Pre-Allocate limit

Came across a new problem, not sure if any checks can be made for it, even if just in the sanity check or better error messages. On my 32GB workstation, I cranked up the Pre-Allocate up to 28GB and started making PRT Volumes. When I started submitting jobs to the farm however, I ended up only really making ~3GB of particles. So while it seems on my local workstation renders than I only need ~3GB to render, any machine on the farm with less than ~28GB was failing to render. What’s worse is that I wasn’t rendering PRT Volumes at all, I had cached them out to PRT Loaders, so the final counts were actually fairly easy to predict from the PRT headers.

  • Chad

We have considered removing the Pre-Allocation because the actual dynamic memory allocation with PRT Volumes is now a lot faster too (with 100MP, it is probably about a second slower than with Pre-Allocation on).
So for the time being, we would suggest not using the Pre-Allocation at all even when working with PRT Volumes.
Given your experience, we might really remove the option from the final build…

Please try working without Pre-Allocation and let us know if you see any performance or memory management issues.