请问再CPU环境下如何测试时间是2-3ms呢，相同模型我测试运行时间是400多ms呢，即使只算推理部分也100多ms

[carter275]

欢迎您使用PaddleClas并反馈相关问题，非常感谢您对PaddleClas的贡献！
提出issue时，辛苦您提供以下信息，方便我们快速定位问题并及时有效地解决您的问题：

1. PaddleClas版本以及PaddlePaddle版本：请您提供您使用的版本号或分支信息，如PaddleClas release/2.2和PaddlePaddle 2.1.0
2. 涉及的其他产品使用的版本号：如您在使用PaddleClas的同时还在使用其他产品，如PaddleServing、PaddleInference等，请您提供其版本号
3. 训练环境信息：
   a. 具体操作系统，如Linux/Windows/MacOS
   b. Python版本号，如Python3.6/7/8
   c. CUDA/cuDNN版本， 如CUDA10.2/cuDNN 7.6.5等
4. 完整的代码(相比于repo中代码，有改动的地方)、详细的错误信息及相关log

[carter275]

== PaddleClas is powered by PaddlePaddle ! ==

== ==
== For more info please go to the following website. ==
== ==
== https://github.com/PaddlePaddle/PaddleClas ==

2024-09-17 00:48:58 INFO: Global :
2024-09-17 00:48:58 INFO: batch_size : 1
2024-09-17 00:48:58 INFO: cpu_num_threads : 10
2024-09-17 00:48:58 INFO: enable_benchmark : True
2024-09-17 00:48:58 INFO: enable_mkldnn : True
2024-09-17 00:48:58 INFO: enable_profile : False
2024-09-17 00:48:58 INFO: gpu_mem : 8000
2024-09-17 00:48:58 INFO: infer_imgs : /home/aistudio/gai/test/9/NG/20240911_153540_790.bmp
2024-09-17 00:48:58 INFO: inference_model_dir : /home/aistudio/deploy
2024-09-17 00:48:58 INFO: ir_optim : True
2024-09-17 00:48:58 INFO: use_fp16 : False
2024-09-17 00:48:58 INFO: use_gpu : False
2024-09-17 00:48:58 INFO: use_tensorrt : False
2024-09-17 00:48:58 INFO: PostProcess :
2024-09-17 00:48:58 INFO: SavePreLabel :
2024-09-17 00:48:58 INFO: save_dir : ./pre_label/
2024-09-17 00:48:58 INFO: Topk :
2024-09-17 00:48:58 INFO: class_id_map_file : /home/aistudio/gai/label.txt
2024-09-17 00:48:58 INFO: topk : 3
2024-09-17 00:48:58 INFO: main_indicator : Topk
2024-09-17 00:48:58 INFO: PreProcess :
2024-09-17 00:48:58 INFO: transform_ops :
2024-09-17 00:48:58 INFO: ResizeImage :
2024-09-17 00:48:58 INFO: resize_short : 256
2024-09-17 00:48:58 INFO: CropImage :
2024-09-17 00:48:58 INFO: size : 224
2024-09-17 00:48:58 INFO: NormalizeImage :
2024-09-17 00:48:58 INFO: channel_num : 3
2024-09-17 00:48:58 INFO: mean : [0.485, 0.456, 0.406]
2024-09-17 00:48:58 INFO: order :
2024-09-17 00:48:58 INFO: scale : 0.00392157
2024-09-17 00:48:58 INFO: std : [0.229, 0.224, 0.225]
2024-09-17 00:48:58 INFO: ToCHWImage : None
预测代码运行时间: 0.1401 秒
20240911_153540_790.bmp: class id(s): [0, 1], score(s): [0.56, 0.44], label_name(s): ['OK', 'NG']
代码运行时间: 0.3865 秒

[carter275]

