refine cube_cli.cpp

core/cube/cube-api/src/cube_cli.cpp

@@ -31,8 +31,9 @@ DEFINE_bool(print_output, false, "print output flag");
 DEFINE_int32(thread_num, 1, "thread num");
 std::atomic<int> g_concurrency(0);
 
-std::vector<uint64_t> time_list;
+std::vector<std::vector<uint64_t>> time_list;
 std::vector<uint64_t> request_list;
+int turns = 1000000 / FLAGS_batch;
 
 namespace {
 inline uint64_t time_diff(const struct timeval& start_time,
@@ -97,7 +98,7 @@ int run(int argc, char** argv, int thread_id) {
   while (g_concurrency.load() >= FLAGS_thread_num) {
   }
   g_concurrency++;
-
+  time_list[thread_id].resize(turns);
   while (index < file_size) {
     // uint64_t key = strtoul(buffer, NULL, 10);
 
@@ -121,47 +122,12 @@ int run(int argc, char** argv, int thread_id) {
       }
       ++seek_counter;
       uint64_t seek_cost = time_diff(seek_start, seek_end);
-      seek_cost_total += seek_cost;
-      if (seek_cost > seek_cost_max) {
-        seek_cost_max = seek_cost;
-      }
-      if (seek_cost < seek_cost_min) {
-        seek_cost_min = seek_cost;
-      }
+      time_list[thread_id][request - 1] = seek_cost;
 
       keys.clear();
       values.clear();
     }
   }
-  /*
-    if (keys.size() > 0) {
-      int ret = 0;
-      values.resize(keys.size());
-      TIME_FLAG(seek_start);
-      ret = cube->seek(FLAGS_dict, keys, &values);
-      TIME_FLAG(seek_end);
-      if (ret != 0) {
-        LOG(WARNING) << "cube seek failed";
-      } else if (FLAGS_print_output) {
-        for (size_t i = 0; i < keys.size(); ++i) {
-          fprintf(stdout,
-                  "key:%lu value:%s\n",
-                  keys[i],
-                  string_to_hex(values[i].buff).c_str());
-        }
-      }
-
-      ++seek_counter;
-      uint64_t seek_cost = time_diff(seek_start, seek_end);
-      seek_cost_total += seek_cost;
-      if (seek_cost > seek_cost_max) {
-        seek_cost_max = seek_cost;
-      }
-      if (seek_cost < seek_cost_min) {
-        seek_cost_min = seek_cost;
-      }
-    }
-  */
   g_concurrency--;
 
   // fclose(key_file);
@@ -171,12 +137,6 @@ int run(int argc, char** argv, int thread_id) {
     LOG(WARNING) << "destroy cube api failed err=" << ret;
   }
 
-  uint64_t seek_cost_avg = seek_cost_total / seek_counter;
-  LOG(INFO) << "seek cost avg = " << seek_cost_avg;
-  LOG(INFO) << "seek cost max = " << seek_cost_max;
-  LOG(INFO) << "seek cost min = " << seek_cost_min;
-
-  time_list[thread_id] = seek_cost_avg;
   request_list[thread_id] = request;
 
   return 0;
@@ -202,24 +162,27 @@ int run_m(int argc, char** argv) {
   uint64_t min_time = 1000000;
   uint64_t request_num = 0;
   for (int i = 0; i < thread_num; i++) {
-    sum_time += time_list[i];
-    if (time_list[i] > max_time) {
-      max_time = time_list[i];
-    }
-    if (time_list[i] < min_time) {
-      min_time = time_list[i];
+    for (int j = 0; j < request_list[i]; j++) {
+      sum_time += time_list[i][j];
+      if (time_list[i][j] > max_time) {
+        max_time = time_list[i][j];
+      }
+      if (time_list[i][j] < min_time) {
+        min_time = time_list[i][j];
+      }
     }
     request_num += request_list[i];
   }
-  uint64_t mean_time = sum_time / thread_num;
+  uint64_t mean_time = sum_time / (thread_num * turns);
   uint64_t main_time = time_diff(main_start, main_end);
-  LOG(INFO) << thread_num << " thread seek cost"
-            << " avg = " << std::to_string(mean_time)
-            << " max = " << std::to_string(max_time)
-            << " min = " << std::to_string(min_time);
-  LOG(INFO) << " total_request = " << std::to_string(request_num)
-            << " speed = " << std::to_string(request_num * 1000000 /
-                                             main_time)  // mean_time us
+  LOG(INFO) << "\n"
+            << thread_num << " thread seek cost"
+            << "\navg = " << std::to_string(mean_time)
+            << "\nmax = " << std::to_string(max_time)
+            << "\nmin = " << std::to_string(min_time);
+  LOG(INFO) << "\ntotal_request = " << std::to_string(request_num)
+            << "\nspeed = " << std::to_string(request_num * 1000000 /
+                                              main_time)  // mean_time us
             << " query per second";
   return 0;
 }

python/examples/criteo_ctr_with_cube/benchmark_cube.sh

 rm profile_log
 
-#wget https://paddle-serving.bj.bcebos.com/unittest/ctr_cube_unittest.tar.gz --no-check-certificate
+wget https://paddle-serving.bj.bcebos.com/unittest/ctr_cube_unittest.tar.gz --no-check-certificate
 tar xf ctr_cube_unittest.tar.gz
 mv models/ctr_client_conf ./
 mv models/ctr_serving_model_kv ./
 mv models/data ./cube/
 
-#wget https://paddle-serving.bj.bcebos.com/others/cube_app.tar.gz --no-check-certificate
+wget https://paddle-serving.bj.bcebos.com/others/cube_app.tar.gz --no-check-certificate
 tar xf cube_app.tar.gz
 mv cube_app/cube* ./cube/
 sh cube_prepare.sh &
@@ -24,7 +24,9 @@ do
     echo "========================================"
     echo "batch size : $batch_size" >> profile_log
     echo "thread num : $thread_num" >> profile_log
+    tail -n 7 profile | head -n 4 >> profile_log
     tail -n 2 profile >> profile_log
+
 done
 done