src/gallium/auxiliary/draw/draw_vs_varient.c

   1 /**************************************************************************
   2  *
   3  * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
   4  * All Rights Reserved.
   5  *
   6  * Permission is hereby granted, free of charge, to any person obtaining a
   7  * copy of this software and associated documentation files (the
   8  * "Software"), to deal in the Software without restriction, including
   9  * without limitation the rights to use, copy, modify, merge, publish,
  10  * distribute, sub license, and/or sell copies of the Software, and to
  11  * permit persons to whom the Software is furnished to do so, subject to
  12  * the following conditions:
  13  *
  14  * The above copyright notice and this permission notice (including the
  15  * next paragraph) shall be included in all copies or substantial portions
  16  * of the Software.
  17  *
  18  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
  19  * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
  20  * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
  21  * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
  22  * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
  23  * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
  24  * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  25  *
  26  **************************************************************************/
  27
  28  /*
  29   * Authors:
  30   *   Keith Whitwell <keith@tungstengraphics.com>
  31   */
  32
  33 #include "pipe/p_util.h"
  34 #include "draw/draw_context.h"
  35 #include "draw/draw_private.h"
  36 #include "draw/draw_vbuf.h"
  37 #include "draw/draw_vertex.h"
  38 #include "draw/draw_vs.h"
  39 #include "translate/translate.h"
  40 #include "translate/translate_cache.h"
  41
  42 /* A first pass at incorporating vertex fetch/emit functionality into
  43  */
  44 struct draw_vs_varient_generic {
  45    struct draw_vs_varient base;
  46
  47    struct draw_vertex_shader *shader;
  48    struct draw_context *draw;
  49
  50    /* Basic plan is to run these two translate functions before/after
  51     * the vertex shader's existing run_linear() routine to simulate
  52     * the inclusion of this functionality into the shader...
  53     *
  54     * Next will look at actually including it.
  55     */
  56    struct translate *fetch;
  57    struct translate *emit;
  58
  59    unsigned temp_vertex_stride;
  60 };
  61
  62
  63
  64
  65
  66 static void vsvg_set_input( struct draw_vs_varient *varient,
  67                             unsigned buffer,
  68                             const void *ptr,
  69                             unsigned stride )
  70 {
  71    struct draw_vs_varient_generic *vsvg = (struct draw_vs_varient_generic *)varient;
  72
  73    vsvg->fetch->set_buffer(vsvg->fetch,
  74                            buffer,
  75                            ptr,
  76                            stride);
  77 }
  78
  79
  80 /* Mainly for debug at this stage:
  81  */
  82 static void do_rhw_viewport( struct draw_vs_varient_generic *vsvg,
  83                              unsigned count,
  84                              void *output_buffer )
  85 {
  86    char *ptr = (char *)output_buffer;
  87    const float *scale = vsvg->base.vs->draw->viewport.scale;
  88    const float *trans = vsvg->base.vs->draw->viewport.translate;
  89    unsigned stride = vsvg->temp_vertex_stride;
  90    unsigned j;
  91
  92    ptr += vsvg->base.vs->position_output * 4 * sizeof(float);
  93
  94    for (j = 0; j < count; j++, ptr += stride) {
  95       float *data = (float *)ptr;
  96       float w = 1.0f / data[3];
  97
  98       data[0] = data[0] * w * scale[0] + trans[0];
  99       data[1] = data[1] * w * scale[1] + trans[1];
 100       data[2] = data[2] * w * scale[2] + trans[2];
 101       data[3] = w;
 102    }
 103 }
 104
 105 static void do_viewport( struct draw_vs_varient_generic *vsvg,
 106                          unsigned count,
 107                          void *output_buffer )
 108 {
 109    char *ptr = (char *)output_buffer;
 110    const float *scale = vsvg->base.vs->draw->viewport.scale;
 111    const float *trans = vsvg->base.vs->draw->viewport.translate;
 112    unsigned stride = vsvg->temp_vertex_stride;
 113    unsigned j;
 114
 115    ptr += vsvg->base.vs->position_output * 4 * sizeof(float);
 116
 117    for (j = 0; j < count; j++, ptr += stride) {
 118       float *data = (float *)ptr;
 119
 120       data[0] = data[0] * scale[0] + trans[0];
 121       data[1] = data[1] * scale[1] + trans[1];
 122       data[2] = data[2] * scale[2] + trans[2];
 123    }
 124 }
 125
 126
 127 static void PIPE_CDECL vsvg_run_elts( struct draw_vs_varient *varient,
 128                                       const unsigned *elts,
 129                                       unsigned count,
 130                                       void *output_buffer)
 131 {
 132    struct draw_vs_varient_generic *vsvg = (struct draw_vs_varient_generic *)varient;
 133    unsigned temp_vertex_stride = vsvg->temp_vertex_stride;
 134    void *temp_buffer = MALLOC( align(count,4) * temp_vertex_stride );
 135
 136    if (0) debug_printf("%s %d \n", __FUNCTION__,  count);
 137
 138    /* Want to do this in small batches for cache locality?
