Page 1 of 1
using your own desktop
Posted: Wed Feb 05, 2020 12:11 pm
by korrupt78
So far I've been running faceswap commands on an unused desktop remotely via SSH, which has worked fine. Today for the first time I tried actually using the desktop, and whenever a faceswap command is running, the desktop GUI becomes almost unusably laggy and slow. I guess the one graphics card can't do machine learning and serve a desktop environment at the same time.
Curious if anyone else has had this experience, and if it could help to switch my monitors to the motherboard's video ports to free up the GeForce card to be used exclusively for faceswap?
Re: using your own desktop
Posted: Wed Feb 05, 2020 4:19 pm
by bryanlyon
If you have a built in IGPU on your CPU, I highly recommend using that for your desktop environment. It'd definitely make training while using the system smoother.
Re: using your own desktop
Posted: Wed Feb 05, 2020 11:50 pm
by korrupt78
bryanlyon wrote: ↑Wed Feb 05, 2020 4:19 pm
If you have a built in IGPU on your CPU, I highly recommend using that for your desktop environment. It'd definitely make training while using the system smoother.
It turns out I don't! Sorry, I'm still catching up to the current state of computer hardware. My whole computing life (1992+), if a motherboard came with a video port, it also came with the ability to drive it. But apparently the ASRock Fatal1ty B450 Gaming-ITX/AC I bought last year (to go with a AMD Ryzen 5 1600) doesn't come with any on-board video, it just delegates that task to another component. And since my Ryzen is a GPU, not APU, literally the only device in my computer capable of driving monitors is my 960.
So, I guess I'll have to content myself with light experimentation locally (while lagging up my desktop UI) and save the real work for the 2160 Super I just bought and am about to install on an Ubuntu Server tower.