Updating documentation for sampling, generating meta files for sampled data

src/network-workloads/README_synthetic.txt

 ************ Synthetic traffic with dragonfly network model **********
 - traffic patterns supported: uniform random, nearest group and nearest neighbor traffic.
-	- Uniform random traffic: sends messages to a randomly selected destination
+	- 1--> Uniform random traffic: sends messages to a randomly selected destination
       node. This traffic pattern is uniformly distributed throughout the
       network and gives a better performance with minimal routing as compared
       to non-minimal or adaptive routing.
-    - Nearest group traffic: with minimal routing, it sends traffic to the
+    - 2--> Nearest group traffic: with minimal routing, it sends traffic to the
       single global channel connecting two groups (it congests the network when
       using minimal routing).  This pattern performs better with non-minimal
       and adaptive routing algorithms.
-    - Nearest neighbor traffic: it sends traffic to the next node, potentially
+    - 3--> Nearest neighbor traffic: it sends traffic to the next node, potentially
       connected to the same router. 
 
 SAMPLING:
     - The modelnet_enable_sampling function takes a sampling interval "t" and
-      an end time. Over this end time, dragonfly model will collect compute
+      an end time in nanosecs. Over this end time, dragonfly model will collect compute
       node and router samples after every "t" simulated nanoseconds. The
-      sampling output files can be specified in the config file using
+      names of the sampling output files can be specified in the config file using
       cn_sample_file and rt_sample_file arguments. By default the compute node
       and router outputs will be sent to dragonfly-cn-sampling-%d.bin and
-      dragonfly-router-sampling-%d.bin. Corresponding metadata files for also
+      dragonfly-router-sampling-%d.bin. Corresponding metadata files are also
       generated that gives information on the file format, dragonfly
       configuration being used, router radix etc. 
       
@@ -26,20 +26,22 @@ SAMPLING:
       text format can be found at
       src/networks/model-net/read-dragonfly-sample.c (Note that the router
       radix aka RADIX needs to be tuned with the dragonfly configuration in the
-      utility to enable continguous array allocation).
+      utility to enable continguous array allocation. By default the radix is
+      set to 16 corresponding to a 1,056 node dragonfly network). The utility
+      can be built using mpicc and it expects the generated binary files to be
+      in the same directory when doing the translation from binary into text.
 
 HOW TO RUN:
 
 ROSS optimistic mode:
-mpirun -np 4 ./src/models/network-workloads/model-net-synthetic --sync=3
---traffic=3 --lp-io-dir=mn_synthetic --lp-io-use-suffix=1  --arrival_time=100.0
--- ../src/models/network-workloads/conf/modelnet-synthetic-dragonfly.conf
+mpirun -np 4 ./bin/model-net-synthetic --sync=3 --traffic=1
+--lp-io-dir=mn_synthetic --lp-io-use-suffix=1  --arrival_time=1000.0 --
+../src/network-workloads/conf/modelnet-synthetic-dragonfly.conf 
 
 ROSS serial mode:
-
-./src/models/network-workloads/model-net-synthetic --sync=1 --traffic=3
---lp-io-dir=mn_synthetic --lp-io-use-suffix=1  --arrival_time=100.0 --
-../src/models/network-workloads/conf/modelnet-synthetic-dragonfly.conf
+./bin/model-net-synthetic --sync=1 --traffic=1
+--lp-io-dir=mn_synthetic --lp-io-use-suffix=1  --arrival_time=1000.0 --
+../src/network-workloads/conf/modelnet-synthetic-dragonfly.conf 
 
 options:
 
@@ -53,11 +55,9 @@ num_msgs: number of messages generated per terminal. Each message has a size of
 traffic: 1 for uniform random traffic, 2 for nearest group traffic and 3 for
 nearest neighbor traffic.
 
-sampling-interval: if time-stepped series sampling is turned on, this parameter
-can be used to configure the sampling interval.
+sampling-interval: this parameter can be used to configure the sampling interval.
 
