-
Notifications
You must be signed in to change notification settings - Fork 9
/
techniqueMegakernel.cuh
1873 lines (1588 loc) · 55.9 KB
/
techniqueMegakernel.cuh
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
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
// Project Whippletree
// http://www.icg.tugraz.at/project/parallel
//
// Copyright (C) 2014 Institute for Computer Graphics and Vision,
// Graz University of Technology
//
// Author(s): Markus Steinberger - steinberger ( at ) icg.tugraz.at
// Michael Kenzel - kenzel ( at ) icg.tugraz.at
// Pedro Boechat - boechat ( at ) icg.tugraz.at
// Bernhard Kerbl - kerbl ( at ) icg.tugraz.at
// Mark Dokter - dokter ( at ) icg.tugraz.at
// Dieter Schmalstieg - schmalstieg ( at ) icg.tugraz.at
//
// Permission is hereby granted, free of charge, to any person obtaining a copy
// of this software and associated documentation files (the "Software"), to deal
// in the Software without restriction, including without limitation the rights
// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
// copies of the Software, and to permit persons to whom the Software is
// furnished to do so, subject to the following conditions:
//
// The above copyright notice and this permission notice shall be included in
// all copies or substantial portions of the Software.
//
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
// THE SOFTWARE.
//
#pragma once
#include <memory>
#include <vector>
#include <tools/utils.h>
#include <tools/cuda_memory.h>
#include <iostream>
#include <unistd.h>
#include <cxxabi.h>
#include "timing.h"
#include "delay.cuh"
#include "techniqueInterface.h"
#include "procinfoTemplate.cuh"
#include "queuingMultiPhase.cuh"
#define PROC_MAX_NUM 64
#define SM_MAX_NUM 50
#define MEGAKERNEL_MAX_PROC_NUM 10
#define BLOCK_PER_SM_MAX_NUM 100
int streamIndex;
namespace SegmentedStorage
{
void checkReinitStorage();
}
/*
template <int type>
struct UserProcType;
*/
template <int type>
struct UserQueueType;
template <class A>
void printClassName()
{
int status;
printf("%s\n", abi::__cxa_demangle(typeid(A).name(), 0, 0, &status));
}
extern __device__ void* queuePointers[PROC_MAX_NUM];
namespace Megakernel
{
// ZHENG Zhen added.
__device__ inline uint zz_smid()
{
uint mysmid;
asm("mov.u32 %0, %smid;" : "=r" (mysmid));
return mysmid;
}
cudaStream_t streamArray[PROC_MAX_NUM];
enum MegakernelStopCriteria
{
// Stop megakernel, when the task queue is empty.
EmptyQueue,
// Stop megakernel, when the task queue is empty,
// and "shutdown" indicator is filled with "true" value.
ShutdownIndicator,
};
extern __device__ volatile int doneCounter[PROC_MAX_NUM];
extern __device__ volatile int endCounter[PROC_MAX_NUM];
// ZHENG Zhen added.