2024-09-17 08:42:38 INFO: Global :
2024-09-17 08:42:38 INFO: batch_size : 1
2024-09-17 08:42:38 INFO: cpu_num_threads : 10
2024-09-17 08:42:38 INFO: enable_benchmark : True
2024-09-17 08:42:38 INFO: enable_mkldnn : True
2024-09-17 08:42:38 INFO: enable_profile : False
2024-09-17 08:42:38 INFO: gpu_mem : 8000
2024-09-17 08:42:38 INFO: infer_imgs : /home/aistudio/gai/test/NG
2024-09-17 08:42:38 INFO: inference_model_dir : /home/aistudio/deploy
2024-09-17 08:42:38 INFO: ir_optim : True
2024-09-17 08:42:38 INFO: use_fp16 : False
2024-09-17 08:42:38 INFO: use_gpu : False
2024-09-17 08:42:38 INFO: use_tensorrt : False
2024-09-17 08:42:38 INFO: PostProcess :
2024-09-17 08:42:38 INFO: SavePreLabel :
2024-09-17 08:42:38 INFO: save_dir : ./pre_label/
2024-09-17 08:42:38 INFO: Topk :
2024-09-17 08:42:38 INFO: class_id_map_file : /home/aistudio/gai/label.txt
2024-09-17 08:42:38 INFO: topk : 3
2024-09-17 08:42:38 INFO: main_indicator : Topk
2024-09-17 08:42:38 INFO: PreProcess :
2024-09-17 08:42:38 INFO: transform_ops :
2024-09-17 08:42:38 INFO: ResizeImage :
2024-09-17 08:42:38 INFO: resize_short : 256
2024-09-17 08:42:38 INFO: CropImage :
2024-09-17 08:42:38 INFO: size : 224
2024-09-17 08:42:38 INFO: NormalizeImage :
2024-09-17 08:42:38 INFO: channel_num : 3
2024-09-17 08:42:38 INFO: mean : [0.485, 0.456, 0.406]
2024-09-17 08:42:38 INFO: order :
2024-09-17 08:42:38 INFO: scale : 0.00392157
2024-09-17 08:42:38 INFO: std : [0.229, 0.224, 0.225]
2024-09-17 08:42:38 INFO: ToCHWImage : None
预测代码运行时间: 0.6187 秒
20240911_153410_348.bmp: class id(s): [1, 0], score(s): [1.00, 0.00], label_name(s): ['NG', 'OK']
预测代码运行时间: 0.0967 秒
20240911_153811_895.bmp: class id(s): [1, 0], score(s): [1.00, 0.00], label_name(s): ['NG', 'OK']
预测代码运行时间: 0.0155 秒
20240911_153812_977.bmp: class id(s): [1, 0], score(s): [0.92, 0.08], label_name(s): ['NG', 'OK']
预测代码运行时间: 0.0050 秒
20240911_153817_810.bmp: class id(s): [1, 0], score(s): [1.00, 0.00], label_name(s): ['NG', 'OK']
预测代码运行时间: 0.0913 秒
20240911_154550_690.bmp: class id(s): [1, 0], score(s): [0.95, 0.05], label_name(s): ['NG', 'OK']
预测代码运行时间: 0.0934 秒
20240911_154613_782.bmp: class id(s): [1, 0], score(s): [0.75, 0.25], label_name(s): ['NG', 'OK']
预测代码运行时间: 0.0945 秒
20240911_154635_839.bmp: class id(s): [1, 0], score(s): [0.51, 0.49], label_name(s): ['NG', 'OK']
预测代码运行时间: 0.0949 秒
20240911_155645_207.bmp: class id(s): [0, 1], score(s): [0.61, 0.39], label_name(s): ['OK', 'NG']
预测代码运行时间: 0.0159 秒
20240911_155703_216.bmp: class id(s): [1, 0], score(s): [1.00, 0.00], label_name(s): ['NG', 'OK']
预测代码运行时间: 0.1067 秒
20240911_155704_282.bmp: class id(s): [1, 0], score(s): [0.82, 0.18], label_name(s): ['NG', 'OK']
预测代码运行时间: 0.0961 秒
20240911_155731_509.bmp: class id(s): [1, 0], score(s): [0.83, 0.17], label_name(s): ['NG', 'OK']
预测代码运行时间: 0.1899 秒
20240911_155739_086.bmp: class id(s): [1, 0], score(s): [1.00, 0.00], label_name(s): ['NG', 'OK']
预测代码运行时间: 0.1879 秒
20240911_155753_362.bmp: class id(s): [1, 0], score(s): [0.93, 0.07], label_name(s): ['NG', 'OK']
预测代码运行时间: 0.0946 秒
20240911_155806_174.bmp: class id(s): [1, 0], score(s): [1.00, 0.00], label_name(s): ['NG', 'OK']
预测代码运行时间: 0.1944 秒
20240911_161528_831.bmp: class id(s): [1, 0], score(s): [0.99, 0.01], label_name(s): ['NG', 'OK']
预测代码运行时间: 0.0064 秒
20240911_161535_665.bmp: class id(s): [1, 0], score(s): [0.98, 0.02], label_name(s): ['NG', 'OK']
预测代码运行时间: 0.0073 秒
20240911_161537_160.bmp: class id(s): [1, 0], score(s): [0.98, 0.02], label_name(s): ['NG', 'OK']
预测代码运行时间: 0.1034 秒
20240911_161538_510.bmp: class id(s): [1, 0], score(s): [1.00, 0.00], label_name(s): ['NG', 'OK']
预测代码运行时间: 0.0124 秒
20240911_161936_063.bmp: class id(s): [1, 0], score(s): [0.99, 0.01], label_name(s): ['NG', 'OK']
预测代码运行时间: 0.0956 秒
20240911_161938_079.bmp: class id(s): [1, 0], score(s): [0.99, 0.01], label_name(s): ['NG', 'OK']
预测代码运行时间: 0.0120 秒
20240911_161955_288.bmp: class id(s): [0, 1], score(s): [0.76, 0.24], label_name(s): ['OK', 'NG']
预测代码运行时间: 0.0949 秒
20240911_161956_487.bmp: class id(s): [1, 0], score(s): [1.00, 0.00], label_name(s): ['NG', 'OK']
预测代码运行时间: 0.0197 秒
20240911_162004_202.bmp: class id(s): [1, 0], score(s): [0.57, 0.43], label_name(s): ['NG', 'OK']
代码运行时间: 3.5929 秒

[zhangyubo0722]

使用的是同一型号的CPU吗

[TingquanGao]

The issue has no response for a long time and will be closed. You can reopen or new another issue if are still confused.

---

From Bot