-sampling-end-time: if time-stepped series sampling is turned on, this parameter
-can be used to configure end time.
+sampling-end-time: this parameter can be used to configure end time.
 
 lp-io-dir: generates network traffic information on dragonfly terminals and
 routers. Here is information on individual files:

src/network-workloads/model-net-mpi-replay.c

@@ -718,7 +718,6 @@ static void codes_exec_mpi_recv_rc(
         nw_message* m,
         tw_lp* lp)
 {
-	num_bytes_recvd -= m->rc.saved_num_bytes;
 	ns->recv_time = m->rc.saved_recv_time;
 	if(m->fwd.found_match >= 0)
 	  {
@@ -781,8 +780,6 @@ static void codes_exec_mpi_recv(
 	m->rc.saved_recv_time = s->recv_time;
     m->rc.saved_num_bytes = mpi_op->u.recv.num_bytes;
 
-	num_bytes_recvd += mpi_op->u.recv.num_bytes;
-
     mpi_msgs_queue * recv_op = (mpi_msgs_queue*) malloc(sizeof(mpi_msgs_queue));
     recv_op->req_init_time = tw_now(lp);
     recv_op->op_type = mpi_op->op_type;
@@ -990,6 +987,7 @@ static void update_arrival_queue_rc(nw_state* s,
 {
 	s->recv_time = m->rc.saved_recv_time;
     s->num_bytes_recvd -= m->fwd.num_bytes;
+    num_bytes_recvd -= m->fwd.num_bytes;
 
     codes_local_latency_reverse(lp);
 
@@ -1045,6 +1043,7 @@ static void update_arrival_queue(nw_state* s, tw_bf * bf, nw_message * m, tw_lp
     //    printf("\n Dest rank %d local rank %d ", m->fwd.dest_rank, s->local_rank);
 	m->rc.saved_recv_time = s->recv_time;
     s->num_bytes_recvd += m->fwd.num_bytes;
+    num_bytes_recvd += m->fwd.num_bytes;
 
     // send a callback to the sender to increment times
     // find the global id of the source
@@ -1270,7 +1269,7 @@ static void get_next_mpi_operation_rc(nw_state* s, tw_bf * bf, nw_message * m, t
 			if(m->op_type == CODES_WK_ISEND)
 				codes_issue_next_event_rc(lp);
 			s->num_sends--;
-            s->num_bytes_sent += m->rc.saved_num_bytes;
+            s->num_bytes_sent -= m->rc.saved_num_bytes;
 			num_bytes_sent -= m->rc.saved_num_bytes;
 		}
 		break;
@@ -1703,7 +1702,9 @@ int main( int argc, char** argv )
    assert(num_net_traces);
 
    if(!g_tw_mynode)
-	printf("\n Total bytes sent %llu recvd %llu \n max runtime %lf ns avg runtime %lf \n max comm time %lf avg comm time %lf \n max send time %lf avg send time %lf \n max recv time %lf avg recv time %lf \n max wait time %lf avg wait time %lf \n", total_bytes_sent, total_bytes_recvd,
+	printf("\n Total bytes sent %llu recvd %llu \n max runtime %lf ns avg runtime %lf \n max comm time %lf avg comm time %lf \n max send time %lf avg send time %lf \n max recv time %lf avg recv time %lf \n max wait time %lf avg wait time %lf \n", 
+            total_bytes_sent, 
+            total_bytes_recvd,
 			max_run_time, avg_run_time/num_net_traces,
 			max_comm_run_time, avg_comm_run_time/num_net_traces,
 			total_max_send_time, total_avg_send_time/num_net_traces,

src/networks/model-net/read-dragonfly-sample.c

@@ -75,7 +75,7 @@ int main( int argc, char** argv )
    long in_sz = st.st_size;
    event_array = malloc(in_sz);
 
-   sprintf(buffer_write, "dragonfly-write-log.%d", my_rank);
+   sprintf(buffer_write, "dragonfly-write-log-%d.dat", my_rank);
    writeFile = fopen(buffer_write, "w");
 
    if(pFile == NULL || writeFile == NULL)
@@ -83,13 +83,21 @@ int main( int argc, char** argv )
 	fputs("\n File error ", stderr);
 	return -1;
    }
+   if(my_rank == 0)
+   {
+        char meta_filename[128];
+        sprintf(meta_filename, "dragonfly-write-log.meta");
+
+        FILE * fp_meta = fopen(meta_filename, "w+");
+        fprintf(fp_meta, "Rank_ID num_finished_chunks data_size_finished(bytes) finished_hops time_spent(ns) busy_time(ns) num_fwd_events num_rev_events sample_end_time(ns)");
+        fclose(fp_meta);
+   }
    fseek(pFile, 0L, SEEK_SET);
    fread(event_array, sizeof(struct dfly_samples), in_sz / sizeof(struct dfly_samples), pFile);
-   fprintf(writeFile, " Rank_ID Finished_chunks Data_size Finished_hops Time_spent busy_time fwd_events rev_events sample_end_time");
    for(i = 0; i < in_sz / sizeof(struct dfly_samples); i++)
    {
     printf("\n Terminal id %ld ", event_array[i].terminal_id);
-    fprintf(writeFile, "\n %ld %ld %ld %lf %lf %lf %ld %ld %lf ", event_array[i].terminal_id,
+    fprintf(writeFile, "%ld %ld %ld %lf %lf %lf %ld %ld %lf \n", event_array[i].terminal_id,
                                                                event_array[i].fin_chunks_sample,
                                                                event_array[i].data_size_sample,
                                                                event_array[i].fin_hops_sample, 
@@ -117,7 +125,7 @@ int main( int argc, char** argv )
     r_event_array = malloc(in_sz_rt);
 
