PhotoAI is very slow - uses integrated GPU rather than dedicated GPU

PhotoAI is horribly slow due to it using the integrated Intel GPU rather than dedicated Radeon GPU.

Steps to reproduce issue:

  1. Attempt to use PhotoAI
  2. Go and make a cup of tea.
  3. Resort to using CPU which is franctionally less aweful.

Topaz Photo AI v3.0.4 on 2019 MacBook Pro. Earlier versions also affected, been like this since I upgraded to 2.7 (I think) from 2.0.

Please send me your system profile information so I can check if your computer hardware is up to date.

You can securely submit your files(s) to my Dropbox using the link below.

Please be sure to send me a note here to let me know you sent something.

Profile uploaded

To help troubleshoot this problem I’d like for you to follow the directions below:

Topaz Photo AI | plist Deletion

I will note that Intel CPU Mac devices are slightly underpowered for AI processing and will have slower performance. We have some more information on our Topaz Labs Documentation page that I will link below for your review:

Topaz Photo AI | System Requirements

Many users with Intel MacOS machines have reported that Topaz Photo AI v2.4.2 is most suited for their needs and experiences better processing times. I have included a link to that previous version below if you would like to revert to this version for your machine:

Topaz Photo AI | v2.4.2 | Mac Installer

Thank you for taking the time to reply, I hadn’t checked the system requirements recently. I can understand having only 4GB of video RAM being a problem for many models and large images.

However, why is the regular brush selection tool so slow? Surely there is very little processing required for that? And once the superpixels have been identified, why is that selection method so slow as well? There is at least 1s lag just moving the brush around the image - not even clicking and painting in!

We have some bugs associated with the brush speed in some of the recent updates, but I can assure you that our developers are hard at work resolving this problem in an up-and-coming version release.

1 Like

Just upgraded to v3.1.0 and it is much improved, actually usable again. Thank you!

1 Like