Removing excessive qhash memory allocation from network models

src/network-workloads/conf/dragonfly-custom/modelnet-test-dragonfly-custom.conf

@@ -49,9 +49,9 @@ PARAMS
 # number of global channels per router 
    num_global_channels="10";
 # network config file for intra-group connections 
-   intra-group-connections="/Users/mmubarak/Documents/software_development/dragonfly-cray/codes-conflict/scripts/gen-cray-topo/intra-custom";
+   intra-group-connections="../src/network-workloads/conf/dragonfly-custom/intra-custom";
 # network config file for inter-group connections
-   inter-group-connections="/Users/mmubarak/Documents/software_development/dragonfly-cray/codes-conflict/scripts/gen-cray-topo/inter-custom";
+   inter-group-connections="../src/network-workloads/conf/dragonfly-custom/inter-custom";
 # routing protocol to be used 
    routing="prog-adaptive";
 }

src/network-workloads/conf/dragonfly-custom/modelnet-test-dragonfly-edison.conf

@@ -48,9 +48,9 @@ PARAMS
 # number of global channels per router 
    num_global_channels="10";
 # network config file for intra-group connections 
-   intra-group-connections="/Users/mmubarak/Documents/software_development/dragonfly-cray/codes-conflict/scripts/gen-cray-topo/intra-edison";
+   intra-group-connections="../src/network-workloads/conf/dragonfly-custom/intra-edison";
 # network config file for inter-group connections
-   inter-group-connections="/Users/mmubarak/Documents/software_development/dragonfly-cray/codes-conflict/scripts/gen-cray-topo/inter-edison";
+   inter-group-connections="../src/network-workloads/conf/dragonfly-custom/inter-edison";
 # routing protocol to be used 
    routing="minimal";
 }

src/network-workloads/conf/dragonfly-custom/modelnet-test-dragonfly-theta.conf

@@ -48,9 +48,9 @@ PARAMS
 # number of global channels per router 
    num_global_channels="10";
 # network config file for intra-group connections 
-   intra-group-connections="/Users/mmubarak/Documents/software_development/test_versions/codes/scripts/gen-cray-topo/intra-theta";
+   intra-group-connections="../src/network-workloads/conf/dragonfly-custom/intra-theta";
 # network config file for inter-group connections
-   inter-group-connections="/Users/mmubarak/Documents/software_development/test_versions/codes/scripts/gen-cray-topo/inter-theta";
+   inter-group-connections="../src/network-workloads/conf/dragonfly-custom/inter-theta";
 # routing protocol to be used 
    routing="adaptive";
 }

src/networks/model-net/dragonfly-custom.C

@@ -22,7 +22,7 @@
 
 #define DUMP_CONNECTIONS 0
 #define CREDIT_SIZE 8
-#define DFLY_HASH_TABLE_SIZE 262144
+#define DFLY_HASH_TABLE_SIZE 5000
 
 // debugging parameters
 #define TRACK -1
@@ -834,11 +834,8 @@ terminal_custom_init( terminal_state * s,
       s->vc_occupancy[i]=0;
     }
 
-   s->rank_tbl = qhash_init(dragonfly_rank_hash_compare, dragonfly_hash_func, DFLY_HASH_TABLE_SIZE);
-
-   if(!s->rank_tbl)
-       tw_error(TW_LOC, "\n Hash table not initialized! ");
 
+   s->rank_tbl = NULL;
    s->terminal_msgs = 
        (terminal_custom_message_list**)malloc(s->num_vcs*sizeof(terminal_custom_message_list*));
    s->terminal_msgs_tail = 
@@ -1479,6 +1476,9 @@ static void packet_arrive(terminal_state * s, tw_bf * bf, terminal_custom_messag
     // NIC aggregation - should this be a separate function?
     // Trigger an event on receiving server
 