extern __device__ volatile int sm_flag[PROC_MAX_NUM * SM_MAX_NUM];
//__device__ int proc_exe_count[PROC_MAX_NUM];
extern __device__ int block_count[PROC_MAX_NUM * SM_MAX_NUM];
extern __device__ int group_done_flag[PROC_MAX_NUM];
//extern int numGroups;
//extern int procGroupArray[PROC_MAX_NUM];
// extern __device__ int *procIdArray_global; //[PROC_MAX_NUM];
extern int taskCountArray[PROC_MAX_NUM];
extern __device__ int resultCounter[PROC_MAX_NUM];
template<class InitProc, class Q>
__global__ void initData(Q* q, int num)
{
int id = blockIdx.x*blockDim.x + threadIdx.x;
for( ; id < num; id += blockDim.x*gridDim.x)
{
InitProc::template init<Q>(q, id);
}
}
// add by ZHENG Zhen，用来insert Data
template<class InitProc, class Q, typename ExpectedData>
__global__ void initData(Q* q, int num, ExpectedData *data)
{
//q->printName();
int id = blockIdx.x*blockDim.x + threadIdx.x;
for( ; id < num; id += blockDim.x*gridDim.x)
{
InitProc::template init<Q>(q, id, data);
// q is CurrentMultiphaseQueue
// q->printName();
}
}
template<class Q>
__global__ void recordData(Q* q)
{
q->record();
}
template<class Q>
__global__ void resetData(Q* q)
{
q->reset();
}
template<class Q, class ProcInfo, class PROC, class CUSTOM, bool Itemized, bool MultiElement>
class FuncCaller;
// ??? 一个block不可以多个task同时
template<class Q, class ProcInfo, class PROC, class CUSTOM>
class FuncCaller<Q, ProcInfo, PROC, CUSTOM, false, false>
{
public:
__device__ __inline__
static void call(Q* queue, void* data, int hasData, uint* shared)
{
//if(threadIdx.x == 0) {printf("1111111111\n");}
int nThreads;
if(PROC::NumThreads != 0)
nThreads = PROC::NumThreads;
else
nThreads = blockDim.x;
if(PROC::NumThreads == 0 || threadIdx.x < nThreads)
{
PROC :: template execute<Q, Context<PROC::NumThreads, false, CUSTOM> >(
threadIdx.x, nThreads, queue,
reinterpret_cast<typename PROC::ExpectedData*>(data), shared);
}
}
};
// ??? 一个block可以多个task同时
template<class Q, class ProcInfo, class PROC, class CUSTOM>
class FuncCaller<Q, ProcInfo, PROC, CUSTOM, false, true>
{
public:
__device__ __inline__
static void call(Q* queue, void* data, int hasData, uint* shared)
{
//if(threadIdx.x == 0) {printf("22222222222222\n");}
if(PROC::NumThreads != 0)
{
int nThreads;
nThreads = PROC::NumThreads;
int tid = threadIdx.x % PROC::NumThreads;
int offset = threadIdx.x / PROC::NumThreads;
if(threadIdx.x < hasData)
{
PROC :: template execute<Q, Context<PROC::NumThreads, true, CUSTOM> >(
tid, nThreads, queue, reinterpret_cast<typename PROC::ExpectedData*>(data),
shared + offset*PROC::sharedMemory/sizeof(uint) );
}
}
else
{
PROC :: template execute<Q, Context<PROC::NumThreads, true, CUSTOM> >(
threadIdx.x, blockDim.x, queue,
reinterpret_cast<typename PROC::ExpectedData*>(data), shared);
}
}
};
// ??? 所有data在同一个block中
template<class Q, class ProcInfo, class PROC, class CUSTOM, bool MultiElement>
class FuncCaller<Q, ProcInfo, PROC, CUSTOM, true, MultiElement>
{
public:
__device__ __inline__
static void call(Q* queue, void* data, int numData, uint* shared)
{
// ZHENG Zhen added, this one
//if(threadIdx.x == 0) {printf("33333333333333\n");}
if(threadIdx.x < numData)
{
PROC :: template execute<Q, Context<PROC::NumThreads, MultiElement, CUSTOM> >(
threadIdx.x, numData, queue,
reinterpret_cast<typename PROC::ExpectedData*>(data), shared);
}
}
};
////////////////////////////////////////////////////////////////////////////////////////
