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
|
/*
* Copyright 2011-2017 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.
*/
/* CUDA kernel entry points */
#ifdef __CUDA_ARCH__
#include "kernel_config.h"
#include "kernel/kernel_compat_cuda.h"
#include "kernel/filter/filter_kernel.h"
/* kernels */
extern "C" __global__ void
CUDA_LAUNCH_BOUNDS(CUDA_THREADS_BLOCK_WIDTH, CUDA_KERNEL_MAX_REGISTERS)
kernel_cuda_filter_divide_shadow(int sample,
TilesInfo *tiles,
float *unfilteredA,
float *unfilteredB,
float *sampleVariance,
float *sampleVarianceV,
float *bufferVariance,
int4 prefilter_rect,
int buffer_pass_stride,
int buffer_denoising_offset,
bool use_split_variance)
{
int x = prefilter_rect.x + blockDim.x*blockIdx.x + threadIdx.x;
int y = prefilter_rect.y + blockDim.y*blockIdx.y + threadIdx.y;
if(x < prefilter_rect.z && y < prefilter_rect.w) {
kernel_filter_divide_shadow(sample,
tiles,
x, y,
unfilteredA,
unfilteredB,
sampleVariance,
sampleVarianceV,
bufferVariance,
prefilter_rect,
buffer_pass_stride,
buffer_denoising_offset,
use_split_variance);
}
}
extern "C" __global__ void
CUDA_LAUNCH_BOUNDS(CUDA_THREADS_BLOCK_WIDTH, CUDA_KERNEL_MAX_REGISTERS)
kernel_cuda_filter_get_feature(int sample,
TilesInfo *tiles,
int m_offset,
int v_offset,
float *mean,
float *variance,
int4 prefilter_rect,
int buffer_pass_stride,
int buffer_denoising_offset,
bool use_split_variance)
{
int x = prefilter_rect.x + blockDim.x*blockIdx.x + threadIdx.x;
int y = prefilter_rect.y + blockDim.y*blockIdx.y + threadIdx.y;
if(x < prefilter_rect.z && y < prefilter_rect.w) {
kernel_filter_get_feature(sample,
tiles,
m_offset, v_offset,
x, y,
mean, variance,
prefilter_rect,
buffer_pass_stride,
buffer_denoising_offset,
use_split_variance);
}
}
extern "C" __global__ void
CUDA_LAUNCH_BOUNDS(CUDA_THREADS_BLOCK_WIDTH, CUDA_KERNEL_MAX_REGISTERS)
kernel_cuda_filter_detect_outliers(float *image,
float *variance,
float *depth,
float *output,
int4 prefilter_rect,
int pass_stride)
{
int x = prefilter_rect.x + blockDim.x*blockIdx.x + threadIdx.x;
int y = prefilter_rect.y + blockDim.y*blockIdx.y + threadIdx.y;
if(x < prefilter_rect.z && y < prefilter_rect.w) {
kernel_filter_detect_outliers(x, y, image, variance, depth, output, prefilter_rect, pass_stride);
}
}
extern "C" __global__ void
CUDA_LAUNCH_BOUNDS(CUDA_THREADS_BLOCK_WIDTH, CUDA_KERNEL_MAX_REGISTERS)
kernel_cuda_filter_combine_halves(float *mean, float *variance, float *a, float *b, int4 prefilter_rect, int r)
{
int x = prefilter_rect.x + blockDim.x*blockIdx.x + threadIdx.x;
int y = prefilter_rect.y + blockDim.y*blockIdx.y + threadIdx.y;
if(x < prefilter_rect.z && y < prefilter_rect.w) {
kernel_filter_combine_halves(x, y, mean, variance, a, b, prefilter_rect, r);
}
}
extern "C" __global__ void
CUDA_LAUNCH_BOUNDS(CUDA_THREADS_BLOCK_WIDTH, CUDA_KERNEL_MAX_REGISTERS)
kernel_cuda_filter_construct_transform(float const* __restrict__ buffer,
float *transform, int *rank,
int4 filter_area, int4 rect,
int radius, float pca_threshold,
int pass_stride)
{
int x = blockDim.x*blockIdx.x + threadIdx.x;
int y = blockDim.y*blockIdx.y + threadIdx.y;
if(x < filter_area.z && y < filter_area.w) {
int *l_rank = rank + y*filter_area.z + x;
float *l_transform = transform + y*filter_area.z + x;
kernel_filter_construct_transform(buffer,
x + filter_area.x, y + filter_area.y,
rect, pass_stride,
l_transform, l_rank,
radius, pca_threshold,
filter_area.z*filter_area.w,
threadIdx.y*blockDim.x + threadIdx.x);
}
}
extern "C" __global__ void
CUDA_LAUNCH_BOUNDS(CUDA_THREADS_BLOCK_WIDTH, CUDA_KERNEL_MAX_REGISTERS)
kernel_cuda_filter_nlm_calc_difference(int dx, int dy,
const float *ccl_restrict weight_image,
