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
|
/*
* Copyright (C) by Argonne National Laboratory
* See COPYRIGHT in top-level directory
*/
#include "mpitest.h"
#include "mpi.h"
#include <stdio.h>
#include <stdlib.h>
#include "dtpools.h"
#include "mtest_dtp.h"
#include <assert.h>
/*
static char MTEST_Descrip[] = "Send flood test";
*/
int world_rank, world_size;
#define MAX_TOTAL_MSG_SIZE (32 * 1024 * 1024)
#define MAXMSG (4096)
static int pingping(int seed, int testsize, int sendcnt, int recvcnt,
const char *basic_type, mtest_mem_type_e sendmem, mtest_mem_type_e recvmem)
{
int errs = 0;
int err;
int rank, size, source, dest;
int minsize = 2, nmsg, maxmsg;
MPI_Aint sendcount, recvcount;
MPI_Comm comm;
MPI_Datatype sendtype, recvtype;
DTP_pool_s dtp;
struct mtest_obj send, recv;
static char test_desc[200];
snprintf(test_desc, 200,
"./pingping -seed=%d -testsize=%d -type=%s -sendcnt=%d -recvcnt=%d -sendmem=%s -recvmem=%s",
seed, testsize, basic_type, sendcnt, recvcnt, MTest_memtype_name(sendmem),
MTest_memtype_name(recvmem));
if (world_rank == 0) {
MTestPrintfMsg(1, " %s\n", test_desc);
}
err = DTP_pool_create(basic_type, sendcnt, seed, &dtp);
if (err != DTP_SUCCESS) {
fprintf(stderr, "Error while creating send pool (%s,%d)\n", basic_type, sendcnt);
fflush(stderr);
}
MTest_dtp_obj_start(&send, "send", dtp, sendmem, 0, false);
MTest_dtp_obj_start(&recv, "recv", dtp, recvmem, 0, false);
int nbytes;
MPI_Type_size(dtp.DTP_base_type, &nbytes);
nbytes *= sendcnt;
maxmsg = MAX_TOTAL_MSG_SIZE / nbytes;
if (maxmsg > MAXMSG)
maxmsg = MAXMSG;
/* The following illustrates the use of the routines to
* run through a selection of communicators and datatypes.
* Use subsets of these for tests that do not involve combinations
* of communicators, datatypes, and counts of datatypes */
while (MTestGetIntracommGeneral(&comm, minsize, 1)) {
if (comm == MPI_COMM_NULL) {
/* for NULL comms, make sure these processes create the
* same number of objects, so the target knows what
* datatype layout to check for */
errs += MTEST_CREATE_AND_FREE_DTP_OBJS(dtp, testsize);
errs += MTEST_CREATE_AND_FREE_DTP_OBJS(dtp, testsize);
continue;
}
/* Determine the sender and receiver */
MPI_Comm_rank(comm, &rank);
MPI_Comm_size(comm, &size);
source = 0;
dest = size - 1;
DTP_pool_update_count(dtp, rank == source ? sendcnt : recvcnt);
/* To improve reporting of problems about operations, we
* change the error handler to errors return */
MPI_Comm_set_errhandler(comm, MPI_ERRORS_RETURN);
for (int i = 0; i < testsize; i++) {
errs += MTest_dtp_create(&send, rank == source);
errs += MTest_dtp_create(&recv, rank == dest);
if (rank == source) {
MTest_dtp_init(&send, 0, 1, sendcnt);
sendcount = send.dtp_obj.DTP_type_count;
sendtype = send.dtp_obj.DTP_datatype;
for (nmsg = 1; nmsg < maxmsg; nmsg++) {
err = MPI_Send((const char *) send.buf + send.dtp_obj.DTP_buf_offset,
sendcount, sendtype, dest, 0, comm);
if (err) {
errs++;
if (errs < 10) {
MTestPrintError(err);
}
}
}
} else if (rank == dest) {
recvcount = recv.dtp_obj.DTP_type_count;
recvtype = recv.dtp_obj.DTP_datatype;
for (nmsg = 1; nmsg < maxmsg; nmsg++) {
MTest_dtp_init(&recv, -1, -1, recvcnt);
MPI_Status status;
err = MPI_Recv((char *) recv.buf + recv.dtp_obj.DTP_buf_offset,
recvcount, recvtype, source, 0, comm, &status);
if (err) {
errs++;
if (errs < 10) {
MTestPrintError(err);
}
}
/* only up to sendcnt should be updated */
errs += MTestCheckStatus(&status, dtp.DTP_base_type, sendcnt, source, 0,
errs < 10);
errs += MTest_dtp_check(&recv, 0, 1, sendcnt, &send, errs < 10);
}
}
MTest_dtp_destroy(&send);
MTest_dtp_destroy(&recv);
if (MTestGetStressLevel() == 0) {
/* NOTE: Without MPI_Barrier, recv side can easily accumulate large unexpected queue
* across multiple batches, especially in an async test. Currently, both libfabric and ucx
* netmod does not handle large message queue well, resulting in exponential slow-downs.
* Adding barrier let the current tests pass.
*/
MPI_Barrier(comm);
}
}
MTestFreeComm(&comm);
}
MTest_dtp_obj_finish(&send);
MTest_dtp_obj_finish(&recv);
DTP_pool_free(dtp);
return errs;
}
int main(int argc, char *argv[])
{
int errs = 0;
MTest_Init(&argc, &argv);
MPI_Comm_size(MPI_COMM_WORLD, &world_size);
MPI_Comm_rank(MPI_COMM_WORLD, &world_rank);
struct dtp_args dtp_args;
dtp_args_init(&dtp_args, MTEST_DTP_PT2PT, argc, argv);
while (dtp_args_get_next(&dtp_args)) {
errs += pingping(dtp_args.seed, dtp_args.testsize,
dtp_args.count, dtp_args.u.pt2pt.recvcnt,
dtp_args.basic_type, dtp_args.u.pt2pt.sendmem, dtp_args.u.pt2pt.recvmem);
}
dtp_args_finalize(&dtp_args);
MTest_Finalize(errs);
return MTestReturnValue(errs);
}
|