template<class Q, class ProcInfo, bool MultiElement>
struct ProcCallCopyVisitor
{
int* execproc;
const uint4 & sharedMem;
Q* q;
void* execData;
uint* s_data;
int hasResult;
__inline__ __device__ ProcCallCopyVisitor(Q* q, int *execproc,
void * execData, uint* s_data, const uint4& sharedMem, int hasResult )
: execproc(execproc), sharedMem(sharedMem), q(q), execData(execData), s_data(s_data) { }
// 调用指定的TProcedure的execute函数
template<class TProcedure, class CUSTOM>
__device__ __inline__ bool visit()
{
if(*execproc == findProcId<ProcInfo, TProcedure>::value)
{
FuncCaller<Q, ProcInfo, TProcedure, CUSTOM, TProcedure :: ItemInput, MultiElement>::call(
q, execData, hasResult, s_data + sharedMem.x + sharedMem.y + sharedMem.w );
return true;
}
return false;
}
};
// TODO： 和Copy的区别是什么？？？
template<class Q, class ProcInfo, bool MultiElement>
struct ProcCallNoCopyVisitor
{
int* execproc;
const uint4 & sharedMem;
Q* q;
void* execData;
uint* s_data;
int hasResult;
int processCount;
__inline__ __device__ ProcCallNoCopyVisitor(Q* q, int *execproc,
void * execData, uint* s_data, const uint4& sharedMem, int hasResult )
: execproc(execproc), sharedMem(sharedMem), q(q), execData(execData), s_data(s_data), hasResult(hasResult) { }
template<class TProcedure, class CUSTOM>
__device__ __inline__ bool visit()
{
if(*execproc == findProcId<ProcInfo, TProcedure>::value)
{
// ZHENG Zhen added, this one
// printf("%s in %s, at line %d\n", __FUNCTION__, __FILE__, __LINE__);
FuncCaller<Q, ProcInfo, TProcedure, CUSTOM, TProcedure :: ItemInput, MultiElement>::call(
q, execData, hasResult, s_data + sharedMem.x + sharedMem.y + sharedMem.w );
int n = TProcedure::NumThreads != 0 ? hasResult / TProcedure ::NumThreads
: (TProcedure ::ItemInput ? hasResult : 1);
processCount = n;
q-> template finishRead<TProcedure>(execproc[1], n);
return true;
}
return false;
}
};
#define PROCCALLNOCOPYPART(LAUNCHNUM) \
template<class Q, class ProcInfo, bool MultiElement> \
struct ProcCallNoCopyVisitorPart ## LAUNCHNUM \
{ \
int* execproc; \
const uint4 & sharedMem; \
Q* q; \
void* execData; \
uint* s_data; \
int hasResult; \
__inline__ __device__ ProcCallNoCopyVisitorPart ## LAUNCHNUM (Q* q, int *execproc, void * execData, uint* s_data, const uint4& sharedMem, int hasResult ) : execproc(execproc), sharedMem(sharedMem), q(q), execData(execData), s_data(s_data), hasResult(hasResult) { } \
template<class TProcedure, class CUSTOM> \
__device__ __inline__ bool visit() \
{ \
if(*execproc == TProcedure::ProcedureId) \
{ \
FuncCaller<Q, ProcInfo, TProcedure, CUSTOM, TProcedure :: ItemInput, MultiElement>::call(q, execData, hasResult, s_data + sharedMem.x + sharedMem.y + sharedMem.w ); \
int n = TProcedure::NumThreads != 0 ? hasResult / TProcedure ::NumThreads : (TProcedure ::ItemInput ? hasResult : 1); \
q-> template finishRead ## LAUNCHNUM <TProcedure>(execproc[1], n); \
return true; \
} \
return false; \
} \
};
PROCCALLNOCOPYPART(1)
PROCCALLNOCOPYPART(2)
PROCCALLNOCOPYPART(3)
#undef PROCCALLNOCOPYPART
extern __device__ int maxConcurrentBlocks[PROC_MAX_NUM];
extern __device__ volatile int maxConcurrentBlockEvalDone[PROC_MAX_NUM];
void initAll()
{
int nblocks[PROC_MAX_NUM];
memset(nblocks, 0, sizeof(int)*PROC_MAX_NUM);
CUDA_CHECKED_CALL(cudaMemcpyToSymbol(maxConcurrentBlocks,
nblocks, sizeof(int)*PROC_MAX_NUM));
CUDA_CHECKED_CALL(cudaMemcpyToSymbol(maxConcurrentBlockEvalDone,
nblocks, sizeof(int)*PROC_MAX_NUM));
