model-net-synthetic.c 11.6 KB
Newer Older
1 2 3 4 5 6 7 8
/*
 * Copyright (C) 2015 University of Chicago.
 * See COPYRIGHT notice in top-level directory.
 *
 */

/*
* The test program generates some synthetic traffic patterns for the model-net network models.
9
* currently it only support the dragonfly network model uniform random and nearest neighbor traffic patterns.
10 11 12 13 14 15 16 17
*/

#include "codes/model-net.h"
#include "codes/lp-io.h"
#include "codes/codes.h"
#include "codes/codes_mapping.h"
#include "codes/configuration.h"
#include "codes/lp-type-lookup.h"
18
#include "codes/net/dragonfly.h"
19

20
#define PAYLOAD_SZ 2048
21 22 23 24 25 26 27

static int net_id = 0;
static int traffic = 1;
static double arrival_time = 1000.0;

/* whether to pull instead of push */
static int num_servers_per_rep = 0;
28 29 30 31
static int num_routers_per_grp = 0;
static int num_nodes_per_grp = 0;
static int num_groups = 0;
static int num_nodes = 0;
32

33 34 35 36 37
static char lp_io_dir[256] = {'\0'};
static lp_io_handle io_handle;
static unsigned int lp_io_use_suffix = 0;
static int do_lp_io = 0;
static int num_msgs = 20;
38 39
static unsigned int sampling_interval = 800000;
static unsigned int sampling_end_time = 1600000;
40

41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59
typedef struct svr_msg svr_msg;
typedef struct svr_state svr_state;

/* global variables for codes mapping */
static char group_name[MAX_NAME_LENGTH];
static char lp_type_name[MAX_NAME_LENGTH];
static int group_index, lp_type_index, rep_id, offset;

/* type of events */
enum svr_event
{
    KICKOFF,	   /* kickoff event */
    REMOTE,        /* remote event */
    LOCAL      /* local event */
};

/* type of synthetic traffic */
enum TRAFFIC
{
60 61 62
	UNIFORM = 1, /* sends message to a randomly selected node */
	NEAREST_GROUP = 2, /* sends message to the node connected to the neighboring router */
	NEAREST_NEIGHBOR = 3 /* sends message to the next node (potentially connected to the same router) */
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
};

struct svr_state
{
    int msg_sent_count;   /* requests sent */
    int msg_recvd_count;  /* requests recvd */
    int local_recvd_count; /* number of local messages received */
    tw_stime start_ts;    /* time that we started sending requests */
    tw_stime end_ts;      /* time that we ended sending requests */
};

struct svr_msg
{
    enum svr_event svr_event_type;
    tw_lpid src;          /* source of this request or ack */
    int incremented_flag; /* helper for reverse computation */
};

static void svr_init(
    svr_state * ns,
    tw_lp * lp);
static void svr_event(
    svr_state * ns,
    tw_bf * b,
    svr_msg * m,
    tw_lp * lp);
static void svr_rev_event(
    svr_state * ns,
    tw_bf * b,
    svr_msg * m,
    tw_lp * lp);
static void svr_finalize(
    svr_state * ns,
    tw_lp * lp);

tw_lptype svr_lp = {
    (init_f) svr_init,
    (pre_run_f) NULL,
    (event_f) svr_event,
    (revent_f) svr_rev_event,
103
    (commit_f) NULL,
104 105 106 107 108
    (final_f)  svr_finalize,
    (map_f) codes_mapping,
    sizeof(svr_state),
};

109 110 111 112
/* setup for the ROSS event tracing
 * can have a different function for  rbev_trace_f and ev_trace_f
 * but right now it is set to the same function for both
 */
113
void svr_event_collect(svr_msg *m, tw_lp *lp, char *buffer, int *collect_flag)
114 115 116 117 118
{
    int type = (int) m->svr_event_type;
    memcpy(buffer, &type, sizeof(type));
}

