CPU - latest regarding speed and multicore?

vray
rhino5
cpu
unhandled
hardware

#1

I’m using Rhino 5 and Vray 3. As I add lights, both modeling and rendering times have crashed. The biggest problems are around doing modeling (rotate, drag, zoom, etc) and vray interactive rendering. For final or artistic renders, I’d like more speed, but I can allow the machine to run while I’m away from the computer.

Thinking about these three CPU options:
i7-7700k (four core) at 4.8ghz
i7-6950x (ten core) @ 4.3/4.1 ghz v3
i9-7900x (ten core) @ 3.3ghz (4.3 “turbo boost”)
To improve interactive rendering I’m looking at the Nvidia GTX 1080ti.

I read the forum and here is what I gathered from posts made about 2014 and 2015 (I imagine I’m over-simplifying):

  • Modeling (rotate, drag, etc) uses CPU - fewer cores or a single core at higher Ghz are faster than a slower multi-core CPU. Modeling tasks are sequential so do not lend themselves to divided CPU threads.
  • Rendering uses multi-core CPU, so if rendering is the priority, then a multi-core CPU is preferred over a faster single or quad CPU. Rendering can be divided among multiple threads.
  • Vray interactive mode relies heavily on GPU, so more Cuda cores are preferred here - rather than faster CPU speed or more cores in a CPU.

Has the thinking on multiple cores evolved any since 2014/2015?

What would you prefer among the i7-7700k, i7-6950X and i9-7900X CPU’s?

I look forward to your comments. I don’t have a lot of gear experience, so let me know if I’m off base.
Thanks!
Jeff


#2

The information on the link below may help explain how the GPU assists Rhino 5:

http://www.simplyrhino.co.uk/support/pny-nvidia-quadro-graphic-cards

The cards tested are not current but the general guidance remains true.

Hope that helps with the GPU part of your investigations.

Paul


#3

with the latest CPUs, you’re pretty much no longer penalized with slower clock speeds due to buying more cores… (still a little… but almost negligible)

the turbo speeds of an 8-core have gotten really close to the speed of a quad… and they can hold that speed for much longer (basically, it will hold the speed indefinitely if using only one or two of the cores).

so, with higher core count cpus now, you can do all of your single core operations on a very fast processor… then when you switch to rendering, you have a lot of cores at your disposable.


#4

I see enough information on the GPU here in the forum, but I look for some general assumptions on hardware, like

  • more CPU cores are better for …
  • higher CPU frequency is better for …
  • higer cpubenchmark.net is better for …
  • more RAM is better for …
  • more GPU RAM is better for …
    (as examples)

Specifically we are are having problems with big/complex files (500MB-1,5TB and more). We started separating the file into blocks, but still we have to work (at least navigate) in a file that contains all stuff. And the viewing experience is somewhat slow … we go to the copy machine in between …