}
template<class Q, MegakernelStopCriteria StopCriteria, bool Maintainer>
class MaintainerCaller;
template<class Q, MegakernelStopCriteria StopCriteria>
class MaintainerCaller<Q, StopCriteria, true>
{
public:
static __inline__ __device__ bool RunMaintainer(Q* q, int* shutdown)
{
if(blockIdx.x == 1)
{
__shared__ bool run;
run = true;
__syncthreads();
int runs = 0;
while(run)
{
q->globalMaintain();
__syncthreads();
if(runs > 10)
{
if(endCounter[0] == 0)
{
if(StopCriteria == MegakernelStopCriteria::EmptyQueue)
run = false;
else if (shutdown)
{
if(*shutdown)
run = false;
}
}
__syncthreads();
}
else
++runs;
}
}
return false;
}
static __inline__ __device__ bool RunMaintainerSpecific(Q* q, int* shutdown, int procId)
{
if(blockIdx.x == 1)
{
__shared__ bool run;
run = true;
__syncthreads();
int runs = 0;
while(run)
{
q->globalMaintain();
__syncthreads();
if(runs > 10)
{
if(endCounter[procId] == 0)
{
if(StopCriteria == MegakernelStopCriteria::EmptyQueue)
run = false;
else if (shutdown)
{
if(*shutdown)
run = false;
}
}
__syncthreads();
}
else
++runs;
}
}
return false;
}
};
template<class Q, MegakernelStopCriteria StopCriteria>
class MaintainerCaller<Q, StopCriteria, false>
{
public:
static __inline__ __device__ bool RunMaintainer(Q* q, int* shutdown)
{
return false;
}
static __inline__ __device__ bool RunMaintainerSpecific(Q* q, int* shutdown, int procId)
{
return false;
}
};
template<class Q, class PROCINFO, class CUSTOM, bool CopyToShared, bool MultiElement, bool tripleCall>
class MegakernelLogics;
template<class Q, class PROCINFO, class CUSTOM, bool MultiElement, bool tripleCall>
class MegakernelLogics<Q, PROCINFO, CUSTOM, true, MultiElement, tripleCall>
{
public:
// 执行PROCINFO中的Procedure的execute函数
static __device__ __inline__ int run(Q* q, uint4 sharedMemDist)
{
extern __shared__ uint s_data[];
// TODO: shared memory存储的数据
void* execData = reinterpret_cast<void*>(s_data + sharedMemDist.x + sharedMemDist.w);
int* execproc = reinterpret_cast<int*>(s_data + sharedMemDist.w);
int hasResult = q-> template dequeue<MultiElement> (execData, execproc,
sizeof(uint)*(sharedMemDist.y + sharedMemDist.z));
__syncthreads();
if(hasResult)
{
ProcCallCopyVisitor<Q, PROCINFO, MultiElement> visitor(q, execproc,
execData, s_data, sharedMemDist, hasResult);
ProcInfoVisitor<PROCINFO, CUSTOM>::template Visit<ProcCallCopyVisitor<Q, PROCINFO, MultiElement> >(visitor);
}
return hasResult;
}
// static __device__ __inline__ int runSpecific(Q* q, int *procIdArray, int procNum, uint4 sharedMemDist)
static __device__ __inline__ int runSpecific(Q* q, uint4 sharedMemDist)
{
extern __shared__ uint s_data[];
// TODO: shared memory存储的数据
void* execData = reinterpret_cast<void*>(s_data + sharedMemDist.x + sharedMemDist.w);
int* execproc = reinterpret_cast<int*>(s_data + sharedMemDist.w);
int hasResult = q-> template dequeue<MultiElement> (execData, *execproc,
sizeof(uint)*(sharedMemDist.y + sharedMemDist.z));
__syncthreads();
if(hasResult)
{
ProcCallCopyVisitor<Q, PROCINFO, MultiElement> visitor(q, execproc,
execData, s_data, sharedMemDist, hasResult);
ProcInfoVisitor<PROCINFO, CUSTOM>::
template Visit<ProcCallCopyVisitor<Q, PROCINFO, MultiElement> >(visitor);
}
return hasResult;
}
};
template<class Q, class PROCINFO, class CUSTOM, bool MultiElement>
class MegakernelLogics<Q, PROCINFO, CUSTOM, false, MultiElement, false>
{
public:
// ZHENG Zhen added, this one.
