adjust outputs and add script for report writing

Makefile

@@ -9,7 +9,8 @@ TEST_DEV ?= GPU
 
 OMP_FLAGS = -Xcompiler -fopenmp
 
-NVFLAGS = -I. -O3 -std=c++14 -DCUDA_SM=$(SM) -arch sm_$(SM) --relocatable-device-code=true --extended-lambda $(OMP_FLAGS) -g
+EXTRA_FLAGS ?= -g
+NVFLAGS = -I. -O3 -std=c++14 -DCUDA_SM=$(SM) -arch sm_$(SM) --relocatable-device-code=true --extended-lambda $(OMP_FLAGS) $(EXTRA_FLAGS)
 
 default:
 	$(NVCC) $(NVFLAGS) gpma_bfs_demo.cu -o gpma_bfs_demo -lgomp

gpma.cuh

@@ -969,7 +969,7 @@ template <dev_type_t DEV>
 __host__ void update_gpma(GPMA<DEV> &gpma, NATIVE_VEC_KEY<DEV> &update_keys, NATIVE_VEC_VALUE<DEV> &update_values) {
     DEBUG_PRINTFLN("DBG: (ENTER UPDATE)update_gpma args, update_keys={}, values={}", rlib::printable_iter(update_keys), rlib::printable_iter(values_for_print(update_values)));
     gpma.print_status("ENTER update_gpma");
-    rlib::printfln("DBG: tree_height={}, update_size={}", gpma.tree_height, update_keys.size());
+    // rlib::printfln("DBG: tree_height={}, update_size={}", gpma.tree_height, update_keys.size());
 
     //LOG_TIME("enter_update_gpma")
 
@@ -1046,5 +1046,5 @@ __host__ void update_gpma(GPMA<DEV> &gpma, NATIVE_VEC_KEY<DEV> &update_keys, NAT
 
     anySync<DEV>();
     gpma.print_status("LEAVE update_gpma");
-    rlib::printfln("DBG: (LEAVE UPDATE) =====================================================================================");
+    //rlib::printfln("DBG: (LEAVE UPDATE) =====================================================================================");
 }

gpma_bfs_demo.cu

@@ -62,7 +62,11 @@ int main(int argc, char **argv) {
 
     LOG_TIME("before init_csr_gpma")
     constexpr size_t cpu_count = 4;
-    GPMA_multidev<cpu_count-1, 1> gpma(node_size);
+#ifndef TEST_CPUS
+#define TEST_CPUS cpu_count-1
+#define TEST_GPUS 1
+#endif
+    GPMA_multidev<TEST_CPUS, TEST_GPUS> gpma(node_size);
     cudaDeviceSynchronize();
 
     LOG_TIME("before update_gpma 1")
@@ -71,9 +75,11 @@ int main(int argc, char **argv) {
     cudaDeviceSynchronize();
 
     LOG_TIME("before first bfs")
+#ifndef _DISABLE_BFS
     gpma_bfs(gpma, node_size, edge_size, bfs_start_node, RAW_PTR(bfs_result));
     int reach_nodes = node_size - thrust::count(bfs_result.begin(), bfs_result.end(), 0);
     printf("start from node %d, number of reachable nodes: %d\n", bfs_start_node, reach_nodes);
+#endif
 
     LOG_TIME("before main loop")
     for (int i = 0; i < num_slide; i++) {
@@ -99,9 +105,11 @@ int main(int argc, char **argv) {
     printf("Graph is updated.\n");
     LOG_TIME("before second bfs")
 
+#ifndef _DISABLE_BFS
     gpma_bfs(gpma, node_size, edge_size, bfs_start_node, RAW_PTR(bfs_result));
     reach_nodes = node_size - thrust::count(bfs_result.begin(), bfs_result.end(), 0);
     printf("start from node %d, number of reachable nodes: %d\n", bfs_start_node, reach_nodes);
+#endif
     LOG_TIME("after second bfs")
 
     return 0;

par_test.fish

+#!/usr/bin/fish
+
+for i in (seq 8)
+    echo CPUS=$i ===========================================
+    make EXTRA_FLAGS="-DTEST_CPUS=$i -DTEST_GPUS=0 -DTEST_DISABLE_BFS"
+    and time ./gpma_bfs_demo /tmp/999999999.pokec.txt 0
+end
+
+
+