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
|
/* StarPU --- Runtime system for heterogeneous multicore architectures.
*
* Copyright (C) 2011, 2012 Centre National de la Recherche Scientifique
* Copyright (C) 2011-2012 Université de Bordeaux 1
*
* StarPU is free software; you can redistribute it and/or modify
* it under the terms of the GNU Lesser General Public License as published by
* the Free Software Foundation; either version 2.1 of the License, or (at
* your option) any later version.
*
* StarPU 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 Lesser General Public License in COPYING.LGPL for more details.
*/
#include <stdarg.h>
#include <mpi.h>
#include <starpu.h>
#include <starpu_data.h>
#include <common/utils.h>
#include <starpu_hash.h>
#include <common/htable32.h>
#include <util/starpu_insert_task_utils.h>
#include <datawizard/coherency.h>
//#define STARPU_MPI_VERBOSE 1
#include <starpu_mpi_private.h>
/* Whether we are allowed to keep copies of remote data. Does not work
* yet: the sender has to know whether the receiver has it, keeping it
* in an array indexed by node numbers. */
//#define MPI_CACHE
#include <starpu_mpi_insert_task_cache.h>
static void _starpu_mpi_tables_init()
{
if (sent_data == NULL) {
int nb_nodes;
int i;
MPI_Comm_size(MPI_COMM_WORLD, &nb_nodes);
_STARPU_MPI_DEBUG("Initialising hash table for cache\n");
sent_data = malloc(nb_nodes * sizeof(struct starpu_htbl32_node *));
for(i=0 ; i<nb_nodes ; i++) sent_data[i] = NULL;
received_data = malloc(nb_nodes * sizeof(struct starpu_htbl32_node *));
for(i=0 ; i<nb_nodes ; i++) received_data[i] = NULL;
}
}
void _starpu_data_deallocate(starpu_data_handle_t data_handle)
{
#ifdef STARPU_DEVEL
#warning _starpu_data_deallocate not implemented yet
#endif
}
int starpu_mpi_insert_task(MPI_Comm comm, struct starpu_codelet *codelet, ...)
{
int arg_type;
va_list varg_list;
int me, do_execute, xrank, nb_nodes;
size_t *size_on_nodes;
size_t arg_buffer_size = 0;
char *arg_buffer;
int dest=0, inconsistent_execute;
_STARPU_MPI_LOG_IN();
MPI_Comm_rank(comm, &me);
MPI_Comm_size(comm, &nb_nodes);
size_on_nodes = (size_t *)calloc(1, nb_nodes * sizeof(size_t));
_starpu_mpi_tables_init();
/* Get the number of buffers and the size of the arguments */
va_start(varg_list, codelet);
arg_buffer_size = _starpu_insert_task_get_arg_size(varg_list);
va_start(varg_list, codelet);
_starpu_codelet_pack_args(arg_buffer_size, &arg_buffer, varg_list);
/* Find out whether we are to execute the data because we own the data to be written to. */
inconsistent_execute = 0;
do_execute = -1;
xrank = -1;
va_start(varg_list, codelet);
while ((arg_type = va_arg(varg_list, int)) != 0) {
if (arg_type==STARPU_EXECUTE_ON_NODE) {
xrank = va_arg(varg_list, int);
_STARPU_MPI_DEBUG("Executing on node %d\n", xrank);
do_execute = 1;
}
else if (arg_type==STARPU_EXECUTE_ON_DATA) {
starpu_data_handle_t data = va_arg(varg_list, starpu_data_handle_t);
xrank = starpu_data_get_rank(data);
_STARPU_MPI_DEBUG("Executing on data node %d\n", xrank);
STARPU_ASSERT(xrank <= nb_nodes);
do_execute = 1;
}
else if (arg_type==STARPU_R || arg_type==STARPU_W || arg_type==STARPU_RW || arg_type==STARPU_SCRATCH || arg_type==STARPU_REDUX) {
starpu_data_handle_t data = va_arg(varg_list, starpu_data_handle_t);
if (data && arg_type & STARPU_R) {
int rank = starpu_data_get_rank(data);
struct starpu_data_interface_ops *ops;
ops = data->ops;
size_on_nodes[rank] += ops->get_size(data);
}
if (arg_type & STARPU_W) {
if (!data) {
/* We don't have anything allocated for this.