static __device__ __inline__ int run(Q* q, uint4 sharedMemDist)
{
extern __shared__ uint s_data[];
void* execData = reinterpret_cast<void*>(s_data + sharedMemDist.x + sharedMemDist.w);
int* execproc = reinterpret_cast<int*>(s_data + sharedMemDist.w);
// 10 registers usage for little demo
int hasResult = q-> template dequeueStartRead<MultiElement> (
execData, execproc, sizeof(uint)*sharedMemDist.z);
// test register usage
//int hasResult = ((int)execproc) % 9 == 0 ? 1 : 0;
__syncthreads();
// too many registers usage here
if(hasResult)
{
ProcCallNoCopyVisitor<Q, PROCINFO, MultiElement> visitor(q, execproc,
execData, s_data, sharedMemDist, hasResult);
ProcInfoVisitor<PROCINFO, CUSTOM>::template Visit<ProcCallNoCopyVisitor<Q, PROCINFO, MultiElement> >(
visitor);
hasResult = visitor.processCount;
}
return hasResult;
}
// static __device__ __inline__ int runSpecific(Q* q, int *procIdArray, int procNum, uint4 sharedMemDist)
static __device__ __inline__ int runSpecific(Q* q, uint4 sharedMemDist)
{
// ZHENG Zhen added, this one.
// printf("%s in %s, at line %d\n", __FUNCTION__, __FILE__, __LINE__);
//if(threadIdx.x == 0)
//{
// q是CurrentMultiphaseQueue
// q->printName();
// printf("address: %p\n", q);
// printf("QueueSize: %d, in %s %s %d\n",
// q->qs.q.getAttach().getAttachment().size(),
// __FUNCTION__, __FILE__, __LINE__);
//}
extern __shared__ uint s_data[];
void* execData = reinterpret_cast<void*>(s_data + sharedMemDist.x + sharedMemDist.w);
int* execproc = reinterpret_cast<int*>(s_data + sharedMemDist.w);
int hasResult = q-> template dequeueStartRead<MultiElement> (
execData, execproc, sizeof(uint)*sharedMemDist.z);
//printf("procId: %d, hasResult: %d\n",
// procIdArray[i], hasResult);
__syncthreads();
if(hasResult)
{
ProcCallNoCopyVisitor<Q, PROCINFO, MultiElement> visitor(q, execproc,
execData, s_data, sharedMemDist, hasResult);
ProcInfoVisitor<PROCINFO, CUSTOM>::template Visit<ProcCallNoCopyVisitor<Q, PROCINFO, MultiElement> >(
visitor);
}
return hasResult;
}
};
/*
template<class Q, class PROCINFO, class CUSTOM, bool MultiElement>
class MegakernelLogics<Q, PROCINFO, CUSTOM, false, MultiElement, true>
{
public:
static __device__ __inline__ int run(Q* q, uint4 sharedMemDist)
{
extern __shared__ uint s_data[];
void* execData = reinterpret_cast<void*>(s_data + sharedMemDist.x + sharedMemDist.w);
int* execproc = reinterpret_cast<int*>(s_data + sharedMemDist.w);
int hasResult = q-> template dequeueStartRead1<MultiElement> (execData, execproc, sizeof(uint)*sharedMemDist.z);
if(hasResult)
{
ProcCallNoCopyVisitorPart1<Q, PROCINFO, MultiElement> visitor(
q, execproc, execData, s_data, sharedMemDist, hasResult);
ProcInfoVisitor<PROCINFO, CUSTOM>
::template Visit<ProcCallNoCopyVisitorPart1<Q, PROCINFO, MultiElement> >(visitor);
return hasResult;
}