119 120 121 122 123 124
void svr_model_stat_collect(svr_state *s, tw_lp *lp, char *buffer)
{
    return;
}

st_model_types svr_model_types[] = {
125
    {(rbev_trace_f) svr_event_collect,
126
     sizeof(int),
127
     (ev_trace_f) svr_event_collect,
128 129 130
     sizeof(int),
     (model_stat_f) svr_model_stat_collect,
     0},
131 132 133
    {0}
};

134
static const st_model_types  *svr_get_model_stat_types(void)
135
{
136
    return(&svr_model_types[0]);
137 138
}

139
void svr_register_model_types()
140
{
141
    st_model_type_register("server", svr_get_model_stat_types());
142 143
}

144 145 146
const tw_optdef app_opt [] =
{
        TWOPT_GROUP("Model net synthetic traffic " ),
147 148
    	TWOPT_UINT("traffic", traffic, "UNIFORM RANDOM=1, NEAREST NEIGHBOR=2 "),
    	TWOPT_UINT("num_messages", num_msgs, "Number of messages to be generated per terminal "),
149 150
    	TWOPT_UINT("sampling-interval", sampling_interval, "the sampling interval "),
    	TWOPT_UINT("sampling-end-time", sampling_end_time, "sampling end time "),
151 152 153
	    TWOPT_STIME("arrival_time", arrival_time, "INTER-ARRIVAL TIME"),
        TWOPT_CHAR("lp-io-dir", lp_io_dir, "Where to place io output (unspecified -> no output"),
        TWOPT_UINT("lp-io-use-suffix", lp_io_use_suffix, "Whether to append uniq suffix to lp-io directory (default 0)"),
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
        TWOPT_END()
};

const tw_lptype* svr_get_lp_type()
{
            return(&svr_lp);
}

static void svr_add_lp_type()
{
  lp_type_register("server", svr_get_lp_type());
}

static void issue_event(
    svr_state * ns,
    tw_lp * lp)
{
    tw_event *e;
    svr_msg *m;
    tw_stime kickoff_time;

    /* each server sends a dummy event to itself that will kick off the real
     * simulation
     */

    /* skew each kickoff event slightly to help avoid event ties later on */
180
    kickoff_time = 1.1 * g_tw_lookahead + tw_rand_exponential(lp->rng, arrival_time);
181 182 183 184 185 186 187 188 189 190 191

    e = tw_event_new(lp->gid, kickoff_time, lp);
    m = tw_event_data(e);
    m->svr_event_type = KICKOFF;
    tw_event_send(e);
}

static void svr_init(
    svr_state * ns,
    tw_lp * lp)
{
192 193
    ns->start_ts = 0.0;

194 195 196 197 198 199 200 201 202 203
    issue_event(ns, lp);
    return;
}

static void handle_kickoff_rev_event(
            svr_state * ns,
            tw_bf * b,
            svr_msg * m,
            tw_lp * lp)
{
204 205 206
    if(m->incremented_flag)
        return;

207 208
    if(b->c1)
        tw_rand_reverse_unif(lp->rng);
209

210
    model_net_event_rc(net_id, lp, PAYLOAD_SZ);
211
	ns->msg_sent_count--;
212
    tw_rand_reverse_unif(lp->rng);
213
}
214 215 216 217 218 219
static void handle_kickoff_event(
	    svr_state * ns,
	    tw_bf * b,
	    svr_msg * m,
	    tw_lp * lp)
{
220
    if(ns->msg_sent_count >= num_msgs)
221 222 223 224
    {
        m->incremented_flag = 1;
        return;
    }
225 226 227 228

    m->incremented_flag = 0;

    char anno[MAX_NAME_LENGTH];
229
    tw_lpid local_dest = -1, global_dest = -1;
230

