP_NEAR_SHIFT)); #else /* !defined(macintosh)) */ GLubyte mask = 0; if (-cx + cw < 0) mask |= CLIP_RIGHT_BIT; if ( cx + cw < 0) mask |= CLIP_LEFT_BIT; if (-cy + cw < 0) mask |= CLIP_TOP_BIT; if ( cy + cw < 0) mask |= CLIP_BOTTOM_BIT; if (-cz + cw < 0) mask |= CLIP_FAR_BIT; if ( cz + cw < 0) mask |= CLIP_NEAR_BIT; #endif /* defined(macintosh) */ clipMask[i] = mask; if (mask) { c++; tmpAndMask &= mask; tmpOrMask |= mask; vProj[i][0] = 0; /* no longer required? */ vProj[i][1] = 0; vProj[i][2] = 0; vProj[i][3] = 1; } else { GLfloat oow = 1.0 / cw; vProj[i][3] = oow; vProj[i][0] = cx * oow; vProj[i][1] = cy * oow; vProj[i][2] = cz * oow; } } *orMask = tmpOrMask; *andMask = (c < count ? 0 : tmpAndMask); proj_vec->flags |= VEC_SIZE_4; proj_vec->size = 3; proj_vec->count = clip_vec->count; return proj_vec; } static GLvector4f * TAG(cliptest_points3)( GLvector4f *clip_vec, GLvector4f *proj_vec, GLubyte clipMask[], GLubyte *orMask, GLubyte *andMask ) { const GLuint stride = clip_vec->stride; const GLuint count = clip_vec->count; const GLfloat *from = (GLfloat *)clip_vec->start; GLubyte tmpOrMask = *orMask; GLubyte tmpAndMask = *andMask; GLuint i; STRIDE_LOOP { const GLfloat cx = from[0], cy = from[1], cz = from[2]; GLubyte mask = 0; if (cx > 1.0) mask |= CLIP_RIGHT_BIT; else if (cx < -1.0) mask |= CLIP_LEFT_BIT; if (cy > 1.0) mask |= CLIP_TOP_BIT; else if (cy < -1.0) mask |= CLIP_BOTTOM_BIT; if (cz > 1.0) mask |= CLIP_FAR_BIT; else if (cz < -1.0) mask |= CLIP_NEAR_BIT; clipMask[i] = mask; tmpOrMask |= mask; tmpAndMask &= mask; } gl_vector4f_clean_elem(proj_vec, count, 3); *orMask = tmpOrMask; *andMask = tmpAndMask; return clip_vec; } static GLvector4f * TAG(cliptest_points2)( GLvector4f *clip_vec, GLvector4f *proj_vec, GLubyte clipMask[], GLubyte *orMask, GLubyte *andMask ) { const GLuint stride = clip_vec->stride; const GLuint count = clip_vec->count; const GLfloat *from = (GLfloat *)clip_vec->start; GLubyte tmpOrMask = *orMask; GLubyte tmpAndMask = *andMask; GLuint i; STRIDE_LOOP { const GLfloat cx = from[0], cy = from[1]; GLubyte mask = 0; if (cx > 1.0) mask |= CLIP_RIGHT_BIT; else if (cx < -1.0) mask |= CLIP_LEFT_BIT; if (cy > 1.0) mask |= CLIP_TOP_BIT; else if (cy < -1.0) mask |= CLIP_BOTTOM_BIT; clipMask[i] = mask; tmpOrMask |= mask; tmpAndMask &= mask; } gl_vector4f_clean_elem(proj_vec, count, 3); *orMask = tmpOrMask; *andMask = tmpAndMask; return clip_vec; } static void TAG(init_c_cliptest)( void ) { gl_clip_tab[4] = TAG(cliptest_points4); gl_clip_tab[3] = TAG(cliptest_points3); gl_clip_tab[2] = TAG(cliptest_points2); } @ 3.9 log @merge from experimental branch upto merge-1 tag @ text @d1 1 a1 1 /* $Id: clip_tmp.h,v 3.8.2.3 1999/07/06 09:28:06 miklos Exp $ */ d58 1 a58 1 /* On PowerPC this is 17% faster */ d64 2 a65 2 mask |= (((cw < cz) << CLIP_NEAR_SHIFT)); mask |= (((cw < -cz) << CLIP_FAR_SHIFT)); @ 3.8 log @ Unitialized proj[n] caused crashes on VMS (and other machines?) @ text @d1 1 a1 1 /* $Id: clip_tmp.h,v 3.7 1999/04/07 22:19:04 brianp Exp $ */ a52 1 GLubyte mask = 0; d57 20 a76 6 if (cx > cw) mask |= CLIP_RIGHT_BIT; if (cx < -cw) mask |= CLIP_LEFT_BIT; if (cy > cw) mask |= CLIP_TOP_BIT; if (cy < -cw) mask |= CLIP_BOTTOM_BIT; if (cz > cw) mask |= CLIP_FAR_BIT; if (cz < -cw) mask |= CLIP_NEAR_BIT; a79 1 clipMask[i] = mask; d81 1 a81 2 vProj[i][3] = 0; vProj[i][0] = 0; d83 2 a84 1 vProj[i][2] = 0; @ 3.8.2.1 log @Quake3 inspired optimizations @ text @d1 1 a1 1 /* $Id: clip_tmp.h,v 3.8 1999/04/21 11:28:45 joukj Exp $ */ d58 6 a63 6 if (-cx + cw < 