summaryrefslogtreecommitdiff
path: root/src/mesa/drivers/dri/i965/brw_vs_state.c
blob: 7285466645b66a1aa7457c44fbccfaaf535f25ae (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
/*
 Copyright (C) Intel Corp.  2006.  All Rights Reserved.
 Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
 develop this 3D driver.
 
 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 (including the
 next paragraph) 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 THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS 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.
 
 **********************************************************************/
 /*
  * Authors:
  *   Keith Whitwell <keith@tungstengraphics.com>
  */
            


#include "brw_context.h"
#include "brw_state.h"
#include "brw_defines.h"
#include "main/macros.h"

struct brw_vs_unit_key {
   unsigned int total_grf;
   unsigned int urb_entry_read_length;
   unsigned int curb_entry_read_length;

   unsigned int curbe_offset;

   unsigned int nr_urb_entries, urb_size;

   unsigned int nr_surfaces;
};

static void
vs_unit_populate_key(struct brw_context *brw, struct brw_vs_unit_key *key)
{
   GLcontext *ctx = &brw->intel.ctx;

   memset(key, 0, sizeof(*key));

   /* CACHE_NEW_VS_PROG */
   key->total_grf = brw->vs.prog_data->total_grf;
   key->urb_entry_read_length = brw->vs.prog_data->urb_read_length;
   key->curb_entry_read_length = brw->vs.prog_data->curb_read_length;

   /* BRW_NEW_URB_FENCE */
   key->nr_urb_entries = brw->urb.nr_vs_entries;
   key->urb_size = brw->urb.vsize;

   /* BRW_NEW_NR_VS_SURFACES */
   key->nr_surfaces = brw->vs.nr_surfaces;

   /* BRW_NEW_CURBE_OFFSETS, _NEW_TRANSFORM */
   if (ctx->Transform.ClipPlanesEnabled) {
      /* Note that we read in the userclip planes as well, hence
       * clip_start:
       */
      key->curbe_offset = brw->curbe.clip_start;
   }
   else {
      key->curbe_offset = brw->curbe.vs_start;
   }
}

static dri_bo *
vs_unit_create_from_key(struct brw_context *brw, struct brw_vs_unit_key *key)
{
   struct brw_vs_unit_state vs;
   dri_bo *bo;
   int chipset_max_threads;

   memset(&vs, 0, sizeof(vs));

   vs.thread0.kernel_start_pointer = brw->vs.prog_bo->offset >> 6; /* reloc */
   vs.thread0.grf_reg_count = ALIGN(key->total_grf, 16) / 16 - 1;
   vs.thread1.floating_point_mode = BRW_FLOATING_POINT_NON_IEEE_754;
   /* Choosing multiple program flow means that we may get 2-vertex threads,
    * which will have the channel mask for dwords 4-7 enabled in the thread,
    * and those dwords will be written to the second URB handle when we
    * brw_urb_WRITE() results.
    */
   vs.thread1.single_program_flow = 0;

   if (BRW_IS_IGDNG(brw))
      vs.thread1.binding_table_entry_count = 0; /* hardware requirement */
   else
      vs.thread1.binding_table_entry_count = key->nr_surfaces;

   vs.thread3.urb_entry_read_length = key->urb_entry_read_length;
   vs.thread3.const_urb_entry_read_length = key->curb_entry_read_length;
   vs.thread3.dispatch_grf_start_reg = 1;
   vs.thread3.urb_entry_read_offset = 0;
   vs.thread3.const_urb_entry_read_offset = key->curbe_offset * 2;

   if (BRW_IS_IGDNG(brw)) {
      switch (key->nr_urb_entries) {
      case 8:
      case 12:
      case 16:
      case 32:
      case 64:
      case 96:
      case 128:
      case 168:
      case 192:
      case 224:
      case 256:
	 vs.thread4.nr_urb_entries = key->nr_urb_entries >> 2;
	 break;
      default:
	 assert(0);
      }
   } else {
      switch (key->nr_urb_entries) {
      case 8:
      case 12:
      case 16:
      case 32:
	 break;
      case 64:
	 assert(BRW_IS_G4X(brw));
	 break;
      default:
	 assert(0);
      }
      vs.thread4.nr_urb_entries = key->nr_urb_entries;
   }

   vs.thread4.urb_entry_allocation_size = key->urb_size - 1;

   if (BRW_IS_IGDNG(brw))
      chipset_max_threads = 72;
   else if (BRW_IS_G4X(brw))
      chipset_max_threads = 32;
   else
      chipset_max_threads = 16;
   vs.thread4.max_threads = CLAMP(key->nr_urb_entries / 2,
				  1, chipset_max_threads) - 1;

   if (INTEL_DEBUG & DEBUG_SINGLE_THREAD)
      vs.thread4.max_threads = 0;

   /* No samplers for ARB_vp programs:
    */
   /* It has to be set to 0 for IGDNG
    */
   vs.vs5.sampler_count = 0;

   if (INTEL_DEBUG & DEBUG_STATS)
      vs.thread4.stats_enable = 1;

   /* Vertex program always enabled:
    */
   vs.vs6.vs_enable = 1;

   bo = brw_upload_cache(&brw->cache, BRW_VS_UNIT,
			 key, sizeof(*key),
			 &brw->vs.prog_bo, 1,
			 &vs, sizeof(vs),
			 NULL, NULL);

   /* Emit VS program relocation */
   dri_bo_emit_reloc(bo,
		     I915_GEM_DOMAIN_INSTRUCTION, 0,
		     vs.thread0.grf_reg_count << 1,
		     offsetof(struct brw_vs_unit_state, thread0),
		     brw->vs.prog_bo);

   return bo;
}

static void prepare_vs_unit(struct brw_context *brw)
{
   struct brw_vs_unit_key key;

   vs_unit_populate_key(brw, &key);

   dri_bo_unreference(brw->vs.state_bo);
   brw->vs.state_bo = brw_search_cache(&brw->cache, BRW_VS_UNIT,
				       &key, sizeof(key),
				       &brw->vs.prog_bo, 1,
				       NULL);
   if (brw->vs.state_bo == NULL) {
      brw->vs.state_bo = vs_unit_create_from_key(brw, &key);
   }
}

const struct brw_tracked_state brw_vs_unit = {
   .dirty = {
      .mesa  = _NEW_TRANSFORM,
      .brw   = (BRW_NEW_CURBE_OFFSETS |
                BRW_NEW_NR_VS_SURFACES |
		BRW_NEW_URB_FENCE),
      .cache = CACHE_NEW_VS_PROG
   },
   .prepare = prepare_vs_unit,
};