* The application knows we won't do anything
* about this task */
/* Yes, the app could actually not call
* insert_task at all itself, this is just a
* safeguard. */
_STARPU_MPI_DEBUG("oh oh\n");
_STARPU_MPI_LOG_OUT();
free(size_on_nodes);
return -EINVAL;
}
int mpi_rank = starpu_data_get_rank(data);
if (mpi_rank == me) {
if (do_execute == 0) {
inconsistent_execute = 1;
}
else {
do_execute = 1;
}
}
else if (mpi_rank != -1) {
if (do_execute == 1) {
inconsistent_execute = 1;
}
else {
do_execute = 0;
dest = mpi_rank;
/* That's the rank which needs the data to be sent to */
}
}
else {
_STARPU_ERROR("rank invalid\n");
}
}
}
else if (arg_type==STARPU_VALUE) {
va_arg(varg_list, void *);
va_arg(varg_list, size_t);
}
else if (arg_type==STARPU_CALLBACK) {
va_arg(varg_list, void (*)(void *));
}
else if (arg_type==STARPU_CALLBACK_WITH_ARG) {
va_arg(varg_list, void (*)(void *));
va_arg(varg_list, void *);
}
else if (arg_type==STARPU_CALLBACK_ARG) {
va_arg(varg_list, void *);
}
else if (arg_type==STARPU_PRIORITY) {
va_arg(varg_list, int);
}
else if (arg_type==STARPU_EXECUTE_ON_NODE) {
va_arg(varg_list, int);
}
else if (arg_type==STARPU_EXECUTE_ON_DATA) {
va_arg(varg_list, int);
}
}
va_end(varg_list);
if (do_execute == -1) {
int i;
size_t max_size = 0;
for(i=0 ; i<nb_nodes ; i++) {
if (size_on_nodes[i] > max_size)
{
max_size = size_on_nodes[i];
xrank = i;
}
}
free(size_on_nodes);
if (xrank != -1) {
_STARPU_MPI_DEBUG("Node %d is having the most R data\n", xrank);
do_execute = 1;
}
}
STARPU_ASSERT(do_execute != -1 && "StarPU needs to see a W or a REDUX data which will tell it where to execute the task");
if (inconsistent_execute == 1) {
if (xrank == -1) {
_STARPU_MPI_DEBUG("Different tasks are owning W data. Needs to specify which one is to execute the codelet, using STARPU_EXECUTE_ON_NODE or STARPU_EXECUTE_ON_DATA\n");
free(size_on_nodes);
return -EINVAL;
}
else {
do_execute = (me == xrank);
dest = xrank;
}
}
else if (xrank != -1) {
do_execute = (me == xrank);
dest = xrank;
}
/* Send and receive data as requested */
va_start(varg_list, codelet);
while ((arg_type = va_arg(varg_list, int)) != 0) {
if (arg_type==STARPU_R || arg_type==STARPU_W || arg_type==STARPU_RW || arg_type == STARPU_SCRATCH) {
starpu_data_handle_t data = va_arg(varg_list, starpu_data_handle_t);
if (data && arg_type & STARPU_R) {
int mpi_rank = starpu_data_get_rank(data);
int mpi_tag = starpu_data_get_tag(data);
STARPU_ASSERT(mpi_tag >= 0 && "StarPU needs to be told the MPI rank of this data, using starpu_data_set_rank");
/* The task needs to read this data */
if (do_execute && mpi_rank != me && mpi_rank != -1) {
/* I will have to execute but I don't have the data, receive */
#ifdef MPI_CACHE
uint32_t key = starpu_crc32_be((uintptr_t)data, 0);
void *already_received = _starpu_htbl_search_32(received_data[mpi_rank], key);
if (!already_received) {
_starpu_htbl_insert_32(&received_data[mpi_rank], key, data);
}
else {
_STARPU_MPI_DEBUG("Do not receive data %p from node %d as it is already available\n", data, mpi_rank);
}
if (!already_received)
#endif
{
_STARPU_MPI_DEBUG("Receive data %p from %d\n", data, mpi_rank);
starpu_mpi_irecv_detached(data, mpi_rank, mpi_tag, comm, NULL, NULL);
}
}
if (!do_execute && mpi_rank == me) {
/* Somebody else will execute it, and I have the data, send it. */
#ifdef MPI_CACHE
uint32_t key = starpu_crc32_be((uintptr_t)data, 0);
