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
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
|
/**************************************************************************
*
* Copyright 2007-2009 VMware, Inc.
* 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, sub license, 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 NON-INFRINGEMENT.
* IN NO EVENT SHALL VMWARE 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.
*
**************************************************************************/
/*
* Rasterization for binned triangles within a tile
*/
#include <limits.h>
#include "util/u_math.h"
#include "lp_debug.h"
#include "lp_perf.h"
#include "lp_rast_priv.h"
#include "lp_tile_soa.h"
/**
* Map an index in [0,15] to an x,y position, multiplied by 4.
* This is used to get the position of each subtile in a 4x4
* grid of edge step values.
* Note: we can use some bit twiddling to compute these values instead
* of using a look-up table, but there's no measurable performance
* difference.
*/
static const int pos_table4[16][2] = {
{ 0, 0 },
{ 4, 0 },
{ 0, 4 },
{ 4, 4 },
{ 8, 0 },
{ 12, 0 },
{ 8, 4 },
{ 12, 4 },
{ 0, 8 },
{ 4, 8 },
{ 0, 12 },
{ 4, 12 },
{ 8, 8 },
{ 12, 8 },
{ 8, 12 },
{ 12, 12 }
};
static const int pos_table16[16][2] = {
{ 0, 0 },
{ 16, 0 },
{ 0, 16 },
{ 16, 16 },
{ 32, 0 },
{ 48, 0 },
{ 32, 16 },
{ 48, 16 },
{ 0, 32 },
{ 16, 32 },
{ 0, 48 },
{ 16, 48 },
{ 32, 32 },
{ 48, 32 },
{ 32, 48 },
{ 48, 48 }
};
/**
* Shade all pixels in a 4x4 block.
*/
static void
block_full_4( struct lp_rasterizer_task *rast_task,
const struct lp_rast_triangle *tri,
int x, int y )
{
lp_rast_shade_quads_all(rast_task->rast,
rast_task->thread_index,
&tri->inputs,
x, y);
}
/**
* Shade all pixels in a 16x16 block.
*/
static void
block_full_16( struct lp_rasterizer_task *rast_task,
const struct lp_rast_triangle *tri,
int x, int y )
{
unsigned ix, iy;
assert(x % 16 == 0);
assert(y % 16 == 0);
for (iy = 0; iy < 16; iy += 4)
for (ix = 0; ix < 16; ix += 4)
block_full_4(rast_task, tri, x + ix, y + iy);
}
/**
* Pass the 4x4 pixel block to the shader function.
* Determination of which of the 16 pixels lies inside the triangle
* will be done as part of the fragment shader.
*/
static void
do_block_4( struct lp_rasterizer_task *rast_task,
const struct lp_rast_triangle *tri,
int x, int y,
int c1,
int c2,
int c3 )
{
lp_rast_shade_quads(rast_task->rast,
rast_task->thread_index,
&tri->inputs,
x, y,
-c1, -c2, -c3);
}
/**
* Evaluate a 16x16 block of pixels to determine which 4x4 subblocks are in/out
* of the triangle's bounds.
*/
static void
do_block_16( struct lp_rasterizer_task *rast_task,
const struct lp_rast_triangle *tri,
int x, int y,
int c0,
int c1,
int c2 )
{
unsigned mask = 0;
int eo[3];
int c[3];
int i, j;
assert(x % 16 == 0);
assert(y % 16 == 0);
eo[0] = tri->eo1 * 4;
eo[1] = tri->eo2 * 4;
eo[2] = tri->eo3 * 4;
c[0] = c0;
c[1] = c1;
c[2] = c2;
for (j = 0; j < 3; j++) {
const int *step = tri->inputs.step[j];
const int cx = c[j] + eo[j];
/* Mask has bits set whenever we are outside any of the edges.
*/
for (i = 0; i < 16; i++) {
int out = cx + step[i] * 4;
mask |= (out >> 31) & (1 << i);
}
}
mask = ~mask & 0xffff;
while (mask) {
int i = ffs(mask) - 1;
int px = x + pos_table4[i][0];
int py = y + pos_table4[i][1];
int cx1 = c0 + tri->inputs.step[0][i] * 4;
int cx2 = c1 + tri->inputs.step[1][i] * 4;
int cx3 = c2 + tri->inputs.step[2][i] * 4;
mask &= ~(1 << i);
/* Don't bother testing if the 4x4 block is entirely in/out of
* the triangle. It's a little faster to do it in the jit code.
*/
LP_COUNT(nr_non_empty_4);
do_block_4(rast_task, tri, px, py, cx1, cx2, cx3);
}
}
/**
* Scan the tile in chunks and figure out which pixels to rasterize
* for this triangle.
*/
void
lp_rast_triangle( struct lp_rasterizer *rast,
unsigned thread_index,
const union lp_rast_cmd_arg arg )
{
struct lp_rasterizer_task *rast_task = &rast->tasks[thread_index];
const struct lp_rast_triangle *tri = arg.triangle;
int x = rast_task->x;
int y = rast_task->y;
int ei[3], eo[3], c[3];
unsigned outmask, inmask, partial_mask;
unsigned i, j;
c[0] = tri->c1 + tri->dx12 * y - tri->dy12 * x;
c[1] = tri->c2 + tri->dx23 * y - tri->dy23 * x;
c[2] = tri->c3 + tri->dx31 * y - tri->dy31 * x;
eo[0] = tri->eo1 * 16;
eo[1] = tri->eo2 * 16;
eo[2] = tri->eo3 * 16;
ei[0] = tri->ei1 * 16;
ei[1] = tri->ei2 * 16;
ei[2] = tri->ei3 * 16;
outmask = 0;
inmask = 0xffff;
for (j = 0; j < 3; j++) {
const int *step = tri->inputs.step[j];
const int cox = c[j] + eo[j];
const int cio = ei[j]- eo[j];
/* Outmask has bits set whenever we are outside any of the
* edges.
*/
/* Inmask has bits set whenever we are inside all of the edges.
*/
for (i = 0; i < 16; i++) {
int out = cox + step[i] * 16;
int in = out + cio;
outmask |= (out >> 31) & (1 << i);
inmask &= ~((in >> 31) & (1 << i));
}
}
assert((outmask & inmask) == 0);
if (outmask == 0xffff)
return;
/* Invert mask, so that bits are set whenever we are at least
* partially inside all of the edges:
*/
partial_mask = ~inmask & ~outmask & 0xffff;
/* Iterate over partials:
*/
while (partial_mask) {
int i = ffs(partial_mask) - 1;
int px = x + pos_table16[i][0];
int py = y + pos_table16[i][1];
int cx1 = c[0] + tri->inputs.step[0][i] * 16;
int cx2 = c[1] + tri->inputs.step[1][i] * 16;
int cx3 = c[2] + tri->inputs.step[2][i] * 16;
partial_mask &= ~(1 << i);
LP_COUNT(nr_partially_covered_16);
do_block_16(rast_task, tri, px, py, cx1, cx2, cx3);
}
/* Iterate over fulls:
*/
while (inmask) {
int i = ffs(inmask) - 1;
int px = x + pos_table16[i][0];
int py = y + pos_table16[i][1];
inmask &= ~(1 << i);
LP_COUNT(nr_fully_covered_16);
block_full_16(rast_task, tri, px, py);
}
}
|