Fix get cuda device bug when only use cpu (#4066)

* fix get cuda device bug when only use cpu

* fix typo

PaddleCV/deeplabv3+/train.py

@@ -164,7 +164,7 @@ total_step = args.total_step
 
 with fluid.program_guard(tp, sp):
     if args.use_py_reader:
-        batch_size_each = batch_size // fluid.core.get_cuda_device_count()
+        batch_size_each = batch_size // utility.get_device_count()
         py_reader = fluid.layers.py_reader(capacity=64,
                                         shapes=[[batch_size_each, 3] + image_shape, [batch_size_each] + image_shape],
                                         dtypes=['float32', 'int32'])
@@ -197,7 +197,7 @@ with fluid.program_guard(tp, sp):
 
 
 exec_strategy = fluid.ExecutionStrategy()
-exec_strategy.num_threads = fluid.core.get_cuda_device_count()
+exec_strategy.num_threads = utility.get_device_count()
 exec_strategy.num_iteration_per_drop_scope = 100
 build_strategy = fluid.BuildStrategy()
 if args.memory_optimize:
@@ -225,11 +225,11 @@ else:
     binary = fluid.compiler.CompiledProgram(tp)
 
 if args.use_py_reader:
-    assert(batch_size % fluid.core.get_cuda_device_count() == 0)
+    assert(batch_size % utility.get_device_count() == 0)
     def data_gen():
         batches = dataset.get_batch_generator(
-            batch_size // fluid.core.get_cuda_device_count(),
-            total_step * fluid.core.get_cuda_device_count(),
+            batch_size // utility.get_device_count(),
+            total_step * utility.get_device_count(),
             use_multiprocessing=args.use_multiprocessing, num_workers=args.num_workers)
         for b in batches:
             yield b[0], b[1]
@@ -266,7 +266,7 @@ print("Training done. Model is saved to", args.save_weights_path)
 save_model()
 
 if args.enable_ce:
-    gpu_num = fluid.core.get_cuda_device_count()
+    gpu_num = utility.get_device_count()
     print("kpis\teach_pass_duration_card%s\t%s" %
           (gpu_num, total_time / epoch_idx))
     print("kpis\ttrain_loss_card%s\t%s" % (gpu_num, train_loss))

PaddleCV/deeplabv3+/utility.py

@@ -78,3 +78,12 @@ def check_gpu(use_gpu):
             sys.exit(1)
     except Exception as e:
         pass
+
+
+def get_device_count():
+    try:
+        device_num = max(fluid.core.get_cuda_device_count(), 1)
+    except:
+        device_num = 1
+
+    return device_num