231 232 233 234 235 236 237 238 239 240 241
    svr_msg * m_local = malloc(sizeof(svr_msg));
    svr_msg * m_remote = malloc(sizeof(svr_msg));

    m_local->svr_event_type = LOCAL;
    m_local->src = lp->gid;

    memcpy(m_remote, m_local, sizeof(svr_msg));
    m_remote->svr_event_type = REMOTE;

    assert(net_id == DRAGONFLY); /* only supported for dragonfly model right now. */
    ns->start_ts = tw_now(lp);
242 243
    codes_mapping_get_lp_info(lp->gid, group_name, &group_index, lp_type_name, &lp_type_index, anno, &rep_id, &offset);
    int local_id = codes_mapping_get_lp_relative_id(lp->gid, 0, 0);
244

245 246 247
   /* in case of uniform random traffic, send to a random destination. */
   if(traffic == UNIFORM)
   {
248
    b->c1 = 1;
249 250
   	local_dest = tw_rand_integer(lp->rng, 0, num_nodes - 1);
   }
251
   else if(traffic == NEAREST_GROUP)
252
   {
253 254
	local_dest = (local_id + num_nodes_per_grp) % num_nodes;
	//printf("\n LP %ld sending to %ld num nodes %d ", local_id, local_dest, num_nodes);
255
   }
256 257
   else if(traffic == NEAREST_NEIGHBOR)
   {
258
	local_dest =  (local_id + 1) % num_nodes;
259 260
//	 printf("\n LP %ld sending to %ld num nodes %d ", rep_id * 2 + offset, local_dest, num_nodes);
   }
261
   assert(local_dest < num_nodes);
262 263
//   codes_mapping_get_lp_id(group_name, lp_type_name, anno, 1, local_dest / num_servers_per_rep, local_dest % num_servers_per_rep, &global_dest);
   global_dest = codes_mapping_get_lpid_from_relative(local_dest, group_name, lp_type_name, NULL, 0);
264 265
   ns->msg_sent_count++;
   model_net_event(net_id, "test", global_dest, PAYLOAD_SZ, 0.0, sizeof(svr_msg), (const void*)m_remote, sizeof(svr_msg), (const void*)m_local, lp);
266

267 268 269 270 271
   issue_event(ns, lp);
   return;
}

static void handle_remote_rev_event(
272
            svr_state * ns,
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
            tw_bf * b,
            svr_msg * m,
            tw_lp * lp)
{
        ns->msg_recvd_count--;
}

static void handle_remote_event(
	    svr_state * ns,
	    tw_bf * b,
	    svr_msg * m,
	    tw_lp * lp)
{
	ns->msg_recvd_count++;
}

static void handle_local_rev_event(
                svr_state * ns,
                tw_bf * b,
                svr_msg * m,
                tw_lp * lp)
{
	ns->local_recvd_count--;
}

static void handle_local_event(
                svr_state * ns,
                tw_bf * b,
                svr_msg * m,
                tw_lp * lp)
{
    ns->local_recvd_count++;
}
/* convert ns to seconds */
static tw_stime ns_to_s(tw_stime ns)
{
    return(ns / (1000.0 * 1000.0 * 1000.0));
}

/* convert seconds to ns */
static tw_stime s_to_ns(tw_stime ns)
{
    return(ns * (1000.0 * 1000.0 * 1000.0));
}

static void svr_finalize(
    svr_state * ns,
    tw_lp * lp)
{
    ns->end_ts = tw_now(lp);

    printf("server %llu recvd %d bytes in %f seconds, %f MiB/s sent_count %d recvd_count %d local_count %d \n", (unsigned long long)lp->gid, PAYLOAD_SZ*ns->msg_recvd_count, ns_to_s(ns->end_ts-ns->start_ts),
        ((double)(PAYLOAD_SZ*ns->msg_sent_count)/(double)(1024*1024)/ns_to_s(ns->end_ts-ns->start_ts)), ns->msg_sent_count, ns->msg_recvd_count, ns->local_recvd_count);
    return;
}

