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
|
/*
* 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 rays that hit the background (sceneintersect
* kernel), and for the rays of state RAY_UPDATE_BUFFER it updates the ray's
* accumulated radiance in the output buffer. This kernel also takes care of
* rays that have been determined to-be-regenerated.
*
* We will empty QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS queue in this kernel.
*
* Typically all rays that are in state RAY_HIT_BACKGROUND, RAY_UPDATE_BUFFER
* will be eventually set to RAY_TO_REGENERATE state in this kernel.
* Finally all rays of ray_state RAY_TO_REGENERATE will be regenerated and put
* in queue QUEUE_ACTIVE_AND_REGENERATED_RAYS.
*
* State of queues when this kernel is called:
* At entry,
* - QUEUE_ACTIVE_AND_REGENERATED_RAYS will be filled with RAY_ACTIVE rays.
* - QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be filled with
* RAY_UPDATE_BUFFER, RAY_HIT_BACKGROUND, RAY_TO_REGENERATE rays.
* At exit,
* - QUEUE_ACTIVE_AND_REGENERATED_RAYS will be filled with RAY_ACTIVE and
* RAY_REGENERATED rays.
* - QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be empty.
*/
ccl_device void kernel_buffer_update(KernelGlobals *kg,
ccl_local_param 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);
int ray_index = ccl_global_id(1) * ccl_global_size(0) + ccl_global_id(0);
if(ray_index == 0) {
/* We will empty this queue in this kernel. */
kernel_split_params.queue_index[QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS] = 0;
}
char enqueue_flag = 0;
ray_index = get_ray_index(kg, ray_index,
QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS,
kernel_split_state.queue_data,
kernel_split_params.queue_size,
1);
#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
int stride = kernel_split_params.stride;
ccl_global char *ray_state = kernel_split_state.ray_state;
ccl_global PathState *state = &kernel_split_state.path_state[ray_index];
PathRadiance *L = &kernel_split_state.path_radiance[ray_index];
ccl_global Ray *ray = &kernel_split_state.ray[ray_index];
ccl_global float3 *throughput = &kernel_split_state.throughput[ray_index];
if(IS_STATE(ray_state, ray_index, RAY_UPDATE_BUFFER)) {
uint work_index = kernel_split_state.work_array[ray_index];
uint sample = get_work_sample(kg, work_index, ray_index) + kernel_split_params.start_sample;
uint tile_x, tile_y, pixel_x, pixel_y;
get_work_pixel_tile_position(kg, &pixel_x, &pixel_y, &tile_x, &tile_y, work_index, ray_index);
ccl_global float *buffer = kernel_split_params.buffer;
buffer += (kernel_split_params.offset + pixel_x + pixel_y*stride) * kernel_data.film.pass_stride;
/* accumulate result in output buffer */
bool is_shadow_catcher = (state->flag & PATH_RAY_SHADOW_CATCHER);
kernel_write_result(kg, buffer, sample, L, is_shadow_catcher);
ASSIGN_RAY_STATE(ray_state, ray_index, RAY_TO_REGENERATE);
}
if(IS_STATE(ray_state, ray_index, RAY_TO_REGENERATE)) {
/* We have completed current work; So get next work */
uint work_index;
int valid_work = get_next_work(kg, &work_index, ray_index);
if(!valid_work) {
/* If work is invalid, this means no more work is available and the thread may exit */
ASSIGN_RAY_STATE(ray_state, ray_index, RAY_INACTIVE);
}
if(IS_STATE(ray_state, ray_index, RAY_TO_REGENERATE)) {
kernel_split_state.work_array[ray_index] = work_index;
/* Get the sample associated with the current work */
uint sample = get_work_sample(kg, work_index, ray_index) + kernel_split_params.start_sample;
/* Get pixel and tile position associated with current work */
uint tile_x, tile_y, pixel_x, pixel_y;
get_work_pixel_tile_position(kg, &pixel_x, &pixel_y, &tile_x, &tile_y, work_index, ray_index);
/* Remap rng_state according to the current work */
ccl_global uint *rng_state = kernel_split_params.rng_state;
rng_state += kernel_split_params.offset + pixel_x + pixel_y*stride;
/* Remap buffer according to the current work */
ccl_global float *buffer = kernel_split_params.buffer;
buffer += (kernel_split_params.offset + pixel_x + pixel_y*stride) * kernel_data.film.pass_stride;
/* Initialize random numbers and ray. */
uint rng_hash;
kernel_path_trace_setup(kg, rng_state, sample, pixel_x, pixel_y, &rng_hash, ray);
if(ray->t != 0.0f) {
/* Initialize throughput, path radiance, Ray, PathState;
* These rays proceed with path-iteration.
*/
*throughput = make_float3(1.0f, 1.0f, 1.0f);
path_radiance_init(L, kernel_data.film.use_light_pass);
path_state_init(kg, &kernel_split_state.sd_DL_shadow[ray_index], state, rng_hash, sample, ray);
#ifdef __SUBSURFACE__
kernel_path_subsurface_init_indirect(&kernel_split_state.ss_rays[ray_index]);
#endif
ASSIGN_RAY_STATE(ray_state, ray_index, RAY_REGENERATED);
enqueue_flag = 1;
}
else {
/* These rays do not participate in path-iteration. */
float4 L_rad = make_float4(0.0f, 0.0f, 0.0f, 0.0f);
/* Accumulate result in output buffer. */
kernel_write_pass_float4(buffer, sample, L_rad);
ASSIGN_RAY_STATE(ray_state, ray_index, RAY_TO_REGENERATE);
}
}
}
#ifndef __COMPUTE_DEVICE_GPU__
}
#endif
/* Enqueue RAY_REGENERATED rays into QUEUE_ACTIVE_AND_REGENERATED_RAYS;
* These rays will be made active during next SceneIntersectkernel.
*/
enqueue_ray_index_local(ray_index,
QUEUE_ACTIVE_AND_REGENERATED_RAYS,
enqueue_flag,
kernel_split_params.queue_size,
local_queue_atomics,
kernel_split_state.queue_data,
kernel_split_params.queue_index);
}
CCL_NAMESPACE_END
|
