You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem? Please describe.
At present, our rendering requires a good GPU to achieve a good rendering effect. At the same time, the average video memory usage is 4-5G. Is it possible to reduce this value?
I ran into this problem when we wanted to develop a Web system that could render to multiple users simultaneously.However, due to the limitations of machine performance and GPU memory, it is difficult to achieve good results when multiple users perform at the same time.
Implementing this feature may be of great help to the algorithm landing.However, my ability is not enough for me to complete this task alone. :(
SophusDavid
changed the title
Can we add some algorithms to speed up rendering and reduce GPU memory usage
[Feature Request]Can we add some algorithms to speed up rendering and reduce GPU memory usage
Jun 2, 2023
Is your feature request related to a problem? Please describe.
At present, our rendering requires a good GPU to achieve a good rendering effect. At the same time, the average video memory usage is 4-5G. Is it possible to reduce this value?
I ran into this problem when we wanted to develop a Web system that could render to multiple users simultaneously.However, due to the limitations of machine performance and GPU memory, it is difficult to achieve good results when multiple users perform at the same time.
Implementing this feature may be of great help to the algorithm landing.However, my ability is not enough for me to complete this task alone. :(
Describe the solution you'd like
I think we can refer to MERF.
Model baking may help to achieve this goal.
Another idea I had was to use a front-end loading model for rendering so that the back end didn't have too much GPU memory pressure.
The text was updated successfully, but these errors were encountered: