Merge pull request #18 from littletomatodonkey/fix_trt_bench

fix benchmark

tools/infer/predict.py

@@ -12,7 +12,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-
 import utils
 import argparse
 import numpy as np
@@ -24,6 +23,7 @@ from paddle.fluid.core import create_paddle_predictor
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
 
+
 def parse_args():
     def str2bool(v):
         return v.lower() in ("true", "t", "1")
@@ -47,8 +47,6 @@ def parse_args():
 def create_predictor(args):
     config = AnalysisConfig(args.model_file, args.params_file)
 
-
-
     if args.use_gpu:
         config.enable_use_gpu(args.gpu_mem, 0)
     else:
@@ -58,7 +56,8 @@ def create_predictor(args):
     config.switch_ir_optim(args.ir_optim)  # default true
     if args.use_tensorrt:
         config.enable_tensorrt_engine(
-                precision_mode=AnalysisConfig.Precision.Half if args.use_fp16 else AnalysisConfig.Precision.Float32,
+            precision_mode=AnalysisConfig.Precision.Half
+            if args.use_fp16 else AnalysisConfig.Precision.Float32,
             max_batch_size=args.batch_size)
 
     config.enable_memory_optim()
@@ -104,39 +103,63 @@ def main():
         assert args.model_name is not None
         assert args.use_tensorrt == True
     # HALF precission predict only work when using tensorrt
-    if args.use_fp16==True:
+    if args.use_fp16 == True:
         assert args.use_tensorrt == True
 
     operators = create_operators()
     predictor = create_predictor(args)
 
     inputs = preprocess(args.image_file, operators)
-    inputs = np.expand_dims(inputs, axis=0).repeat(args.batch_size, axis=0).copy()
+    inputs = np.expand_dims(
+        inputs, axis=0).repeat(
+            args.batch_size, axis=0).copy()
 
     input_names = predictor.get_input_names()
     input_tensor = predictor.get_input_tensor(input_names[0])
-    input_tensor.copy_from_cpu(inputs)
+
+    output_names = predictor.get_output_names()
+    output_tensor = predictor.get_output_tensor(output_names[0])
+
+    test_num = 500
+    test_time = 0.0
     if not args.enable_benchmark:
+        inputs = preprocess(args.image_file, operators)
+        inputs = np.expand_dims(
+            inputs, axis=0).repeat(
+                args.batch_size, axis=0).copy()
+        input_tensor.copy_from_cpu(inputs)
+
         predictor.zero_copy_run()
+
+        output = output_tensor.copy_to_cpu()
+        output = output.flatten()
+        cls = np.argmax(output)
+        score = output[cls]
+        logger.info("class: {0}".format(cls))
+        logger.info("score: {0}".format(score))
     else:
-        for i in range(0,1010):
-            if i == 10:
-                start = time.time()
-            predictor.zero_copy_run()
+        for i in range(0, test_num + 10):
+            inputs = np.random.rand(args.batch_size, 3, 224,
+                                    224).astype(np.float32)
+            start_time = time.time()
+            input_tensor.copy_from_cpu(inputs)
 
-        end = time.time()
-        fp_message = "FP16" if args.use_fp16 else "FP32"
-        logger.info("{0}\t{1}\tbatch size: {2}\ttime(ms): {3}".format(args.model_name, fp_message, args.batch_size, end-start))
+            predictor.zero_copy_run()
 
-    output_names = predictor.get_output_names()
-    output_tensor = predictor.get_output_tensor(output_names[0])
             output = output_tensor.copy_to_cpu()
             output = output.flatten()
+            if i >= 10:
+                test_time += time.time() - start_time
             cls = np.argmax(output)
             score = output[cls]
             logger.info("class: {0}".format(cls))
             logger.info("score: {0}".format(score))
 
+        fp_message = "FP16" if args.use_fp16 else "FP32"
+        logger.info("{0}\t{1}\tbatch size: {2}\ttime(ms): {3}".format(
+            args.model_name, fp_message, args.batch_size, 1000 * test_time /
+            test_num))
+
 
 if __name__ == "__main__":
     main()