Extremely low performance on google cloud L4 Machines

[aihkla]

Have I written custom code (as opposed to using a stock example script provided in MediaPipe)

None

OS Platform and Distribution

Google Cloud L4 Window 2022 server

MediaPipe Tasks SDK version

No response

Task name (e.g. Image classification, Gesture recognition etc.)

Everything

Programming Language and version (e.g. C++, Python, Java)

N/A

Describe the actual behavior

Nothing, just opening the mediapipe offical studio on chrome

Describe the expected behaviour

Run smooth like a regular nvidia PC like a 3080

Standalone code/steps you may have used to try to get what you need

Basically I just turned on the google L4 cloud machine with nvidia driver 551.78.
With Intel Xeon CPU 2.20 X 16, 64GB ram and L4 GPU

Other info / Complete Logs

No response

[aihkla]

The performance is crazy, and i can confirm that the GPU performance runs perfectly fine on 3D application. In task manager, the GPU is not being used at all, the performance is at 0%

[kuaashish]

Hi @aihkla,

Our API layer currently supports GPU functionalities on macOS, regular Linux, and Windows for JavaScript, but not for Python on Windows. The scenario you are trying has not been tested, so sorry we can not provide information about its performance.

Thank you!!

[aihkla]

Hi @aihkla,

Our API layer currently supports GPU functionalities on macOS, regular Linux, and Windows for JavaScript, but not for Python on Windows. The scenario you are trying has not been tested, so sorry we can not provide information about its performance.

Thank you!!

Hi, we are indeed using javascript on web in window PC using chrome. We are NOT using python on windows. May I know if it is possible to allow mediapipe to run on CUDA core on the L4 with such setup?

Cuda with Mediapipe on JS web <---is this possible? It is quite important for us, thank you.

[kuaashish]

Hi @aihkla,

We currently do not have documentation or test scenarios for this setup. If you are interested in support for this platform, please raise a feature request here. Based on future demand and discussions, we may implement this in the near future. We are sorry, but we can not help with this right now.

Thank you!!

[github-actions[bot]]

This issue has been marked stale because it has no recent activity since 7 days. It will be closed if no further activity occurs. Thank you.

[github-actions[bot]]

This issue was closed due to lack of activity after being marked stale for past 7 days.

[google-ml-butler[bot]]

Are you satisfied with the resolution of your issue? Yes No