+    if(!s->rank_tbl)
+        s->rank_tbl = qhash_init(dragonfly_rank_hash_compare, dragonfly_hash_func, DFLY_HASH_TABLE_SIZE);
+    
     struct dfly_hash_key key;
     key.message_id = msg->message_id; 
     key.sender_id = msg->sender_lp;
@@ -2026,8 +2026,10 @@ dragonfly_custom_terminal_final( terminal_state * s,
 
     //if(s->packet_gen != s->packet_fin)
     //    printf("\n generated %d finished %d ", s->packet_gen, s->packet_fin);
+   
+    if(s->rank_tbl)
+        qhash_finalize(s->rank_tbl);
     
-    qhash_finalize(s->rank_tbl);
     rc_stack_destroy(s->st);
     free(s->vc_occupancy);
     free(s->terminal_msgs);

src/networks/model-net/dragonfly.c

@@ -33,7 +33,7 @@
 #define COLLECTIVE_COMPUTATION_DELAY 5700
 #define DRAGONFLY_FAN_OUT_DELAY 20.0
 #define WINDOW_LENGTH 0
-#define DFLY_HASH_TABLE_SIZE 262144
+#define DFLY_HASH_TABLE_SIZE 5000
 
 // debugging parameters
 #define TRACK -1
@@ -757,11 +757,8 @@ terminal_init( terminal_state * s,
       s->vc_occupancy[i]=0;
     }
 
-   s->rank_tbl = qhash_init(dragonfly_rank_hash_compare, dragonfly_hash_func, DFLY_HASH_TABLE_SIZE);
-
-   if(!s->rank_tbl)
-       tw_error(TW_LOC, "\n Hash table not initialized! ");
 
+   s->rank_tbl = NULL;
    s->terminal_msgs = 
        (terminal_message_list**)malloc(1*sizeof(terminal_message_list*));
    s->terminal_msgs_tail = 
@@ -1491,6 +1488,9 @@ static void packet_arrive(terminal_state * s, tw_bf * bf, terminal_message * msg
     // NIC aggregation - should this be a separate function?
     // Trigger an event on receiving server
 
+    if(!s->rank_tbl)
+        s->rank_tbl = qhash_init(dragonfly_rank_hash_compare, dragonfly_hash_func, DFLY_HASH_TABLE_SIZE);
+    
     struct dfly_hash_key key;
     key.message_id = msg->message_id; 
     key.sender_id = msg->sender_lp;
@@ -2363,8 +2363,10 @@ dragonfly_terminal_final( terminal_state * s,
 
     //if(s->packet_gen != s->packet_fin)
     //    printf("\n generated %d finished %d ", s->packet_gen, s->packet_fin);
+   
+    if(s->rank_tbl)
+        qhash_finalize(s->rank_tbl);
     
-    qhash_finalize(s->rank_tbl);
     rc_stack_destroy(s->st);
     free(s->vc_occupancy);
     free(s->terminal_msgs);

src/networks/model-net/fattree.c

@@ -17,7 +17,7 @@
 #define MEAN_PROCESS 1.0
 
 #define TERMINAL_GUID_PREFIX ((uint64_t)(64) << 32)
-#define FTREE_HASH_TABLE_SIZE 262144
+#define FTREE_HASH_TABLE_SIZE 5000
 
 // debugging parameters
 #define TRACK_PKT -1
@@ -965,9 +965,9 @@ void ft_terminal_init( ft_terminal_state * s, tw_lp * lp )
    s->in_send_loop = 0;
    s->issueIdle = 0;
 