hasResult = q-> template dequeueStartRead2<MultiElement> (
execData, execproc, sizeof(uint)*sharedMemDist.z);
if(hasResult)
{
ProcCallNoCopyVisitorPart2<Q, PROCINFO, MultiElement> visitor(
q, execproc, execData, s_data, sharedMemDist, hasResult);
ProcInfoVisitor<PROCINFO, CUSTOM>
::template Visit<ProcCallNoCopyVisitorPart2<Q, PROCINFO, MultiElement> >(visitor);
return hasResult;
}
hasResult = q-> template dequeueStartRead3<MultiElement> (
execData, execproc, sizeof(uint)*sharedMemDist.z);
if(hasResult)
{
ProcCallNoCopyVisitorPart3<Q, PROCINFO, MultiElement> visitor(
q, execproc, execData, s_data, sharedMemDist, hasResult);
ProcInfoVisitor<PROCINFO, CUSTOM>
::template Visit<ProcCallNoCopyVisitorPart3<Q, PROCINFO, MultiElement> >(visitor);
}
return hasResult;
}
static __device__ __inline__ int runSpecific(Q* q, int procId, uint4 sharedMemDist)
{
extern __shared__ uint s_data[];
void* execData = reinterpret_cast<void*>(s_data + sharedMemDist.x + sharedMemDist.w);
int* execproc = reinterpret_cast<int*>(s_data + sharedMemDist.w);
*execproc = procId;
int hasResult = q-> template dequeueStartRead1Specific<MultiElement> (
execData, *execproc, sizeof(uint)*sharedMemDist.z);
if(hasResult)
{
ProcCallNoCopyVisitorPart1<Q, PROCINFO, MultiElement> visitor(
q, execproc, execData, s_data, sharedMemDist, hasResult);
ProcInfoVisitor<PROCINFO, CUSTOM>
::template Visit<ProcCallNoCopyVisitorPart1<Q, PROCINFO, MultiElement> >(visitor);
return hasResult;
}
hasResult = q-> template dequeueStartRead2Specific<MultiElement> (
execData, *execproc, sizeof(uint)*sharedMemDist.z);
if(hasResult)
{
ProcCallNoCopyVisitorPart2<Q, PROCINFO, MultiElement> visitor(
q, execproc, execData, s_data, sharedMemDist, hasResult);
ProcInfoVisitor<PROCINFO, CUSTOM>
::template Visit<ProcCallNoCopyVisitorPart2<Q, PROCINFO, MultiElement> >(visitor);
return hasResult;
}
hasResult = q-> template dequeueStartRead3Specific<MultiElement> (
execData, *execproc, sizeof(uint)*sharedMemDist.z);
if(hasResult)
{
ProcCallNoCopyVisitorPart3<Q, PROCINFO, MultiElement> visitor(
q, execproc, execData, s_data, sharedMemDist, hasResult);
ProcInfoVisitor<PROCINFO, CUSTOM>
::template Visit<ProcCallNoCopyVisitorPart3<Q, PROCINFO, MultiElement> >(visitor);
}
return hasResult;
}
};
*/
template<unsigned long long StaticLimit, bool Dynamic>
struct TimeLimiter;
template<>
struct TimeLimiter<0, false>
{
__device__ __inline__ TimeLimiter() { }
__device__ __inline__ bool stop(int tval)
{
return false;
}
};
template<unsigned long long StaticLimit>
struct TimeLimiter<StaticLimit, false>
{
unsigned long long TimeLimiter_start;
__device__ __inline__ TimeLimiter()
{
// kernel内计时
if(threadIdx.x == 0)
TimeLimiter_start = clock64();
}
__device__ __inline__ bool stop(int tval)
{
return (clock64() - TimeLimiter_start) > StaticLimit;
}
};
template<>
