Batch Processing 3.01 Feedback for developers

I have just completed running a batch of 1,000+ images through Topaz AI and would like to give the following feedback to the developers.

  1. That there continues to be a memory leakage problem.
  2. That you appear to be running a subroutine to clear the memory
  3. That as the memory fills, the load is shifted from the GPU to the CPU, and processing slows.
  4. After the memory is cleared the balance goes back to the GPU.
  5. If the memory clearing sub routine was triggered sooner / more often processing speeds would be improved with the GPU being more utilized and the CPU less loaded.
  6. My guess is that you are checking for when memory utilization is above 90+%
  7. This drops to 40% after the memory is cleared.
  8. Run it once the memory is over 70% or 80% to keep processing loads / speeds up.
1 Like

Thanks, this has been well documented in the release threads.