Last-half timer

Author: seth-planet

src/modules/ObjectDetectionCoral/objectdetection_coral_multitpu.py

@@ -231,6 +231,8 @@ def main():
 
   thread_cnt = 16
   tot_infr_time = 0 
+  half_wall_start = 0 
+  half_infr_count = 0 
   if args.count > 1:
     with concurrent.futures.ThreadPoolExecutor(max_workers=thread_cnt) as executor:
       start = time.perf_counter()
@@ -240,13 +242,20 @@ def main():
         for f in concurrent.futures.as_completed(fs):
           _, infr_time, _ = f.result()
           tot_infr_time += infr_time
+
+          # Start a timer for the last ~half of the run for more accurate benchmark
+          if chunk_i > (args.count-1 / 3):
+            half_infr_count += 1
+            if half_wall_start == 0:
+              half_wall_start = time.perf_counter()
 
         # Uncomment for testing
         # import random
         # logging.info("Pause")
-        # time.sleep(random.randint(0,INTERPRETER_LIFESPAN_SECONDS*2))
+        # time.sleep(random.randint(0,INTERPRETER_LIFESPAN_SECONDS*3))
   else:
     start = time.perf_counter()
+    half_wall_start = time.perf_counter()
 
   # snapshot = tracemalloc.take_snapshot()
   # top_stats = snapshot.statistics('lineno')
@@ -256,18 +265,17 @@ def main():
   start_one = time.perf_counter()
   objs, infr_time, _ = _tpu_runner.process_image(options, copy.copy(image), args.threshold)
   tot_infr_time += infr_time
+  half_infr_count += 1
   wall_time = time.perf_counter() - start
+  half_wall_time = time.perf_counter() - half_wall_start
+  
   print('completed one run every %.2fms for %d runs; %.2fms wall time for a single run' %
                             (wall_time * 1000 / args.count, args.count,
                             (time.perf_counter() - start_one) * 1000))
 
-  # Optimizing the number of segments used for a model would result in the
-  # lowest average time spent adjusted for number of TPUs used. At some point,
-  # adding additional segments just removes from the pool of TPUs you can use
-  # for parallelism.
-  print('%.2fms avg time blocked across %d threads; %.2f avg TPU * ms / run' %
+  print('%.2fms avg time blocked across %d threads; %.2fms ea for final %d inferences' %
                             (tot_infr_time / args.count, thread_cnt,
-                             len(_tpu_runner.pipe.tpu_list) * wall_time * 1000 / args.count))
+                             half_wall_time * 1000 / half_infr_count, half_infr_count)
 
   print('-------RESULTS--------')
   if not objs: