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
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
|
/*
* Adapted from code Copyright 2009-2010 NVIDIA Corporation,
* and code copyright 2009-2012 Intel Corporation
*
* Modifications Copyright 2011-2013, Blender Foundation.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#ifdef __QBVH__
# include "qbvh_shadow_all.h"
#endif
#if BVH_FEATURE(BVH_HAIR)
# define NODE_INTERSECT bvh_node_intersect
#else
# define NODE_INTERSECT bvh_aligned_node_intersect
#endif
/* This is a template BVH traversal function, where various features can be
* enabled/disabled. This way we can compile optimized versions for each case
* without new features slowing things down.
*
* BVH_INSTANCING: object instancing
* BVH_HAIR: hair curve rendering
* BVH_MOTION: motion blur rendering
*
*/
ccl_device bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg,
const Ray *ray,
Intersection *isect_array,
const uint max_hits,
uint *num_hits)
{
/* todo:
* - likely and unlikely for if() statements
* - test restrict attribute for pointers
*/
/* traversal stack in CUDA thread-local memory */
int traversalStack[BVH_STACK_SIZE];
traversalStack[0] = ENTRYPOINT_SENTINEL;
/* traversal variables in registers */
int stackPtr = 0;
int nodeAddr = kernel_data.bvh.root;
/* ray parameters in registers */
const float tmax = ray->t;
float3 P = ray->P;
float3 dir = bvh_clamp_direction(ray->D);
float3 idir = bvh_inverse_direction(dir);
int object = OBJECT_NONE;
float isect_t = tmax;
#if BVH_FEATURE(BVH_MOTION)
Transform ob_itfm;
#endif
#if BVH_FEATURE(BVH_INSTANCING)
int num_hits_in_instance = 0;
#endif
*num_hits = 0;
isect_array->t = tmax;
#if defined(__KERNEL_SSE2__)
const shuffle_swap_t shuf_identity = shuffle_swap_identity();
const shuffle_swap_t shuf_swap = shuffle_swap_swap();
const ssef pn = cast(ssei(0, 0, 0x80000000, 0x80000000));
ssef Psplat[3], idirsplat[3];
# if BVH_FEATURE(BVH_HAIR)
ssef tnear(0.0f), tfar(isect_t);
# endif
shuffle_swap_t shufflexyz[3];
Psplat[0] = ssef(P.x);
Psplat[1] = ssef(P.y);
Psplat[2] = ssef(P.z);
ssef tsplat(0.0f, 0.0f, -isect_t, -isect_t);
gen_idirsplat_swap(pn, shuf_identity, shuf_swap, idir, idirsplat, shufflexyz);
#endif /* __KERNEL_SSE2__ */
IsectPrecalc isect_precalc;
triangle_intersect_precalc(dir, &isect_precalc);
/* traversal loop */
do {
do {
/* traverse internal nodes */
while(nodeAddr >= 0 && nodeAddr != ENTRYPOINT_SENTINEL) {
int nodeAddrChild1, traverse_mask;
float dist[2];
float4 cnodes = kernel_tex_fetch(__bvh_nodes, nodeAddr+0);
#if !defined(__KERNEL_SSE2__)
traverse_mask = NODE_INTERSECT(kg,
P,
# if BVH_FEATURE(BVH_HAIR)
dir,
# endif
idir,
isect_t,
nodeAddr,
PATH_RAY_SHADOW,
dist);
#else // __KERNEL_SSE2__
traverse_mask = NODE_INTERSECT(kg,
P,
dir,
# if BVH_FEATURE(BVH_HAIR)
tnear,
tfar,
# endif
tsplat,
Psplat,
idirsplat,
shufflexyz,
nodeAddr,
PATH_RAY_SHADOW,
dist);
#endif // __KERNEL_SSE2__
