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
|
/*
* Copyright 2011, Blender Foundation.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version 2
* of the License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software Foundation,
* Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
*
* Contributor:
* Jeroen Bakker
* Monique Dewanchand
*/
#include <list>
#include <stdio.h>
#include "BKE_global.h"
#include "COM_WorkScheduler.h"
#include "COM_CPUDevice.h"
#include "COM_OpenCLDevice.h"
#include "COM_OpenCLKernels.cl.h"
#include "OCL_opencl.h"
#include "COM_WriteBufferOperation.h"
#include "PIL_time.h"
#include "BLI_threads.h"
#if COM_CURRENT_THREADING_MODEL == COM_TM_NOTHREAD
# ifndef DEBUG /* test this so we dont get warnings in debug builds */
# warning COM_CURRENT_THREADING_MODEL COM_TM_NOTHREAD is activated. Use only for debugging.
# endif
#elif COM_CURRENT_THREADING_MODEL == COM_TM_QUEUE
/* do nothing - default */
#else
# error COM_CURRENT_THREADING_MODEL No threading model selected
#endif
/// @brief list of all CPUDevices. for every hardware thread an instance of CPUDevice is created
static vector<CPUDevice *> g_cpudevices;
#if COM_CURRENT_THREADING_MODEL == COM_TM_QUEUE
/// @brief list of all thread for every CPUDevice in cpudevices a thread exists
static ListBase g_cputhreads;
/// @brief all scheduled work for the cpu
static ThreadQueue *g_cpuqueue;
static ThreadQueue *g_gpuqueue;
#ifdef COM_OPENCL_ENABLED
static cl_context g_context;
static cl_program g_program;
/// @brief list of all OpenCLDevices. for every OpenCL GPU device an instance of OpenCLDevice is created
static vector<OpenCLDevice *> g_gpudevices;
/// @brief list of all thread for every GPUDevice in cpudevices a thread exists
static ListBase g_gputhreads;
/// @brief all scheduled work for the gpu
#ifdef COM_OPENCL_ENABLED
static bool g_openclActive = false;
#endif
#endif
#endif
#define MAX_HIGHLIGHT 8
extern "C" {
int g_highlightIndex;
void ** g_highlightedNodes;
void ** g_highlightedNodesRead;
#define HIGHLIGHT(wp) \
{ \
ExecutionGroup *group = wp->getExecutionGroup(); \
if (group->isComplex()) { \
NodeOperation *operation = group->getOutputNodeOperation(); \
if (operation->isWriteBufferOperation()) { \
WriteBufferOperation *writeOperation = (WriteBufferOperation *)operation; \
NodeOperation *complexOperation = writeOperation->getInput(); \
bNode *node = complexOperation->getbNode(); \
if (node) { \
if (node->original) { \
node = node->original; \
} \
if (g_highlightIndex < MAX_HIGHLIGHT) { \
g_highlightedNodes[g_highlightIndex++] = node; \
} \
} \
} \
} \
}
void COM_startReadHighlights()
{
if (g_highlightedNodesRead) {
delete [] g_highlightedNodesRead;
}
g_highlightedNodesRead = g_highlightedNodes;
g_highlightedNodes = new void *[MAX_HIGHLIGHT];
g_highlightIndex = 0;
for (int i = 0 ; i < MAX_HIGHLIGHT; i++) {
g_highlightedNodes[i] = 0;
}
}
int COM_isHighlightedbNode(bNode *bnode)
{
if (!g_highlightedNodesRead) return false;
for (int i = 0 ; i < MAX_HIGHLIGHT; i++) {
void *p = g_highlightedNodesRead[i];
if (!p) return false;
if (p == bnode) return true;
}
return false;
}
} // end extern "C"