void *already_sent = _starpu_htbl_search_32(sent_data[dest], key);
if (!already_sent) {
_starpu_htbl_insert_32(&sent_data[dest], key, data);
}
else {
_STARPU_MPI_DEBUG("Do not sent data %p to node %d as it has already been sent\n", data, dest);
}
if (!already_sent)
#endif
{
_STARPU_MPI_DEBUG("Send data %p to %d\n", data, dest);
starpu_mpi_isend_detached(data, dest, mpi_tag, comm, NULL, NULL);
}
}
}
}
else if (arg_type==STARPU_VALUE) {
va_arg(varg_list, void *);
va_arg(varg_list, size_t);
}
else if (arg_type==STARPU_CALLBACK) {
va_arg(varg_list, void (*)(void *));
}
else if (arg_type==STARPU_CALLBACK_WITH_ARG) {
va_arg(varg_list, void (*)(void *));
va_arg(varg_list, void *);
}
else if (arg_type==STARPU_CALLBACK_ARG) {
va_arg(varg_list, void *);
}
else if (arg_type==STARPU_PRIORITY) {
va_arg(varg_list, int);
}
else if (arg_type==STARPU_EXECUTE_ON_NODE) {
va_arg(varg_list, int);
}
else if (arg_type==STARPU_EXECUTE_ON_DATA) {
va_arg(varg_list, starpu_data_handle_t);
}
}
va_end(varg_list);
if (do_execute) {
_STARPU_MPI_DEBUG("Execution of the codelet %p (%s)\n", codelet, codelet->name);
va_start(varg_list, codelet);
struct starpu_task *task = starpu_task_create();
int ret = _starpu_insert_task_create_and_submit(arg_buffer, codelet, &task, varg_list);
_STARPU_MPI_DEBUG("ret: %d\n", ret);
STARPU_ASSERT(ret==0);
}
if (inconsistent_execute) {
va_start(varg_list, codelet);
while ((arg_type = va_arg(varg_list, int)) != 0) {
if (arg_type==STARPU_R || arg_type==STARPU_W || arg_type==STARPU_RW || arg_type == STARPU_SCRATCH) {
starpu_data_handle_t data = va_arg(varg_list, starpu_data_handle_t);
if (arg_type & STARPU_W) {
int mpi_rank = starpu_data_get_rank(data);
int mpi_tag = starpu_data_get_tag(data);
STARPU_ASSERT(mpi_tag >= 0 && "StarPU needs to be told the MPI rank of this data, using starpu_data_set_rank");
if (mpi_rank == me) {
if (xrank != -1 && me != xrank) {
_STARPU_MPI_DEBUG("Receive data %p back from the task %d which executed the codelet ...\n", data, dest);
starpu_mpi_irecv_detached(data, dest, mpi_tag, comm, NULL, NULL);
}
}
else if (do_execute) {
_STARPU_MPI_DEBUG("Send data %p back to its owner %d...\n", data, mpi_rank);
starpu_mpi_isend_detached(data, mpi_rank, mpi_tag, comm, NULL, NULL);
}
}
}
else if (arg_type==STARPU_VALUE) {
va_arg(varg_list, void *);
va_arg(varg_list, size_t);
}
else if (arg_type==STARPU_CALLBACK) {
va_arg(varg_list, void (*)(void *));
}
else if (arg_type==STARPU_CALLBACK_WITH_ARG) {
va_arg(varg_list, void (*)(void *));
va_arg(varg_list, void *);
}
else if (arg_type==STARPU_CALLBACK_ARG) {
va_arg(varg_list, void *);
}
else if (arg_type==STARPU_PRIORITY) {
va_arg(varg_list, int);
}
else if (arg_type==STARPU_EXECUTE_ON_NODE) {
va_arg(varg_list, int);
}
else if (arg_type==STARPU_EXECUTE_ON_DATA) {
va_arg(varg_list, starpu_data_handle_t);
}
}
va_end(varg_list);
}
va_start(varg_list, codelet);
while ((arg_type = va_arg(varg_list, int)) != 0) {
if (arg_type==STARPU_R || arg_type==STARPU_W || arg_type==STARPU_RW || arg_type == STARPU_SCRATCH) {
starpu_data_handle_t data = va_arg(varg_list, starpu_data_handle_t);
#ifdef MPI_CACHE
if (arg_type & STARPU_W) {
uint32_t key = starpu_crc32_be((uintptr_t)data, 0);
if (do_execute) {
/* Note that all copies I've sent to neighbours are now invalid */
int n, size;
MPI_Comm_size(comm, &size);
for(n=0 ; n<size ; n++) {
void *already_sent = _starpu_htbl_search_32(sent_data[n], key);
