Windows 10 Graphics Settings: App Preference

Haven’t seen this discussed here, and not quite sure how this works, doesn’t work, or if it works at all with Shotcut. Just discovered this today.

I have multiple versions of Shotcut installed and discovered this was set to a really old version, I just changed it to the current version. For those who don’t install multiple versions, you only have to do this once.


ApplicationFrameHost_2020-05-02_17-26-37

Windows 10 Home 1909 (Build 18363.778)
GTX 1070

My understanding is that this is a hint to Windows as to which Graphics processor to use for an application. The application can still over-ride your request. Also, if (as seems likely from your screenshot) you only have one GPU then it is irrelevant, since the app will run on that GPU, since it is the only one.

GPU Preference Description
System default To let Windows decide the best GPU for your application.
Power saving A request to run the application on the most power saving GPU available.
High performance A request to run the application on the most high performance GPU available.

Generally, the power saving GPU is the integrated GPU on a system, and the high performance GPU is the discrete GPU or external GPU. If you have both a discrete GPU and an external GPU on a system, the external GPU is considered the high performance GPU.

For the point of conversation I have an i7-7700k, that would be Intel HD Graphics 630. I don’t have any monitors plugged into the motherboard, just my 2 monitors on the GTX 1070. I wouldn’t set things up with the HD Graphics 630 anyway. Motherboard is a ASUS ROG Maximus IX Hero Z270. I built the computer myself and it’s highly possible I don’t have settings correct in the EUFI.

I’m curious if would this help out with exporting “Use hardware encoder” and Parallel processing. Perhaps I’m mixing up terms that don’t have anything to do with each other. Also interested in hearing if people have used this, and what settings they used and if they have noticed improvements or not with performance.

The two terms are different. In simple terms:

Parallel processing involves splitting up the work into separate tasks and running these tasks concurrently on threads allocated to the CPU.

Hardware encoding involves the GPU doing one piece of work but on many of the same items at the same time.

Simple example - A restaurant kitchen:

The different staff are doing different jobs at at the same time: waiters, manager, sous-chef, chef, etc… They are parallel processing (multiple tasks or people doing a different job of work at the same time).

The automatic dishwasher is cleaning many pieces of cutlery, dishes and plates at the same time, which is like hardware encoding (1 “processor” GPU or dishwater) operating on multiple items at the same time)

A setting like this applies to laptops with switchable graphics(where both cards are connected to the monitor simultaniously), a desktop however has to use whatever gpu a display is attatched to to drive applications on that display. The only exception to this is pure compute where later model intel chips can still use things like quicksync for video decoding/encoding on the cpu however NVENC would be a better choice considering the 1070 you have(quicksync encode might make sense if you had say a GT1030 though)

This topic was automatically closed after 90 days. New replies are no longer allowed.