#if COM_CURRENT_THREADING_MODEL == COM_TM_QUEUE
void *WorkScheduler::thread_execute_cpu(void *data)
{
Device *device = (Device *)data;
WorkPackage *work;
while ((work = (WorkPackage *)BLI_thread_queue_pop(g_cpuqueue))) {
HIGHLIGHT(work);
device->execute(work);
delete work;
}
return NULL;
}
void *WorkScheduler::thread_execute_gpu(void *data)
{
Device *device = (Device *)data;
WorkPackage *work;
while ((work = (WorkPackage *)BLI_thread_queue_pop(g_gpuqueue))) {
HIGHLIGHT(work);
device->execute(work);
delete work;
}
return NULL;
}
#endif
void WorkScheduler::schedule(ExecutionGroup *group, int chunkNumber)
{
WorkPackage *package = new WorkPackage(group, chunkNumber);
#if COM_CURRENT_THREADING_MODEL == COM_TM_NOTHREAD
CPUDevice device;
device.execute(package);
delete package;
#elif COM_CURRENT_THREADING_MODEL == COM_TM_QUEUE
#ifdef COM_OPENCL_ENABLED
if (group->isOpenCL() && g_openclActive) {
BLI_thread_queue_push(g_gpuqueue, package);
}
else {
BLI_thread_queue_push(g_cpuqueue, package);
}
#else
BLI_thread_queue_push(cpuqueue, package);
#endif
#endif
}
void WorkScheduler::start(CompositorContext &context)
{
#if COM_CURRENT_THREADING_MODEL == COM_TM_QUEUE
unsigned int index;
g_cpuqueue = BLI_thread_queue_init();
BLI_init_threads(&g_cputhreads, thread_execute_cpu, g_cpudevices.size());
for (index = 0; index < g_cpudevices.size(); index++) {
Device *device = g_cpudevices[index];
BLI_insert_thread(&g_cputhreads, device);
}
#ifdef COM_OPENCL_ENABLED
if (context.getHasActiveOpenCLDevices()) {
g_gpuqueue = BLI_thread_queue_init();
BLI_init_threads(&g_gputhreads, thread_execute_gpu, g_gpudevices.size());
for (index = 0; index < g_gpudevices.size(); index++) {
Device *device = g_gpudevices[index];
BLI_insert_thread(&g_gputhreads, device);
}
g_openclActive = true;
}
else {
g_openclActive = false;
}
#endif
#endif
}
void WorkScheduler::finish()
{
#if COM_CURRENT_THREADING_MODEL == COM_TM_QUEUE
#ifdef COM_OPENCL_ENABLED
if (g_openclActive) {
BLI_thread_queue_wait_finish(g_gpuqueue);
BLI_thread_queue_wait_finish(g_cpuqueue);
}
else {
BLI_thread_queue_wait_finish(g_cpuqueue);
}
#else
BLI_thread_queue_wait_finish(cpuqueue);
#endif
#endif
}
void WorkScheduler::stop()
{
#if COM_CURRENT_THREADING_MODEL == COM_TM_QUEUE
BLI_thread_queue_nowait(g_cpuqueue);
BLI_end_threads(&g_cputhreads);
BLI_thread_queue_free(g_cpuqueue);
g_cpuqueue = NULL;
#ifdef COM_OPENCL_ENABLED
if (g_openclActive) {
BLI_thread_queue_nowait(g_gpuqueue);
BLI_end_threads(&g_gputhreads);
BLI_thread_queue_free(g_gpuqueue);
g_gpuqueue = NULL;
}
#endif
#endif
}
bool WorkScheduler::hasGPUDevices()
{
#if COM_CURRENT_THREADING_MODEL == COM_TM_QUEUE
#ifdef COM_OPENCL_ENABLED
return g_gpudevices.size() > 0;
#else
return 0;
#endif
#else
return 0;
#endif
}
extern void clContextError(const char *errinfo, const void *private_info, size_t cb, void *user_data)
{
printf("OPENCL error: %s\n", errinfo);
}
void WorkScheduler::initialize()
{
g_highlightedNodesRead = 0;
g_highlightedNodes = 0;
COM_startReadHighlights();
#if COM_CURRENT_THREADING_MODEL == COM_TM_QUEUE
int numberOfCPUThreads = BLI_system_thread_count();
for (int index = 0; index < numberOfCPUThreads; index++) {