 139     */
 140
 141    vsvg->fetch->run_elts( vsvg->fetch,
 142                           elts,
 143                           count,
 144                           temp_buffer );
 145
 146    vsvg->base.vs->run_linear( vsvg->base.vs,
 147                               temp_buffer,
 148                               temp_buffer,
 149                               (const float (*)[4])vsvg->base.vs->draw->pt.user.constants,
 150                               count,
 151                               temp_vertex_stride,
 152                               temp_vertex_stride);
 153
 154
 155    if (vsvg->base.key.clip) {
 156       /* not really handling clipping, just do the rhw so we can
 157        * see the results...
 158        */
 159       do_rhw_viewport( vsvg,
 160                        count,
 161                        temp_buffer );
 162    }
 163    else if (vsvg->base.key.viewport) {
 164       do_viewport( vsvg,
 165                    count,
 166                    temp_buffer );
 167    }
 168
 169
 170    vsvg->emit->set_buffer( vsvg->emit,
 171                            0,
 172                            temp_buffer,
 173                            temp_vertex_stride );
 174
 175    vsvg->emit->set_buffer( vsvg->emit,
 176                            1,
 177                            &vsvg->draw->rasterizer->point_size,
 178                            0);
 179
 180    vsvg->emit->run( vsvg->emit,
 181                     0, count,
 182                     output_buffer );
 183
 184    FREE(temp_buffer);
 185 }
 186
 187
 188 static void PIPE_CDECL vsvg_run_linear( struct draw_vs_varient *varient,
 189                                         unsigned start,
 190                                         unsigned count,
 191                                         void *output_buffer )
 192 {
 193    struct draw_vs_varient_generic *vsvg = (struct draw_vs_varient_generic *)varient;
 194    unsigned temp_vertex_stride = vsvg->temp_vertex_stride;
 195    void *temp_buffer = MALLOC( align(count,4) * temp_vertex_stride );
 196
 197    if (0) debug_printf("%s %d %d (sz %d, %d)\n", __FUNCTION__, start, count,
 198                        vsvg->base.key.output_stride,
 199                        temp_vertex_stride);
 200
 201    vsvg->fetch->run( vsvg->fetch,
 202                      start,
 203                      count,
 204                      temp_buffer );
 205
 206    vsvg->base.vs->run_linear( vsvg->base.vs,
 207                               temp_buffer,
 208                               temp_buffer,
 209                               (const float (*)[4])vsvg->base.vs->draw->pt.user.constants,
 210                               count,
 211                               temp_vertex_stride,
 212                               temp_vertex_stride);
 213
 214    if (vsvg->base.key.clip) {
 215       /* not really handling clipping, just do the rhw so we can
 216        * see the results...