static void svr_rev_event(
    svr_state * ns,
    tw_bf * b,
    svr_msg * m,
    tw_lp * lp)
{
    switch (m->svr_event_type)
    {
	case REMOTE:
		handle_remote_rev_event(ns, b, m, lp);
		break;
	case LOCAL:
		handle_local_rev_event(ns, b, m, lp);
		break;
	case KICKOFF:
		handle_kickoff_rev_event(ns, b, m, lp);
		break;
	default:
		assert(0);
		break;
    }
}

static void svr_event(
    svr_state * ns,
    tw_bf * b,
    svr_msg * m,
    tw_lp * lp)
{
   switch (m->svr_event_type)
    {
        case REMOTE:
            handle_remote_event(ns, b, m, lp);
            break;
        case LOCAL:
            handle_local_event(ns, b, m, lp);
            break;
	case KICKOFF:
	    handle_kickoff_event(ns, b, m, lp);
	    break;
        default:
            printf("\n Invalid message type %d ", m->svr_event_type);
            assert(0);
        break;
    }
}

int main(
    int argc,
    char **argv)
{
    int nprocs;
    int rank;
    int num_nets;
    int *net_ids;

    tw_opt_add(app_opt);
    tw_init(&argc, &argv);

    if(argc < 2)
    {
            printf("\n Usage: mpirun <args> --sync=2/3 mapping_file_name.conf (optional --nkp) ");
            MPI_Finalize();
            return 0;
    }

    MPI_Comm_rank(MPI_COMM_WORLD, &rank);
    MPI_Comm_size(MPI_COMM_WORLD, &nprocs);

    configuration_load(argv[2], MPI_COMM_WORLD, &config);

    model_net_register();
    svr_add_lp_type();

403 404
    if (g_st_ev_trace || g_st_model_stats)
        svr_register_model_types();
405

406 407 408
    codes_mapping_setup();

    net_ids = model_net_configure(&num_nets);
409
    //assert(num_nets==1);
410 411 412
    net_id = *net_ids;
    free(net_ids);

413 414
    /* 5 days of simulation time */
    g_tw_ts_end = s_to_ns(5 * 24 * 60 * 60);
415
    model_net_enable_sampling(sampling_interval, sampling_end_time);
416

417
    if(net_id != DRAGONFLY)
418
    {
419 420 421
	printf("\n The test works with dragonfly model configuration only! ");
        MPI_Finalize();
        return 0;
422
    }
423 424
    num_servers_per_rep = codes_mapping_get_lp_count("MODELNET_GRP", 1, "server",
            NULL, 1);
425
    configuration_get_value_int(&config, "PARAMS", "num_routers", NULL, &num_routers_per_grp);
426

427 428 429
    num_groups = (num_routers_per_grp * (num_routers_per_grp/2) + 1);
    num_nodes = num_groups * num_routers_per_grp * (num_routers_per_grp / 2);
    num_nodes_per_grp = num_routers_per_grp * (num_routers_per_grp / 2);
430

431 432
    assert(num_nodes);

433
    if(lp_io_dir[0])
434
    {
435 436 437 438
        do_lp_io = 1;
        int flags = lp_io_use_suffix ? LP_IO_UNIQ_SUFFIX : 0;
        int ret = lp_io_prepare(lp_io_dir, flags, &io_handle, MPI_COMM_WORLD);
        assert(ret == 0 || !"lp_io_prepare failure");
439 440
    }
    tw_run();
441 442 443
    if (do_lp_io){
        int ret = lp_io_flush(io_handle, MPI_COMM_WORLD);
        assert(ret == 0 || !"lp_io_flush failure");
444
    }
445
    model_net_report_stats(net_id);
446 447 448 449
    tw_end();
    return 0;
}

Jonathan Jenkins's avatar
Jonathan Jenkins committed
450 451 452 453 454 455 456 457
/*
 * Local variables:
 *  c-indent-level: 4
 *  c-basic-offset: 4
 * End:
 *
 * vim: ft=c ts=8 sts=4 sw=4 expandtab
 */