const float *ccl_restrict variance_image,
float *difference_image,
int4 rect, int w,
int channel_offset,
float a, float k_2)
{
int x = blockDim.x*blockIdx.x + threadIdx.x + rect.x;
int y = blockDim.y*blockIdx.y + threadIdx.y + rect.y;
if(x < rect.z && y < rect.w) {
kernel_filter_nlm_calc_difference(x, y, dx, dy, weight_image, variance_image, difference_image, rect, w, channel_offset, a, k_2);
}
}
extern "C" __global__ void
CUDA_LAUNCH_BOUNDS(CUDA_THREADS_BLOCK_WIDTH, CUDA_KERNEL_MAX_REGISTERS)
kernel_cuda_filter_nlm_blur(const float *ccl_restrict difference_image, float *out_image, int4 rect, int w, int f)
{
int x = blockDim.x*blockIdx.x + threadIdx.x + rect.x;
int y = blockDim.y*blockIdx.y + threadIdx.y + rect.y;
if(x < rect.z && y < rect.w) {
kernel_filter_nlm_blur(x, y, difference_image, out_image, rect, w, f);
}
}
extern "C" __global__ void
CUDA_LAUNCH_BOUNDS(CUDA_THREADS_BLOCK_WIDTH, CUDA_KERNEL_MAX_REGISTERS)
kernel_cuda_filter_nlm_calc_weight(const float *ccl_restrict difference_image, float *out_image, int4 rect, int w, int f)
{
int x = blockDim.x*blockIdx.x + threadIdx.x + rect.x;
int y = blockDim.y*blockIdx.y + threadIdx.y + rect.y;
if(x < rect.z && y < rect.w) {
kernel_filter_nlm_calc_weight(x, y, difference_image, out_image, rect, w, f);
}
}
extern "C" __global__ void
CUDA_LAUNCH_BOUNDS(CUDA_THREADS_BLOCK_WIDTH, CUDA_KERNEL_MAX_REGISTERS)
kernel_cuda_filter_nlm_update_output(int dx, int dy,
const float *ccl_restrict difference_image,
const float *ccl_restrict image,
float *out_image, float *accum_image,
int4 rect, int w,
int f)
{
int x = blockDim.x*blockIdx.x + threadIdx.x + rect.x;
int y = blockDim.y*blockIdx.y + threadIdx.y + rect.y;
if(x < rect.z && y < rect.w) {
kernel_filter_nlm_update_output(x, y, dx, dy, difference_image, image, out_image, accum_image, rect, w, f);
}
}
extern "C" __global__ void
CUDA_LAUNCH_BOUNDS(CUDA_THREADS_BLOCK_WIDTH, CUDA_KERNEL_MAX_REGISTERS)
kernel_cuda_filter_nlm_normalize(float *out_image, const float *ccl_restrict accum_image, int4 rect, int w)
{
int x = blockDim.x*blockIdx.x + threadIdx.x + rect.x;
int y = blockDim.y*blockIdx.y + threadIdx.y + rect.y;
if(x < rect.z && y < rect.w) {
kernel_filter_nlm_normalize(x, y, out_image, accum_image, rect, w);
}
}
extern "C" __global__ void
CUDA_LAUNCH_BOUNDS(CUDA_THREADS_BLOCK_WIDTH, CUDA_KERNEL_MAX_REGISTERS)
kernel_cuda_filter_nlm_construct_gramian(int dx, int dy,
const float *ccl_restrict difference_image,
const float *ccl_restrict buffer,
float const* __restrict__ transform,
int *rank,
float *XtWX,
float3 *XtWY,
int4 rect,
int4 filter_rect,
int w, int h, int f,
int pass_stride)
{
int x = blockDim.x*blockIdx.x + threadIdx.x + max(0, rect.x-filter_rect.x);
int y = blockDim.y*blockIdx.y + threadIdx.y + max(0, rect.y-filter_rect.y);
if(x < min(filter_rect.z, rect.z-filter_rect.x) && y < min(filter_rect.w, rect.w-filter_rect.y)) {
kernel_filter_nlm_construct_gramian(x, y,
dx, dy,
difference_image,
buffer,
transform, rank,
XtWX, XtWY,
rect, filter_rect,
w, h, f,
pass_stride,
threadIdx.y*blockDim.x + threadIdx.x);
}
}
extern "C" __global__ void
CUDA_LAUNCH_BOUNDS(CUDA_THREADS_BLOCK_WIDTH, CUDA_KERNEL_MAX_REGISTERS)
kernel_cuda_filter_finalize(int w, int h,
float *buffer, int *rank,
float *XtWX, float3 *XtWY,
int4 filter_area, int4 buffer_params,
int sample)
{
int x = blockDim.x*blockIdx.x + threadIdx.x;
int y = blockDim.y*blockIdx.y + threadIdx.y;
if(x < filter_area.z && y < filter_area.w) {
int storage_ofs = y*filter_area.z+x;
rank += storage_ofs;
XtWX += storage_ofs;
XtWY += storage_ofs;
kernel_filter_finalize(x, y, w, h, buffer, rank, filter_area.z*filter_area.w, XtWX, XtWY, buffer_params, sample);
}
}
#endif
|