0) mask |= CLIP_RIGHT_BIT; if ( cx + cw < 0) mask |= CLIP_LEFT_BIT; if (-cy + cw < 0) mask |= CLIP_TOP_BIT; if ( cy + cw < 0) mask |= CLIP_BOTTOM_BIT; if (-cz + cw < 0) mask |= CLIP_FAR_BIT; if ( cz + cw < 0) mask |= CLIP_NEAR_BIT; d69 2 a70 1 vProj[i][0] = 0; /* no longer required? */ d72 1 a72 2 vProj[i][2] = 0; vProj[i][3] = 1; @ 3.8.2.2 log @Removed SGIS multitexture, added FX/X86 assm directory @ text @d1 1 a1 1 /* $Id: clip_tmp.h,v 3.8.2.1 1999/05/21 21:29:25 keithw Exp $ */ a63 2 clipMask[i] = mask; d67 1 @ 3.8.2.3 log @Cliptest_points4 optimalizations for PowerPC. @ text @d1 1 a1 1 /* $Id: clip_tmp.h,v 3.8.2.2 1999/06/19 15:04:13 keithw Exp $ */ d53 1 a57 11 #if defined(macintosh) /* On PowerPC this is 17% faster */ GLuint mask; mask = (((cw < cx) << CLIP_RIGHT_SHIFT)); mask |= (((cw < -cx) << CLIP_LEFT_SHIFT)); mask |= (((cw < cy) << CLIP_TOP_SHIFT)); mask |= (((cw < -cy) << CLIP_BOTTOM_SHIFT)); mask |= (((cw < cz) << CLIP_NEAR_SHIFT)); mask |= (((cw < -cz) << CLIP_FAR_SHIFT)); #else /* !defined(macintosh)) */ GLubyte mask = 0; a63 1 #endif /* defined(macintosh) */ @ 3.8.2.4 log @Typo fixed in PowerPC optimalizations. @ text @d1 1 a1 1 /* $Id: clip_tmp.h,v 3.8.2.3 1999/07/06 09:28:06 miklos Exp $ */ d58 1 a58 1 /* On PowerPC cliptest is 17% faster in this way*/ d64 2 a65 2 mask |= (((cw < cz) << CLIP_FAR_SHIFT)); mask |= (((cw < -cz) << CLIP_NEAR_SHIFT)); @ 3.7 log @inserted copyright info @ text @d1 1 a1 1 /* $Id$ */ d69 4 @ 3.6 log @user-clip bug fixes, faster FX vertex snapping @ text @d1 30 @ 3.5 log @Compiled vertex arrays @ text @d84 1 a84 1 if (proj_vec->flags & VEC_DIRTY_3) gl_vector4f_clean_elem(proj_vec, 3); d116 1 a116 1 if (proj_vec->flags & VEC_DIRTY_3) gl_vector4f_clean_elem(proj_vec, 3); @ 3.4 log @fixed IRIX compiler warnings @ text @a6 2 * * KW: Removed old CLIP_4D flag as part of fix for Jouk's FPE problems. a23 1 const GLfloat cw = from[3]; d27 1 d40 1 a40 1 GLfloat oow = 1.0 / from[3]; d84 1 a84 1 if (proj_vec->flags & VEC_DIRTY_3) gl_clean_elem(3, proj_vec); d116 1 a116 1 if (proj_vec->flags & VEC_DIRTY_3) gl_clean_elem(3, proj_vec); @ 3.3 log @Removed CLIP_4D_BIT, added CLIP_CULLED_BIT. Clipmask is now used to drive culling in vertex transformation, allowing us to skip both clipped and culled vertices with a single test. @ text @d126 1 a126 4 static void TAG(init_c_cliptest)() a131 3 @ 3.2 log @ Patch got from joshv@@planet.net, which solves the overflow problem with gears @ text @d7 2 a8 6 * * For clipped primitives with W all +ve, we can ignore planes not * in the union of the bitmasks of the vertices. With mixed +,- * this doesn't seem to be possible. Thus the test for cw < 0 and * the new flag. This test could alternately be done in * viewclip_polygon_4. d26 1 d30 6 a35 7 const GLfloat cw = from[3]; if (cx > cw) mask |= CLIP_RIGHT_BIT; else if (cx < -cw) mask |= CLIP_LEFT_BIT; if (cy > cw) mask |= CLIP_TOP_BIT; else if (cy < -cw) mask |= CLIP_BOTTOM_BIT; if (cz > cw) mask |= CLIP_FAR_BIT; else if (cz < -cw) mask |= CLIP_NEAR_BIT; a36 1 if (cw <= 0.0) mask |= CLIP_4D; /* can't skip planes in clipping */ a40 12 vProj[i][3] = 0.0; vProj[i][0] = 0.0; vProj[i][1] = 0.0; vProj[i][2] = 0.0; #if 0 } else if (cw == 0) { /* only get here for 0,0,0,0 - not really sure what * the correct behaviour should be - at the moment * I'm inclined to ignore it. */ clipMask[i] = tmpOrMask = CLIP_ALL_BITS|CLIP_4D; #endif d42 1 a42 1 GLfloat oow = 1.0 / cw; d46 2 a47 2 vProj[i][2] = cz * oow; } d49 1 @ 3.1 log @Merged in kw3 patch @ text @d46 4 @