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
|
/*
* Copyright 2011-2015 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.
*/
CCL_NAMESPACE_BEGIN
/*This kernel takes care of setting up ray for the next iteration of
* path-iteration and accumulating radiance corresponding to AO and
* direct-lighting
*
* Ray state of rays that are terminated in this kernel are changed
* to RAY_UPDATE_BUFFER.
*
* Note on queues:
* This kernel fetches rays from the queue QUEUE_ACTIVE_AND_REGENERATED_RAYS
* and processes only the rays of state RAY_ACTIVE.
* There are different points in this kernel where a ray may terminate and
* reach RAY_UPDATE_BUFF state. These rays are enqueued into
* QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS queue. These rays will still be present
* in QUEUE_ACTIVE_AND_REGENERATED_RAYS queue, but since their ray-state has
* been changed to RAY_UPDATE_BUFF, there is no problem.
*
* State of queues when this kernel is called:
* At entry,
* - QUEUE_ACTIVE_AND_REGENERATED_RAYS will be filled with RAY_ACTIVE,
* RAY_REGENERATED, RAY_UPDATE_BUFFER rays.
* - QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be filled with
* RAY_TO_REGENERATE and RAY_UPDATE_BUFFER rays.
* At exit,
* - QUEUE_ACTIVE_AND_REGENERATED_RAYS will be filled with RAY_ACTIVE,
* RAY_REGENERATED and more RAY_UPDATE_BUFFER rays.
* - QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be filled with
* RAY_TO_REGENERATE and more RAY_UPDATE_BUFFER rays.
*/
ccl_device void kernel_next_iteration_setup(KernelGlobals *kg)
{
ccl_local unsigned int local_queue_atomics;
if(ccl_local_id(0) == 0 && ccl_local_id(1) == 0) {
local_queue_atomics = 0;
}
ccl_barrier(CCL_LOCAL_MEM_FENCE);
if(ccl_global_id(0) == 0 && ccl_global_id(1) == 0) {
/* If we are here, then it means that scene-intersect kernel
* has already been executed atleast once. From the next time,
* scene-intersect kernel may operate on queues to fetch ray index
*/
*kernel_split_params.use_queues_flag = 1;
/* Mark queue indices of QUEUE_SHADOW_RAY_CAST_AO_RAYS and
* QUEUE_SHADOW_RAY_CAST_DL_RAYS queues that were made empty during the
* previous kernel.
*/
kernel_split_params.queue_index[QUEUE_SHADOW_RAY_CAST_AO_RAYS] = 0;
kernel_split_params.queue_index[QUEUE_SHADOW_RAY_CAST_DL_RAYS] = 0;
}
char enqueue_flag = 0;
int ray_index = ccl_global_id(1) * ccl_global_size(0) + ccl_global_id(0);
ray_index = get_ray_index(kg, ray_index,
QUEUE_ACTIVE_AND_REGENERATED_RAYS,
kernel_split_state.queue_data,
kernel_split_params.queue_size,
0);
#ifdef __COMPUTE_DEVICE_GPU__
/* If we are executing on a GPU device, we exit all threads that are not
* required.
*
* If we are executing on a CPU device, then we need to keep all threads
* active since we have barrier() calls later in the kernel. CPU devices,
* expect all threads to execute barrier statement.
*/
if(ray_index == QUEUE_EMPTY_SLOT) {
return;
}
#endif
#ifndef __COMPUTE_DEVICE_GPU__
if(ray_index != QUEUE_EMPTY_SLOT) {
#endif
/* Load ShaderData structure. */
PathRadiance *L = NULL;
ccl_global PathState *state = NULL;
ccl_global char *ray_state = kernel_split_state.ray_state;
/* Path radiance update for AO/Direct_lighting's shadow blocked. */
if(IS_FLAG(ray_state, ray_index, RAY_SHADOW_RAY_CAST_DL) ||
IS_FLAG(ray_state, ray_index, RAY_SHADOW_RAY_CAST_AO))
{
state = &kernel_split_state.path_state[ray_index];
L = &kernel_split_state.path_radiance[ray_index];
float3 _throughput = kernel_split_state.throughput[ray_index];
if(IS_FLAG(ray_state, ray_index, RAY_SHADOW_RAY_CAST_AO)) {
float3 shadow = kernel_split_state.ao_light_ray[ray_index].P;
// TODO(mai): investigate correctness here
char update_path_radiance = (char)kernel_split_state.ao_light_ray[ray_index].t;
if(update_path_radiance) {
path_radiance_accum_ao(L,
_throughput,
kernel_split_state.ao_alpha[ray_index],
kernel_split_state.ao_bsdf[ray_index],
shadow,
state->bounce);
}
REMOVE_RAY_FLAG(ray_state, ray_index, RAY_SHADOW_RAY_CAST_AO);
}
if(IS_FLAG(ray_state, ray_index, RAY_SHADOW_RAY_CAST_DL)) {
float3 shadow = kernel_split_state.light_ray[ray_index].P;
// TODO(mai): investigate correctness here
char update_path_radiance = (char)kernel_split_state.light_ray[ray_index].t;
if(update_path_radiance) {
BsdfEval L_light = kernel_split_state.bsdf_eval[ray_index];
path_radiance_accum_light(L,
_throughput,
&L_light,
shadow,
1.0f,
state->bounce,
kernel_split_state.is_lamp[ray_index]);
}
REMOVE_RAY_FLAG(ray_state, ray_index, RAY_SHADOW_RAY_CAST_DL);
}
}
if(IS_STATE(ray_state, ray_index, RAY_ACTIVE)) {
ccl_global float3 *throughput = &kernel_split_state.throughput[ray_index];
ccl_global Ray *ray = &kernel_split_state.ray[ray_index];
ccl_global RNG *rng = &kernel_split_state.rng[ray_index];
state = &kernel_split_state.path_state[ray_index];
L = &kernel_split_state.path_radiance[ray_index];
/* Compute direct lighting and next bounce. */
if(!kernel_path_surface_bounce(kg, rng, &kernel_split_state.sd[ray_index], throughput, state, L, ray)) {
ASSIGN_RAY_STATE(ray_state, ray_index, RAY_UPDATE_BUFFER);
enqueue_flag = 1;
}
}
#ifndef __COMPUTE_DEVICE_GPU__
}
#endif
/* Enqueue RAY_UPDATE_BUFFER rays. */
enqueue_ray_index_local(ray_index,
QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS,
enqueue_flag,
kernel_split_params.queue_size,
&local_queue_atomics,
kernel_split_state.queue_data,
kernel_split_params.queue_index);
}
CCL_NAMESPACE_END
|