struct TimeLimiter<0, true>
{
unsigned long long TimeLimiter_start;
__device__ __inline__ TimeLimiter()
{
if(threadIdx.x == 0)
TimeLimiter_start = clock64();
}
__device__ __inline__ bool stop(int tval)
{
return (clock64() - TimeLimiter_start)/1024 > tval;
}
};
template<class Q, class PROCINFO, class CUSTOM, bool CopyToShared, bool MultiElement,
bool Maintainer, class TimeLimiter, MegakernelStopCriteria StopCriteria>
__global__ void megakernel(Q* q, uint4 sharedMemDist, int t, int* shutdown)
{
if(q == 0)
{
if(maxConcurrentBlockEvalDone[0] != 0)
return;
if(threadIdx.x == 0)
{
atomicAdd(&maxConcurrentBlocks[0], 1);
// doneCounter[0] = 0;
// endCounter[0] = 2597;
}
/*
if(threadIdx.x == 0)
{
printf("Block: %d\n", blockIdx.x);
}
*/
DelayFMADS<10000,4>::delay();
__syncthreads();
maxConcurrentBlockEvalDone[0] = 1;
__threadfence();
return;
}
__shared__ volatile int runState;
#ifdef MAX_CLOCK_MILISEC
__shared__ long long clockzz;
#endif
if(MaintainerCaller<Q, StopCriteria, Maintainer>::RunMaintainer(q, shutdown))
return;
__shared__ TimeLimiter timelimiter;
if(threadIdx.x == 0)
{
#ifdef MAX_CLOCK_MILISEC
clockzz = clock64();
#endif
if(endCounter[0] == 0)
runState = 0;
else
{
atomicAdd((int*)&doneCounter[0],1);
if(atomicAdd((int*)&endCounter[0],1) == 2597)
atomicSub((int*)&endCounter[0], 2597);
runState = 1;
}
}
q->workerStart();
__syncthreads();
while(runState)
{
// 从queue中取出数据，执行对应Procedure的execute函数
int hasResult = MegakernelLogics<Q, PROCINFO, CUSTOM, CopyToShared,
MultiElement, Q::needTripleCall>::run(q, sharedMemDist);
/*
if(!hasResult && q->isEmpty())
{
runState = 0;
}
*/
if(threadIdx.x == 0)
{
#ifdef MAX_CLOCK_MILISEC
if((clock64() - clockzz) / 705500 > MAX_CLOCK_MILISEC)
{
runState = 0;
}
#endif
if(timelimiter.stop(t))
{
runState = 0;
}
else if(hasResult) // queue中仍然有数据
{
if(runState == 3)
{
//back on working
runState = 1;
atomicAdd((int*)&doneCounter[0],1);
atomicAdd((int*)&endCounter[0],1);
}
else if(runState == 2)
{
//back on working
runState = 1;
atomicAdd((int*)&doneCounter[0],1);
}
}
else // 这次没有从Queue中取出数据
{
//RUNSTATE UPDATES
if(runState == 1)
{
//first time we are out of work
atomicSub((int*)&doneCounter[0],1);
runState = 2;
}
else if(runState == 2)
{
if(doneCounter[0] == 0)
{
//everyone seems to be out of work -> get ready for end
atomicSub((int*)&endCounter[0],1);
runState = 3;
}
}
else if(runState == 3)
{
//int d = doneCounter[-1];
//int e = endCounter[0];
//printf("%d %d %d\n",blockIdx.x, d, e);
if(doneCounter[0] != 0)
{
//someone started to work again
atomicAdd((int*)&endCounter[0],1);
runState = 2;
}
else if(endCounter[0] == 0)
{
//everyone is really out of work
if(StopCriteria == MegakernelStopCriteria::EmptyQueue)
runState = 0;
else if (shutdown)
{
if(*shutdown)
runState = 0;
}
}
}
}
}
__syncthreads();
q->workerMaintain();
}
q->workerEnd();