     int sample_size = sizeof(struct dfly_rtr_sample);
-    sprintf(buffer_rtr_write, "dragonfly-rtr-write-log.%d", my_rank);
+    sprintf(buffer_rtr_write, "dragonfly-rtr-write-%d.dat", my_rank);
     writeRouterFile = fopen(buffer_rtr_write, "w");
 
     if(writeRouterFile == NULL || pFile == NULL)
@@ -127,13 +135,23 @@ int main( int argc, char** argv )
         return -1;
     }
     fseek(pFile, 0L, SEEK_SET);
+   
+    if(my_rank == 0)
+    {
+        char rtr_meta_filename[128];
+        sprintf(rtr_meta_filename, "dragonfly-rtr-write-log.meta");
+
+        FILE * fp_rtr_meta = fopen(rtr_meta_filename, "w+");
+        fprintf(fp_rtr_meta, "%d entries for busy time and link traffic \n", RADIX);
+        fprintf(fp_rtr_meta, "Format: Router_ID Busy_time_per_channel(ns) Link_traffic_per_channel(ns) Sample_end_time(ns) fwd_events reverse_events");
+        fclose(fp_rtr_meta);
+   }
     fread(r_event_array, sample_size, in_sz_rt / sample_size, pFile); 
-    fprintf(writeRouterFile, "\n Router_ID Busy_time_per_channel Link_traffic_per_channel Sample_end_time fwd_events reverse_events");
     //printf("\n Sample size %d in_sz_rt %ld ", in_sz_rt / sample_size, in_sz_rt);
     for(i = 0; i < in_sz_rt / sample_size; i++)
     {
         //printf("\n %ld ", r_event_array[i].router_id);
-        fprintf(writeRouterFile, "\n %ld ", r_event_array[i].router_id);
+        fprintf(writeRouterFile, "%ld ", r_event_array[i].router_id);
         
         for(j = 0; j < RADIX; j++ )
         {
@@ -146,7 +164,7 @@ int main( int argc, char** argv )
             //printf("\n link traffic %ld ", r_event_array[i].link_traffic[j]);
             fprintf(writeRouterFile, " %ld ", r_event_array[i].link_traffic[j]);
         }
-        fprintf(writeRouterFile, " %lf \n", r_event_array[i].end_time);
+       fprintf(writeRouterFile, " %lf ", r_event_array[i].end_time);
         fprintf(writeRouterFile, " %ld ", r_event_array[i].fwd_events);
         fprintf(writeRouterFile, " %ld \n", r_event_array[i].rev_events);
     }
@@ -163,10 +181,19 @@ int main( int argc, char** argv )
 
     mpi_event_array = malloc(in_sz_mpi);
     int mpi_sample_sz = sizeof(struct mpi_workload_sample);
-    sprintf(buffer_rtr_write, "dragonfly-mpi-write-logs-%d", my_rank);
+    sprintf(buffer_rtr_write, "dragonfly-mpi-write-logs-%d.dat", my_rank);
     writeFile = fopen(buffer_rtr_write, "w+");
     assert(writeFile);
 
+    if(my_rank == 0)
+    {
+        char ops_meta_filename[128];
+        sprintf(ops_meta_filename, "dragonfly-mpi-write-logs.meta");
+
+        FILE * fp_ops_meta = fopen(ops_meta_filename, "w+");
+        fprintf(fp_ops_meta, "network_node_id app_id num_sends num_bytes_sent num_waits sample_end_time(ns)");
+        fclose(fp_ops_meta);
+   }
     fread(mpi_event_array, mpi_sample_sz, in_sz_mpi / mpi_sample_sz, pFile);
     for(i = 0; i < in_sz_mpi / mpi_sample_sz; i++)
     {