-   s->rank_tbl = qhash_init(fattree_rank_hash_compare, fattree_hash_func, FTREE_HASH_TABLE_SIZE);
-   if(!s->rank_tbl)
-     tw_error(TW_LOC, "\n Hash table not initialized! ");
+   s->rank_tbl = NULL;
+   //if(!s->rank_tbl)
+   //  tw_error(TW_LOC, "\n Hash table not initialized! ");
 
    s->params->num_terminals = codes_mapping_get_lp_count(lp_group_name, 0,
       LP_CONFIG_NM, s->anno, 0);
@@ -2381,6 +2381,8 @@ void ft_packet_arrive_rc(ft_terminal_state * s, tw_bf * bf, fattree_message * ms
 void ft_packet_arrive(ft_terminal_state * s, tw_bf * bf, fattree_message * msg,
     tw_lp * lp) {
 
+  if(!s->rank_tbl)  
+    s->rank_tbl = qhash_init(fattree_rank_hash_compare, fattree_hash_func, FTREE_HASH_TABLE_SIZE);
   //Establish msg hash keys
   struct ftree_hash_key key;
   key.message_id = msg->message_id;

src/networks/model-net/slimfly.c

@@ -25,7 +25,7 @@
 #define MEAN_PROCESS 1.0
 
 /* collective specific parameters */
-#define DFLY_HASH_TABLE_SIZE 65536
+#define DFLY_HASH_TABLE_SIZE 5000
 
 // debugging parameters
 #define TRACK 4
@@ -840,10 +840,9 @@ void slim_terminal_init( terminal_state * s,
       s->vc_occupancy[i]=0;
     }
 
-   s->rank_tbl = qhash_init(slimfly_rank_hash_compare, slimfly_hash_func, DFLY_HASH_TABLE_SIZE);
-
-   if(!s->rank_tbl)
-       tw_error(TW_LOC, "\n Hash table not initialized! ");
+    s->rank_tbl = NULL;
+//   if(!s->rank_tbl)
+//       tw_error(TW_LOC, "\n Hash table not initialized! ");
 
    s->terminal_msgs =
        (slim_terminal_message_list**)malloc(1*sizeof(slim_terminal_message_list*));
@@ -1664,6 +1663,9 @@ void slim_packet_arrive(terminal_state * s, tw_bf * bf, slim_terminal_message *
     // NIC aggregation - should this be a separate function?
     // Trigger an event on receiving server
 
+  
+  if(!s->rank_tbl)
+     s->rank_tbl = qhash_init(slimfly_rank_hash_compare, slimfly_hash_func, DFLY_HASH_TABLE_SIZE);
   tw_stime ts = g_tw_lookahead + s->params->credit_delay + tw_rand_unif(lp->rng);
 
   if(msg->packet_ID == TRACK)
@@ -1996,7 +1998,9 @@ void slimfly_terminal_final( terminal_state * s,
     if(s->terminal_msgs[0] != NULL)
 //      printf("[%lu] leftover terminal messages \n", lp->gid);
 
-    qhash_finalize(s->rank_tbl);
+    if(s->rank_tbl)
+        qhash_finalize(s->rank_tbl);
+    
     rc_stack_destroy(s->st);
     free(s->vc_occupancy);
     free(s->terminal_msgs);

tests/Makefile.subdir

@@ -12,6 +12,7 @@ check_PROGRAMS += tests/lp-io-test \
  src/network-workloads/model-net-synthetic \
  src/network-workloads/model-net-synthetic-fattree \
  src/network-workloads/model-net-synthetic-slimfly \
+ src/network-workloads/model-net-synthetic-custom-dfly \
  tests/concurrent-msg-recv tests/modelnet-simplep2p-test \
  tests/modelnet-test-collective \
  tests/modelnet-prio-sched-test \
@@ -31,6 +32,7 @@ TESTS += tests/lp-io-test.sh \
  tests/modelnet-test-dragonfly.sh \
  tests/modelnet-test-slimfly.sh \
  tests/modelnet-test-dragonfly-synthetic.sh \
+ tests/modelnet-test-dragonfly-custom-synthetic.sh \
  tests/modelnet-test-fattree-synthetic.sh \
  tests/modelnet-test-slimfly-synthetic.sh \
  tests/modelnet-p2p-bw-loggp.sh \