With the POP Solver, I noticed I could get a 27% speed up in large particle sims by using a Switch DOP to disable Collisions.
I thought that perhaps if I set Enable Collision Attributes to Set Initial, and initial was 0, I would get similar speedups; but it was the same (a hair slower).
Should the Enable Solver be that slow? I haven't done much low-level DOPs work, so it could be that I'm missing something. I was just surprised at how much faster Switch + Null was, compared to the Enable Solver, especially considering Enables are all over the POP Solver.
The attached file has a modified POP Solver, and here are some of the results I was seeing with a 300,000 particles being birthed + some noise and color (and gravity), for 72 frames (POP Solver time only):
- No Modifications: 4.59s
- Set Inital (on Enable Collision Attrs) to 0: 4.52s
- Disconn. Enable Collisions and Reaping: 3.41s
- Disabled Coll. Attrs via Switches:3.60s
- Disabled Coll. Attrs + Reaping via Switches: 2.34s
I know reaping and collision attributes aren't something you'd normally get rid of; I noticed in the Performance Monitor those two areas were a big chunk of sim time, and could imagine scenarios where you don't have any collisions, you'd want to avoid paying that cost for no reason.
Thanks!
POP Solver: Enable Solver vs Switch DOP
1805 0 0- goldleaf
- Staff
- 4170 posts
- Joined: Sept. 2007
- Offline
-
- Quick Links