自己做了一个物体检测项目，Win10环境，发表通用小设备运行正常，能预测图片。

安装了cuda11，cuDNN8，并通过vs2019测试了cuda项目的例子成功。

1.发布服务器版SDK，下载后，运行sdk服务成功，但是打开图片进行预测时网页报错：Internal Server Error

问题：在vs2019测试自带例程代码时，开始也报错。后面网上查询，发现时vs2019的cuda项目需要修改gpu算力跟自己显卡算力一致，修改后可以运行。

          请问EasyDL的服务器sdk，需要修改GPU算力吗？在哪里改？

下附日志文件：

2021-08-05 09:17:12,213 DEBUG [EasyEdge] 9704 CUDA Driver Version: 11.0, CUDA Runtime Version: 11.0
2021-08-05 09:17:12,216 DEBUG [EasyEdge] 9704 The recommended threshold is 0.2
2021-08-05 09:17:12,221 DEBUG [EasyEdge] 9704 Rescale mode is 0, target_size: 0, max_size: 0
2021-08-05 09:17:13,611 DEBUG [EasyEdge] 9704 Serial num was set by global_controller->set_licence_key
2021-08-05 09:17:14,974 DEBUG [EasyEdge] Local license is ok.
2021-08-05 09:17:14,974 DEBUG [EasyEdge] 9704 Load graph, this may take a while. (PADDLEFLUID)
2021-08-05 09:17:16,984 DEBUG [EasyEdge] 12308 Stat done(0)
2021-08-05 09:17:19,972 DEBUG [EasyEdge] 9704 Thread num 1, use gpu=1, device=0
2021-08-05 09:17:23,559 INFO [EasyEdge] 9704 Allocate graph success.
2021-08-05 09:17:23,602 INFO [EasyEdge] 9704 HTTP is now serving at 127.0.0.1:24401, holding 1 instances
2021-08-05 09:17:23,755 INFO [EasyEdge] 10848 [access_log] "GET /get/service_id" 200
2021-08-05 09:17:28,403 DEBUG [EasyEdge] 2412 Decoding request data, size 586728
2021-08-05 09:17:28,691 DEBUG [EasyEdge] 2412 Input image size: 2448, 2048
2021-08-05 09:17:28,691 DEBUG [EasyEdge] 2412 Lock "infer" aquired, waited 0us
2021-08-05 09:17:28,794 DEBUG [EasyEdge] 2412 Image will be resized to 608,608
2021-08-05 09:17:28,923 DEBUG [EasyEdge] 2412 Image will be converted to RGB

--------------------------------------
C++ Traceback (most recent call last):
--------------------------------------
Not support stack backtrace yet.

----------------------
Error Message Summary:
----------------------
ExternalError: Cuda error(209), no kernel image is available for execution on the device.
[Advise: Please search for the error code(209) on website( https://docs.nvidia.com/cuda/archive/9.0/cuda-runtime-api/group__CUDART__TYPES.html#group__CUDART__TYPES_1g3f51e3575c2178246db0a94a430e0038 ) to get Nvidia's official solution about CUDA Error.] (at D:\code\Paddle\paddle\fluid\platform\gpu_info.cc:459)

2021-08-05 09:17:36,913 INFO [EasyEdge] 2412 [access_log] "POST /" 500 "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/92.0.4515.131 Safari/537.36"