 217        */
 218       do_rhw_viewport( vsvg,
 219                        count,
 220                        temp_buffer );
 221    }
 222    else if (vsvg->base.key.viewport) {
 223       do_viewport( vsvg,
 224                    count,
 225                    temp_buffer );
 226    }
 227
 228    vsvg->emit->set_buffer( vsvg->emit,
 229                            0,
 230                            temp_buffer,
 231                            temp_vertex_stride );
 232
 233    vsvg->emit->set_buffer( vsvg->emit,
 234                            1,
 235                            &vsvg->draw->rasterizer->point_size,
 236                            0);
 237
 238    vsvg->emit->run( vsvg->emit,
 239                     0, count,
 240                     output_buffer );
 241
 242    FREE(temp_buffer);
 243 }
 244
 245
 246
 247
 248
 249 static void vsvg_destroy( struct draw_vs_varient *varient )
 250 {
 251    FREE(varient);
 252 }
 253
 254
 255 struct draw_vs_varient *draw_vs_varient_generic( struct draw_vertex_shader *vs,
 256                                                  const struct draw_vs_varient_key *key )
 257 {
 258    unsigned i;
 259    struct translate_key fetch, emit;
 260
 261    struct draw_vs_varient_generic *vsvg = CALLOC_STRUCT( draw_vs_varient_generic );
 262    if (vsvg == NULL)
 263       return NULL;
 264
 265    vsvg->base.key = *key;
 266    vsvg->base.vs = vs;
 267    vsvg->base.set_input     = vsvg_set_input;
 268    vsvg->base.run_elts      = vsvg_run_elts;
 269    vsvg->base.run_linear    = vsvg_run_linear;
 270    vsvg->base.destroy       = vsvg_destroy;
 271
 272    vsvg->draw = vs->draw;
 273
 274    vsvg->temp_vertex_stride = MAX2(key->nr_inputs,
 275                                    vsvg->base.vs->info.num_outputs) * 4 * sizeof(float);
 276
 277    /* Build free-standing fetch and emit functions:
 278     */
 279    fetch.nr_elements = key->nr_inputs;
 280    fetch.output_stride = vsvg->temp_vertex_stride;
 281    for (i = 0; i < key->nr_inputs; i++) {
 282       fetch.element[i].input_format = key->element[i].in.format;
 283       fetch.element[i].input_buffer = key->element[i].in.buffer;
 284       fetch.element[i].input_offset = key->element[i].in.offset;
 285       fetch.element[i].output_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
 286       fetch.element[i].output_offset = i * 4 * sizeof(float);
 287       assert(fetch.element[i].output_offset < fetch.output_stride);
 288    }
 289
 290
 291    emit.nr_elements = key->nr_outputs;
 292    emit.output_stride = key->output_stride;
 293    for (i = 0; i < key->nr_outputs; i++) {
 294       if (key->element[i].out.format != EMIT_1F_PSIZE)
 295       {
 296          emit.element[i].input_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
 297          emit.element[i].input_buffer = 0;
 298          emit.element[i].input_offset = key->element[i].out.vs_output * 4 * sizeof(float);
 299          emit.element[i].output_format = draw_translate_vinfo_format(key->element[i].out.format);
 300          emit.element[i].output_offset = key->element[i].out.offset;
 301          assert(emit.element[i].input_offset <= fetch.output_stride);
 302       }
 303       else {
 304          emit.element[i].input_format = PIPE_FORMAT_R32_FLOAT;
 305          emit.element[i].input_buffer = 1;
 306          emit.element[i].input_offset = 0;
 307          emit.element[i].output_format = PIPE_FORMAT_R32_FLOAT;
 308          emit.element[i].output_offset = key->element[i].out.offset;
 309       }
 310    }
 311
 312    vsvg->fetch = draw_vs_get_fetch( vs->draw, &fetch );
 313    vsvg->emit = draw_vs_get_emit( vs->draw, &emit );
 314
 315    return &vsvg->base;
 316 }
 317
 318
 319
 320
 321