Finally, I wonder what the purpose of cacheless simulation is. I first thought that it would avoid using the hard drive during the frame in process, but now I noticed that it doesn’t write any cache file.
So, you created this option just to speed up (a little) the simulation for testing purpose before running the “final” simulation ?
If I’m wrong, can you explain a little ? There is still nothing in the online help.
Hey, that’s right. It does not write any cache files, so the hard disk will not be a factor. It keeps the last frame in memory, so you can render it in 3ds Max right after the simulation frame, or write images from the classic GPU Preview.
hmmm, I wonder if I would ever use that. Would it be an option, would it be possible, and would it be useful, to have it keeping every 5 or 10 last frames in memory, and write to disk only every 5th or 10th frame ?
I see a benefit if it behaves like a RAM disk, but we need to keep the cache files.
Otherwise, there is a significant speedup with Cacheless when you are simulating large cache files because it skips the slow particle compression which also does not utilize CPU well.
well, the idea is that phoenixFD doesn’t write on disk at each frame but rather a bucket of frames. I don’t know if there would be a speed improve.
step01: frames from 1 to 10 are computed without any disk write (faster, I suppose)
step02: frames from 1 to 10 are written
step03: frames from 11 to 20 are computed
and so on…
yes, but what’s the point to simulated without keeping the cache ? Or is it written at some moment ?
As I said, you can render via maxscript right after each frame is simulated, while it’s kept in RAM - you just gotta uncomment the code that comes in the default Phoenix maxsript. Or if you have a fire/smoke simulation, you can use the GPU Preview’s image saving.