# 1FPS using Cuda (Nvidia Jetson Orin Nano) **URL:** \ **Category:** Python\ **Tags:** dnn, cuda, nvidia\ **Created:** [June 27, 2024, 6:50pm UTC](https://forum.opencv.org/t/1fps-using-cuda-nvidia-jetson-orin-nano/18105 "2024-06-27T18:50:04Z")\ **Posts on this page:** 5\ **Page:** 1
**Author:** ![CRP](https://avatars.discourse-cdn.com/v4/letter/c/97f17d/32.png) [@CRP](https://forum.opencv.org/u/CRP)\ **Post date:** [June 27, 2024, 6:50pm UTC](https://forum.opencv.org/t/1fps-using-cuda-nvidia-jetson-orin-nano/18105/1 "2024-06-27T18:50:05Z")
I am not sure why, but after building OpenCV with CUDA support, I am only able to achieve 1 fps, regardless of the input video quality. Does anyone know what might be causing this? I am unable to figure this out. As far as I understand, this should not be happening on the Nvidia Jetson Orin Nano. Demonstration: [Watch OPENCV with CUDA LOW FPS JETSON ORIN NANO | Streamable](https://streamable.com/xd1vy7) ---
**Author:** ![cudawarped](https://avatars.discourse-cdn.com/v4/letter/c/9dc877/32.png) [@cudawarped](https://forum.opencv.org/u/cudawarped)\ **Post date:** [June 27, 2024, 6:59pm UTC](https://forum.opencv.org/t/1fps-using-cuda-nvidia-jetson-orin-nano/18105/2 "2024-06-27T18:59:37Z")
What are you doing with CUDA, DNN inference? If so do you have FPS figures for the model you are using on different hardware? ---
**Author:** ![CRP](https://avatars.discourse-cdn.com/v4/letter/c/97f17d/32.png) [@CRP](https://forum.opencv.org/u/CRP)\ **Post date:** [June 27, 2024, 7:34pm UTC](https://forum.opencv.org/t/1fps-using-cuda-nvidia-jetson-orin-nano/18105/3 "2024-06-27T19:34:12Z")
I’m using CUDA with OpenCV for webcam processing in an emotion detection model. When the webcam preview opens, the FPS is very low. However, on different hardware, specifically with my RTX 4090, I get 30 FPS. ---
**Author:** ![crackwitz](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.opencv.org/crackwitz/32/14_2.png) [@crackwitz](https://forum.opencv.org/u/crackwitz)\ **Post date:** [June 28, 2024, 6:00am UTC](https://forum.opencv.org/t/1fps-using-cuda-nvidia-jetson-orin-nano/18105/4 "2024-06-28T06:00:05Z")
according to this, it should perform well enough for common models: > **[Jetson Benchmarks](https://developer.nvidia.com/embedded/jetson-benchmarks)** you have provided no debuggable details. I would guess that you aren’t actually using the GPU. something went wrong. ---
**Author:** ![cudawarped](https://avatars.discourse-cdn.com/v4/letter/c/9dc877/32.png) [@cudawarped](https://forum.opencv.org/u/cudawarped)\ **Post date:** [June 28, 2024, 7:45am UTC](https://forum.opencv.org/t/1fps-using-cuda-nvidia-jetson-orin-nano/18105/5 "2024-06-28T07:45:54Z")
As @crackwitz said we need an [MRE](https://stackoverflow.com/help/minimal-reproducible-example). That said if you can only get 30FPS on an RTX 4090 the back of the envelope calculation of fps on the jetson based purely on flops (not tensor cores/fp16 etc. as we don’t have an [MRE](https://stackoverflow.com/help/minimal-reproducible-example)) would be something like 30 \* 1280 GLOPS ([Jetson Orin Nano 8GB](https://www.techpowerup.com/gpu-specs/jetson-orin-nano-8-gb.c4082)) / 82000 GFLOPS ([RTX 4090](https://www.techpowerup.com/gpu-specs/geforce-rtx-4090.c3889)) ~= 0.4 fps