search

hustvl / TopFormer

TopFormer: Token Pyramid Transformer for Mobile Semantic Segmentation, CVPR2022
Other
373 stars 42 forks source link

How do you calculate the latency? #12

Open xuemanshanzhong opened 2 years ago

xuemanshanzhong commented 2 years ago

Hi, can you tell me the way you calculate the latency of the model? Do you use the benchmark.py or other codes? Thanks.

mulinmeng commented 2 years ago

We report the latency on mobile device. To test the inference speed in mobile device, please refer to tnn_runtime. If you want to calculate the latency on GPU, you can indeed refer to benchmark.py.

You can give us feedback if you have any questions or results.

wmkai commented 1 year ago

We report the latency on mobile device. To test the inference speed in mobile device, please refer to tnn_runtime. If you want to calculate the latency on GPU, you can indeed refer to benchmark.py.

You can give us feedback if you have any questions or results.

Hi, I tried to follow tnn_runtime and calculate the latency in my Mi10 phone with snapdragon 865. But results show the tested latency is 3~4 times longer than the results in the paper. I don't know if there is something wrong with my method.