fix

python/examples/pipeline/bert/benchmark.sh

@@ -6,9 +6,9 @@ ps -ef | grep web_service | awk '{print $2}' | xargs kill -9
 sleep 3
 python3 benchmark.py yaml local_predictor 1 gpu 
 rm -rf profile_log_$modelname
-for thread_num in 1 
+for thread_num in 1 8 16 
 do
-  for batch_size in 20
+  for batch_size in 1 10 100
   do
     echo "----Bert thread num: $thread_num batch size: $batch_size mode:http ----" >>profile_log_$modelname
     rm -rf PipelineServingLogs
@@ -34,9 +34,9 @@ ps -ef | grep web_service | awk '{print $2}' | xargs kill -9
 sleep 3
 python3 benchmark.py yaml local_predictor 1 gpu
 
-for thread_num in 1 
+for thread_num in 1 8 16
 do
-  for batch_size in 20
+  for batch_size in 1 10 100
   do
     echo "----Bert thread num: $thread_num batch size: $batch_size mode:rpc ----" >>profile_log_$modelname
     rm -rf PipelineServingLogs

python/examples/pipeline/bert/config.yml

@@ -14,4 +14,4 @@ op:
       - pooled_output
       model_config: bert_seq128_model/
 rpc_port: 9998
-worker_num: 1
+worker_num: 20

python/examples/pipeline/ocr/benchmark.sh

@@ -6,7 +6,7 @@ ps -ef | grep web_service | awk '{print $2}' | xargs kill -9
 sleep 3
 python3 benchmark.py yaml local_predictor 1 gpu 
 rm -rf profile_log_$modelname
-for thread_num in 1 
+for thread_num in 1 8 16
 do
   for batch_size in 1
   do
@@ -34,7 +34,7 @@ ps -ef | grep web_service | awk '{print $2}' | xargs kill -9
 sleep 3
 python3 benchmark.py yaml local_predictor 1 gpu
 
-for thread_num in 1 
+for thread_num in 1 8 16
 do
   for batch_size in 1
   do

python/examples/pipeline/ocr/config.yml

@@ -6,7 +6,7 @@ http_port: 9999
 
 #worker_num, 最大并发数。当build_dag_each_worker=True时, 框架会创建worker_num个进程，每个进程内构建grpcSever和DAG
 ##当build_dag_each_worker=False时，框架会设置主线程grpc线程池的max_workers=worker_num
-worker_num: 1
+worker_num: 5
 
 #build_dag_each_worker, False，框架在进程内创建一条DAG；True，框架会每个进程内创建多个独立的DAG
 build_dag_each_worker: false
@@ -20,6 +20,9 @@ dag:
 
     #使用性能分析, True，生成Timeline性能数据，对性能有一定影响；False为不使用
     use_profile: false
+    tracer:
+        interval_s: 10
+
 op:
     det:
         #并发数，is_thread_op=True时，为线程并发；否则为进程并发
@@ -37,7 +40,7 @@ op:
             fetch_list: ["concat_1.tmp_0"]
 
             #计算硬件ID，当devices为""或不写时为CPU预测；当devices为"0", "0,1,2"时为GPU预测，表示使用的GPU卡
-            devices: "0"
+            devices: "2"
     rec:
         #并发数，is_thread_op=True时，为线程并发；否则为进程并发
         concurrency: 2
@@ -61,4 +64,4 @@ op:
             fetch_list: ["ctc_greedy_decoder_0.tmp_0", "softmax_0.tmp_0"] 
 
             #计算硬件ID，当devices为""或不写时为CPU预测；当devices为"0", "0,1,2"时为GPU预测，表示使用的GPU卡
-            devices: "0"
+            devices: "2"

python/examples/pipeline/ocr/web_service.py

@@ -12,9 +12,9 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 try:
-    from paddle_serving_server.web_service import WebService, Op
-except ImportError:
     from paddle_serving_server_gpu.web_service import WebService, Op
+except ImportError:
+    from paddle_serving_server.web_service import WebService, Op
 import logging
 import numpy as np
 import cv2