Selur's Little Message Board

Full Version: What do I need to do for Hybrid on an Nvidia GPU?
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
I was using an AMD GPU before, and now I have an NVIDIA GPU. What do I need to do to use Hybrid more effectively?
Too broad of a question to answer anything but vague.
Some filters support using gpu (opencl/cuda) others don't.
Enabling gpu support for those filters might speed up things, but it also can slow down stuff, depending on the source and the whole filter script.
Sometimes configuring the source filter to use gpu decoding might speed things up, other times it might low things down.
NVEncC would use the encoder chip on your graphic card, using 'only use encoder', would disable Avisynth&Vapoursynth filtering (and restrict the filtering options a lot), but speed up things since decoding&filtering&encoding will all be done on the graphic card.
Aside from NVEncC, none of the encoders will/can use an nvidia card in a meaningful way.
=> Only thing you can do is look at your filter configurations and test whether they have gpu support or not and whether enabling the support helps or not.

Cu Selur

Ps.: If you NVIDIA gpu is new enough, you can also use the torch addon.
does torch work with filters?
Using the torch-addon will not change the behavior of the existing filters.
The torch-addon adds additional, machine learning based, filters which use the gpu.
In general, they are way slower than conventional filters, but might to stuff normal filters do not.


Cu Selur
And is there anything I can do for the filters, apart from what you said?
No, sadly, there is no: "Best use my gpu for filtering ig it makes sense and don't if it does not."
i see, thanks mr selur