Loading...
> >

What spec contributes the most toward faster physics simulations?

GordigGordig Posts: 9,900
in Technical Help (nuts n bolts)

I've been playing around a lot with physics in Cinema 4D lately. My current computer is as follows:

Gigabyte Z170X UD5 TH

Intel Core i7-7700K Kaby Lake Quad-Core 4.2 GHz

32GB DDR4 2400 

EVGA Nvidia RTX 2060 Super

What should I look out for, either in terms of upgrading my current computer or for a future build, that would help process physics faster and smoother: RAM, CPU cores, CPU clock speed, GPU, other? I've got a build put together on Newegg with a 32-core Threadripper and 256 GB of RAM, but it will be quite a while before I can afford that. I can afford a Titan RTX to supplement my 2060 Super, but am waiting for the right time to buy (and will definitely wait until we have final specs for Ampere). I've also been eyeing an additional 32 GB of RAM, because my MB can go up to 64. What should I prioritize?

Comments

  • SevrinSevrin Posts: 6,305

    https://www.daz3d.com/forums/categories/cinema-4d-discussion

  • GordigGordig Posts: 9,900
    edited August 2020
    Sevrin said:

    https://www.daz3d.com/forums/categories/cinema-4d-discussion

    techhelp.PNG
    1566 x 331 - 32K
    Post edited by Gordig on
  • Richard HaseltineRichard Haseltine Posts: 99,440

    Yes, but if you put the question in the Cinema 4D forum (you can edit the first post to move it) you will get people who may know the answer. leaving it here people will assume it's for DS and either not notice that it isn't and give incorrect answers or just leave without answering.

  • GordigGordig Posts: 9,900

    It's not just a C4D question, though. I think it's reasonable to assume that Max, Maya, Blender et al process their physics in more or less the same way, and what improves physics sims in one software will work for another.

  • skinklizzardskinklizzard Posts: 218
    edited August 2020

    thats a mostly on the assumption, it will depend largely on the type of simulation whether it can efficiently be run in parralel or not. lots of parallel calculations and your greatest speed up will be more cores, if the simulation has a non parallel bottle neck calculation (calculating total system energy of multiple objects/nodes for instance) where only one core can feasibly calculate the result at once then a faster core speed will improve things.

    In general you'll probably see an improvement in most cases with more cores, 3D physics engines fake a fair amount anyway so a lot can be done in parallel.

    example using cloth sim in dforce in daz studio, simple clothing item dropping to floor, 43 seconds on my gtx 970 with 1664 cuda cores at ~1.1-1.2 ghz, 2 minutes 8 seconds on my i9-9900k with 8 cores at ~4.6 ghz

    edit: some of that speed up can probably be attributed to the design of cuda cores vs general purpose cpu cores though

    Post edited by skinklizzard on
  • kenshaw011267kenshaw011267 Posts: 3,805

    No it isn't. I have no idea how C4D does physics. Check with them.

Sign In or Register to comment.