CNN Model Body

lib/utils/parallel.py

@@ -0,0 +1,23 @@
+import multiprocessing as mp
+import time
+
+
+def run_n_in_parallel(f, n, **kwargs):
+    output = mp.Queue()
+    kwargs.update(output=output)
+    # Setup a list of processes that we want to run
+    processes = [mp.Process(target=f, kwargs=kwargs) for _ in range(n)]
+    # Run processes
+    results = []
+    for p in processes:
+        p.start()
+    while len(results) != n:
+        time.sleep(1)
+        # Get process results from the output queue
+        results.extend([output.get() for _ in processes])
+
+    # Exit the completed processes
+    for p in processes:
+        p.join()
+
+    return results