CPUDevice *device = new CPUDevice();
device->initialize();
g_cpudevices.push_back(device);
}
#ifdef COM_OPENCL_ENABLED
g_context = NULL;
g_program = NULL;
if (clCreateContextFromType) {
cl_uint numberOfPlatforms = 0;
cl_int error;
error = clGetPlatformIDs(0, 0, &numberOfPlatforms);
if (error != CL_SUCCESS) { printf("CLERROR[%d]: %s\n", error, clewErrorString(error)); }
if (G.f & G_DEBUG) printf("%d number of platforms\n", numberOfPlatforms);
cl_platform_id *platforms = new cl_platform_id[numberOfPlatforms];
error = clGetPlatformIDs(numberOfPlatforms, platforms, 0);
unsigned int indexPlatform;
for (indexPlatform = 0; indexPlatform < numberOfPlatforms; indexPlatform++) {
cl_platform_id platform = platforms[indexPlatform];
cl_uint numberOfDevices = 0;
clGetDeviceIDs(platform, CL_DEVICE_TYPE_GPU, 0, 0, &numberOfDevices);
if (numberOfDevices>0) {
cl_device_id *cldevices = new cl_device_id[numberOfDevices];
clGetDeviceIDs(platform, CL_DEVICE_TYPE_GPU, numberOfDevices, cldevices, 0);
g_context = clCreateContext(NULL, numberOfDevices, cldevices, clContextError, NULL, &error);
if (error != CL_SUCCESS) { printf("CLERROR[%d]: %s\n", error, clewErrorString(error)); }
const char *cl_str[2] = {clkernelstoh_COM_OpenCLKernels_cl, NULL};
g_program = clCreateProgramWithSource(g_context, 1, cl_str, 0, &error);
error = clBuildProgram(g_program, numberOfDevices, cldevices, 0, 0, 0);
if (error != CL_SUCCESS) {
cl_int error2;
size_t ret_val_size = 0;
printf("CLERROR[%d]: %s\n", error, clewErrorString(error));
error2 = clGetProgramBuildInfo(g_program, cldevices[0], CL_PROGRAM_BUILD_LOG, 0, NULL, &ret_val_size);
if (error2 != CL_SUCCESS) { printf("CLERROR[%d]: %s\n", error, clewErrorString(error)); }
char *build_log = new char[ret_val_size + 1];
error2 = clGetProgramBuildInfo(g_program, cldevices[0], CL_PROGRAM_BUILD_LOG, ret_val_size, build_log, NULL);
if (error2 != CL_SUCCESS) { printf("CLERROR[%d]: %s\n", error, clewErrorString(error)); }
build_log[ret_val_size] = '\0';
printf("%s", build_log);
delete build_log;
}
else {
unsigned int indexDevices;
for (indexDevices = 0; indexDevices < numberOfDevices; indexDevices++) {
cl_device_id device = cldevices[indexDevices];
cl_int vendorID = 0;
cl_int error2 = clGetDeviceInfo(device, CL_DEVICE_VENDOR_ID, sizeof(cl_int), &vendorID, NULL);
if (error2 != CL_SUCCESS) { printf("CLERROR[%d]: %s\n", error2, clewErrorString(error2)); }
OpenCLDevice *clDevice = new OpenCLDevice(g_context, device, g_program, vendorID);
clDevice->initialize();
g_gpudevices.push_back(clDevice);
}
}
delete[] cldevices;
}
}
delete[] platforms;
}
#endif
#endif
}
void WorkScheduler::deinitialize()
{
#if COM_CURRENT_THREADING_MODEL == COM_TM_QUEUE
Device *device;
while (g_cpudevices.size() > 0) {
device = g_cpudevices.back();
g_cpudevices.pop_back();
device->deinitialize();
delete device;
}
#ifdef COM_OPENCL_ENABLED
while (g_gpudevices.size() > 0) {
device = g_gpudevices.back();
g_gpudevices.pop_back();
device->deinitialize();
delete device;
}
if (g_program) {
clReleaseProgram(g_program);
g_program = NULL;
}
if (g_context) {
clReleaseContext(g_context);
g_context = NULL;
}
#endif
#endif
}
|