nodeAddr = __float_as_int(cnodes.z);
nodeAddrChild1 = __float_as_int(cnodes.w);
if(traverse_mask == 3) {
/* Both children were intersected, push the farther one. */
bool closestChild1 = (dist[1] < dist[0]);
if(closestChild1) {
int tmp = nodeAddr;
nodeAddr = nodeAddrChild1;
nodeAddrChild1 = tmp;
}
++stackPtr;
kernel_assert(stackPtr < BVH_STACK_SIZE);
traversalStack[stackPtr] = nodeAddrChild1;
}
else {
/* One child was intersected. */
if(traverse_mask == 2) {
nodeAddr = nodeAddrChild1;
}
else if(traverse_mask == 0) {
/* Neither child was intersected. */
nodeAddr = traversalStack[stackPtr];
--stackPtr;
}
}
}
/* if node is leaf, fetch triangle list */
if(nodeAddr < 0) {
float4 leaf = kernel_tex_fetch(__bvh_leaf_nodes, (-nodeAddr-1));
int primAddr = __float_as_int(leaf.x);
#if BVH_FEATURE(BVH_INSTANCING)
if(primAddr >= 0) {
#endif
const int primAddr2 = __float_as_int(leaf.y);
const uint type = __float_as_int(leaf.w);
const uint p_type = type & PRIMITIVE_ALL;
/* pop */
nodeAddr = traversalStack[stackPtr];
--stackPtr;
/* primitive intersection */
while(primAddr < primAddr2) {
kernel_assert(kernel_tex_fetch(__prim_type, primAddr) == type);
bool hit;
/* todo: specialized intersect functions which don't fill in
* isect unless needed and check SD_HAS_TRANSPARENT_SHADOW?
* might give a few % performance improvement */
switch(p_type) {
case PRIMITIVE_TRIANGLE: {
hit = triangle_intersect(kg, &isect_precalc, isect_array, P, PATH_RAY_SHADOW, object, primAddr);
break;
}
#if BVH_FEATURE(BVH_MOTION)
case PRIMITIVE_MOTION_TRIANGLE: {
hit = motion_triangle_intersect(kg, isect_array, P, dir, ray->time, PATH_RAY_SHADOW, object, primAddr);
break;
}
#endif
#if BVH_FEATURE(BVH_HAIR)
case PRIMITIVE_CURVE:
case PRIMITIVE_MOTION_CURVE: {
if(kernel_data.curve.curveflags & CURVE_KN_INTERPOLATE)
hit = bvh_cardinal_curve_intersect(kg, isect_array, P, dir, PATH_RAY_SHADOW, object, primAddr, ray->time, type, NULL, 0, 0);
else
hit = bvh_curve_intersect(kg, isect_array, P, dir, PATH_RAY_SHADOW, object, primAddr, ray->time, type, NULL, 0, 0);
break;
}
#endif
default: {
hit = false;
break;
}
}
/* shadow ray early termination */
if(hit) {
/* detect if this surface has a shader with transparent shadows */
/* todo: optimize so primitive visibility flag indicates if
* the primitive has a transparent shadow shader? */
int prim = kernel_tex_fetch(__prim_index, isect_array->prim);
int shader = 0;
#ifdef __HAIR__
if(kernel_tex_fetch(__prim_type, isect_array->prim) & PRIMITIVE_ALL_TRIANGLE)
#endif
{
shader = kernel_tex_fetch(__tri_shader, prim);
}
#ifdef __HAIR__
else {
float4 str = kernel_tex_fetch(__curves, prim);
shader = __float_as_int(str.z);
}
#endif
int flag = kernel_tex_fetch(__shader_flag, (shader & SHADER_MASK)*2);
/* if no transparent shadows, all light is blocked */
if(!(flag & SD_HAS_TRANSPARENT_SHADOW)) {
return true;
}
/* if maximum number of hits reached, block all light */
else if(*num_hits == max_hits) {