if (already_sent) {
_STARPU_MPI_DEBUG("Posting request to clear send cache for data %p\n", data);
_starpu_mpi_clear_cache_request(data, n, _STARPU_MPI_CLEAR_SENT_DATA);
}
}
}
else {
int mpi_rank = starpu_data_get_rank(data);
void *already_received = _starpu_htbl_search_32(received_data[mpi_rank], key);
if (already_received) {
/* Somebody else will write to the data, so discard our cached copy if any */
/* TODO: starpu_mpi could just remember itself. */
_STARPU_MPI_DEBUG("Posting request to clear receive cache for data %p\n", data);
_starpu_mpi_clear_cache_request(data, mpi_rank, _STARPU_MPI_CLEAR_RECEIVED_DATA);
_starpu_data_deallocate(data);
}
}
}
#else
/* We allocated a temporary buffer for the received data, now drop it */
if ((arg_type & STARPU_R) && do_execute) {
int mpi_rank = starpu_data_get_rank(data);
if (mpi_rank != me && mpi_rank != -1) {
_starpu_data_deallocate(data);
}
}
#endif
}
else if (arg_type==STARPU_VALUE) {
va_arg(varg_list, void *);
va_arg(varg_list, size_t);
}
else if (arg_type==STARPU_CALLBACK) {
va_arg(varg_list, void (*)(void *));
}
else if (arg_type==STARPU_CALLBACK_WITH_ARG) {
va_arg(varg_list, void (*)(void *));
va_arg(varg_list, void *);
}
else if (arg_type==STARPU_CALLBACK_ARG) {
va_arg(varg_list, void *);
}
else if (arg_type==STARPU_PRIORITY) {
va_arg(varg_list, int);
}
else if (arg_type==STARPU_EXECUTE_ON_NODE) {
va_arg(varg_list, int);
}
else if (arg_type==STARPU_EXECUTE_ON_DATA) {
va_arg(varg_list, starpu_data_handle_t);
}
}
va_end(varg_list);
_STARPU_MPI_LOG_OUT();
return 0;
}
void starpu_mpi_get_data_on_node_detached(MPI_Comm comm, starpu_data_handle_t data_handle, int node, void (*callback)(void*), void *arg)
{
int me, rank, tag;
rank = starpu_data_get_rank(data_handle);
tag = starpu_data_get_tag(data_handle);
MPI_Comm_rank(comm, &me);
if (node == rank) return;
if (me == node)
{
starpu_mpi_irecv_detached(data_handle, rank, tag, comm, callback, arg);
}
else if (me == rank)
{
starpu_mpi_isend_detached(data_handle, node, tag, comm, NULL, NULL);
}
}
void starpu_mpi_get_data_on_node(MPI_Comm comm, starpu_data_handle_t data_handle, int node)
{
int me, rank, tag;
rank = starpu_data_get_rank(data_handle);
tag = starpu_data_get_tag(data_handle);
MPI_Comm_rank(comm, &me);
if (node == rank) return;
if (me == node)
{
MPI_Status status;
starpu_mpi_recv(data_handle, rank, tag, comm, &status);
}
else if (me == rank)
{
starpu_mpi_send(data_handle, node, tag, comm);
}
}
void starpu_mpi_redux_data(MPI_Comm comm, starpu_data_handle_t data_handle)
{
int me, rank, tag, nb_nodes;
rank = starpu_data_get_rank(data_handle);
tag = starpu_data_get_tag(data_handle);
MPI_Comm_rank(comm, &me);
MPI_Comm_size(comm, &nb_nodes);
_STARPU_MPI_DEBUG("Doing reduction for data %p on node %d with %d nodes ...\n", data_handle, rank, nb_nodes);
// need to count how many nodes have the data in redux mode
if (me == rank) {
int i;
for(i=0 ; i<nb_nodes ; i++) {
if (i != rank) {
starpu_data_handle_t new_handle;
starpu_data_register_same(&new_handle, data_handle);
_STARPU_MPI_DEBUG("Receiving redux handle from %d in %p ...\n", i, new_handle);
starpu_mpi_irecv_detached(new_handle, i, tag, comm, NULL, NULL);
starpu_insert_task(data_handle->redux_cl,
STARPU_RW, data_handle,
STARPU_R, new_handle,
0);
}
}
}
else {
_STARPU_MPI_DEBUG("Sending redux handle to %d ...\n", rank);
starpu_mpi_isend_detached(data_handle, rank, tag, comm, NULL, NULL);
}
}
|
