In the ‘early days of PL9.0’ i think some similar, even wonder on GC (and write long comments about that), and so on.
I like to note: other forum users test this export stuff (performance) a lot. May worth to check thru. Seems something issue may happen with export - however i guess its on some constellation.
But now i don’t want to talk about Export, but at general why ‘GPU GC’ / ‘Free up VRAM’ may not helps at all.
To get some idea:
Once you use any AI mask → AI model loaded to GPU VRAM. If you use Deeprime rendering → DP NR also load to GPU VRAM, and so on. And if some ‘free up’ from GPU VRAM happen → you may not get any advantage → As its loaded again / used up again in the next photo. So, VRAM at least maxed out of all things you use. And make no sense to ‘release it from GPU VRAM’. As during Export you also able to continue editing the photos → some thing need to do (allocate) ‘twice’ (at least in concept), like: NR for editing, and also for Export NR happen, AI mask used for Editing, and Export AI mask handling also needed.
Its may not exact fully, but to get the idea.
Tip: may try to kill DopCor. Try with 1 (one) paralel export, etc.
Agreed! My main headshaker is that I use PL for everything start to finish. I used 8 for a year with DP2 on every image and NEVER… had an issue. Never 1. PL8 was flawless performance. So WHY… if I’m not using Ai at all within PL9 for anything because I had 600 images to get through after Assateague trip…. WHY… would it fail to execute now? It’s still a PL9 issue. And I literallly just ran the 9.3.0 update an after a dozen images or so.. failed to execute. It is not releasing something that’s backbuilding an crashing. I’m no computer guy, but something is way wrong with the management of how it “carries” into image after image until it crashes.
I’m trying to learn all the geeky stuff along the way here. As I never really needed to with 8. It just worked on my new gaming PC during Covid when I bought it. That’s it. So I didn’t have a VRAM or Graphics settings, or cashe issues. I worked all day only using DP2 and standard edits. But I have found today after testing some sunset photos from last night doing all kinds of masks, DP2, whatever…. There was a mention regarding potential bottlenecking of my system performance of the GPU vs graphics… so I made some changes in performance on the system via Predator Sense from balanced to performance on the system to boost the fans on the laptop. They were set to optimize, wanting to keep things quiet. I boosted it performance mode and the fans were howling during those processes. But I made it further that usual until a crash. Reset, starts working again.
I have asked time and time again what setting my system details are best to make sure settings were made to handle the intense processes. I only use DXO Photolab 9. Nothing else. I keep all my cards in a safe, and back up finished work on external drives. I wonder if I need sidecar files… But it’s always dead silence. Just an email reply with “send us your performance data”, over and over… with no real answers.
Unfortunately Photolab has a long time reputation for negligible support, along with a lot of other software companies. Support means employing people. employing people means costs. So silence or kicking it into the very long grass is the norm.
Sidecars/.dop files provide a backup should your PL database be corrupted in some way … I recommend generating and keeping sidecar files associated with each source image.
It’s not transferable because no-one wants to buy it. A couple of code changes and everything is transferable. To be honest, I wouldn’t sell it to anyone. I am not that bad a guy.