return true;
}
/* move on to next entry in intersections array */
isect_array++;
(*num_hits)++;
#if BVH_FEATURE(BVH_INSTANCING)
num_hits_in_instance++;
#endif
isect_array->t = isect_t;
}
primAddr++;
}
}
#if BVH_FEATURE(BVH_INSTANCING)
else {
/* instance push */
object = kernel_tex_fetch(__prim_object, -primAddr-1);
# if BVH_FEATURE(BVH_MOTION)
bvh_instance_motion_push(kg, object, ray, &P, &dir, &idir, &isect_t, &ob_itfm);
# else
bvh_instance_push(kg, object, ray, &P, &dir, &idir, &isect_t);
# endif
triangle_intersect_precalc(dir, &isect_precalc);
num_hits_in_instance = 0;
isect_array->t = isect_t;
# if defined(__KERNEL_SSE2__)
Psplat[0] = ssef(P.x);
Psplat[1] = ssef(P.y);
Psplat[2] = ssef(P.z);
tsplat = ssef(0.0f, 0.0f, -isect_t, -isect_t);
# if BVH_FEATURE(BVH_HAIR)
tfar = ssef(isect_t);
# endif
gen_idirsplat_swap(pn, shuf_identity, shuf_swap, idir, idirsplat, shufflexyz);
# endif
++stackPtr;
kernel_assert(stackPtr < BVH_STACK_SIZE);
traversalStack[stackPtr] = ENTRYPOINT_SENTINEL;
nodeAddr = kernel_tex_fetch(__object_node, object);
}
}
#endif /* FEATURE(BVH_INSTANCING) */
} while(nodeAddr != ENTRYPOINT_SENTINEL);
#if BVH_FEATURE(BVH_INSTANCING)
if(stackPtr >= 0) {
kernel_assert(object != OBJECT_NONE);
if(num_hits_in_instance) {
float t_fac;
# if BVH_FEATURE(BVH_MOTION)
bvh_instance_motion_pop_factor(kg, object, ray, &P, &dir, &idir, &t_fac, &ob_itfm);
# else
bvh_instance_pop_factor(kg, object, ray, &P, &dir, &idir, &t_fac);
# endif
triangle_intersect_precalc(dir, &isect_precalc);
/* scale isect->t to adjust for instancing */
for(int i = 0; i < num_hits_in_instance; i++)
(isect_array-i-1)->t *= t_fac;
}
else {
float ignore_t = FLT_MAX;
# if BVH_FEATURE(BVH_MOTION)
bvh_instance_motion_pop(kg, object, ray, &P, &dir, &idir, &ignore_t, &ob_itfm);
# else
bvh_instance_pop(kg, object, ray, &P, &dir, &idir, &ignore_t);
# endif
triangle_intersect_precalc(dir, &isect_precalc);
}
isect_t = tmax;
isect_array->t = isect_t;
# if defined(__KERNEL_SSE2__)
Psplat[0] = ssef(P.x);
Psplat[1] = ssef(P.y);
Psplat[2] = ssef(P.z);
tsplat = ssef(0.0f, 0.0f, -isect_t, -isect_t);
# if BVH_FEATURE(BVH_HAIR)
tfar = ssef(isect_t);
# endif
gen_idirsplat_swap(pn, shuf_identity, shuf_swap, idir, idirsplat, shufflexyz);
# endif
object = OBJECT_NONE;
nodeAddr = traversalStack[stackPtr];
--stackPtr;
}
#endif /* FEATURE(BVH_INSTANCING) */
} while(nodeAddr != ENTRYPOINT_SENTINEL);
return false;
}
ccl_device_inline bool BVH_FUNCTION_NAME(KernelGlobals *kg,
const Ray *ray,
Intersection *isect_array,
const uint max_hits,
uint *num_hits)
{
#ifdef __QBVH__
if(kernel_data.bvh.use_qbvh) {
return BVH_FUNCTION_FULL_NAME(QBVH)(kg,
ray,
isect_array,
max_hits,
num_hits);
}
else
#endif
{
kernel_assert(kernel_data.bvh.use_qbvh == false);
return BVH_FUNCTION_FULL_NAME(BVH)(kg,
ray,
isect_array,
max_hits,
num_hits);
}
}
#undef BVH_FUNCTION_NAME
#undef BVH_FUNCTION_FEATURES
#undef NODE_INTERSECT
|