If you queue up several videos in Topaz Video Enhance Ai, the software will complete the first one, and will not release the RAM it is using, and then start the second one. After a few videos are complete the software grinds to a halt as all the RAM on your computer is used up.
I would respectfully suggest you open a formal support ticket as the devs should defintely be interested in getting that issue resolved.
I had this exactly the same issue since yesterday(never happened before) after upgrading Nvidia Game Ready Driver to the latest version 496.49(GPU is RTX 2070) using the latest VEAI version 2.4.0, and the issue had been solved after switching to Nvidia Studio Driver before submitting a support ticket.
I have conflicting observation with TVEAI 2.4.0.
I have a macbook pro (not the M1, but Intel CPU) with buildin Intel GPU, and only 8GB RAM. if I use AHQ on a batch wmv file conversion, I see each processed file seems to add some swap memory usage, and the frame/sec getting longer, from the initial 1.5frame/sec increasing to 5.5frame/sec. Stop the batch, clean up the queue, and then re-add the files into queue, that will fix the frame/sec issue right away.
In this case, the Intel buildin GPU does not seem to be used, it is like CPU operation (I could be wrong here). And virtual memory (swap) is my current prime suspect.
On a different Windows PC, where I have 32GB RAM, and an AMD Radeon Pro GPU, but this time, I am using DTDS model, I do not see any such slowing, memory issue, and it seems all memory used is from that GPU (2GB dedicated GPU memory usage, and 18.9GB shared GPU memory usage, whatever those numbers mean)
Obviously, the observation is on different hardware (GPU, and total RAM, OS), and also different model used. I do not have time to do systematic study on this. But currently I lean toward pointing my finger to RAM and swap usage.