}
template<class Q, class PROCINFO, class CUSTOM, bool CopyToShared, bool MultiElement,
bool Maintainer, class TimeLimiter, MegakernelStopCriteria StopCriteria>
__global__ void megakernelSpecific(int groupId, int taskCount, Q* q, uint4 sharedMemDist, int t, int* shutdown)
{
if(q == 0)
{
if(maxConcurrentBlockEvalDone[groupId] != 0)
return;
if(threadIdx.x == 0)
{
atomicAdd(&maxConcurrentBlocks[groupId], 1);
}
DelayFMADS<10000,4>::delay();
__syncthreads();
maxConcurrentBlockEvalDone[groupId] = 1;
__threadfence();
return;
}
__shared__ volatile int runState;
__shared__ TimeLimiter timelimiter;
__shared__ int blockReturn;
#ifdef MAX_CLOCK_MILISEC
__shared__ long long clockzz;
#endif
int smid = zz_smid();
if(sm_flag[groupId * SM_MAX_NUM + smid] == 0) // || // procedures in one megakernel can resist on same SM
// MaintainerCaller<Q, StopCriteria, Maintainer>::RunMaintainer(q, shutdown))
{
return;
}
blockReturn = 0;
__syncthreads();
if(threadIdx.x == 0)
{
int blockLeft = atomicSub(&(block_count[groupId * SM_MAX_NUM + smid]), 1);
if(blockLeft <= 0)
{
blockReturn = 1;
}
/*
atomicAdd((int*)&doneCounter[groupId],1);
if(atomicAdd((int*)&endCounter[groupId],1) == 2597)
{
atomicSub((int*)&endCounter[groupId], 2597);
}
*/
runState = 1;
#ifdef MAX_CLOCK_MILISEC
clockzz = clock64();
#endif
}
__syncthreads();
if(blockReturn == 1)
{
return;
}
while(runState)
{
// 从queue中取出数据，执行对应Procedure的execute函数
int hasResult = MegakernelLogics<Q, PROCINFO, CUSTOM, CopyToShared,
MultiElement, Q::needTripleCall>::run(q, sharedMemDist);
if(threadIdx.x == 0)
{
#ifdef MAX_CLOCK_MILISEC
if((clock64() - clockzz) / 705500 > MAX_CLOCK_MILISEC)
{
runState = 0;
break;
}
#endif
// 判断是否为stream pipeline
//#ifndef STREAM_STYLE_PIPELINE
if(taskCount > 0)
{
int a = hasResult ? atomicAdd(&resultCounter[groupId], hasResult) + hasResult : resultCounter[groupId];
if(a >= taskCount)
//if(a >= taskCount-1)
{
runState = 0;
}
}
else if(hasResult) // queue中仍然有数据
{
if(runState == 3)
{
//back on working
runState = 1;
atomicAdd((int*)&doneCounter[0],1);
atomicAdd((int*)&endCounter[0],1);
}
else if(runState == 2)
{
//back on working
runState = 1;
atomicAdd((int*)&doneCounter[0],1);
}
}
else // if(taskCount <= 0) // 这次没有从Queue中取出数据
// && group_done_flag[groupId] == 1)
{
//RUNSTATE UPDATES
if(runState == 1)
{
//first time we are out of work
atomicSub((int*)&doneCounter[0],1);
runState = 2;
}
else if(runState == 2)
{
if(doneCounter[0] <= 0)
{
//everyone seems to be out of work -> get ready for end
atomicSub((int*)&endCounter[0],1);
runState = 3;
}
}
else if(runState == 3)
{
//int d = doneCounter[-1];
//int e = endCounter[0];
//printf("%d %d %d\n",blockIdx.x, d, e);
if(doneCounter[0] > 0)
{
//someone started to work again
atomicAdd((int*)&endCounter[0],1);
runState = 2;
}
else
{
atomicSub((int*)&endCounter[0], 1);
}