Author Topic: About more CPU's  (Read 2742 times)

2017-12-09, 17:03:02

Aldoren

  • Active Users
  • **
  • Posts: 25
    • View Profile
Hello

I am currently on 2x Xeon 2650V4 and thinking about buying a few render nodes. I am not that in to hardware so my question is, if i go from 2 to 4 CPU's, can i expect speed increase in single renders or will this mostly improve when rendering alot of frames etc? How big speed increase could i expect if any?

When looking at the benchmarks i can see for example that Intel(R) Xeon(R) CPU E5-2686 v3 @ 2.00GHz (×2) performs better then a setup with x4 of the same CPU's. Seems odd.

So. Simply put, if i want more power on single rendering, should i just get the best CPU's that i can fit in one single computer, or will rendernodes do the trick?

Reason for asking is that i want to get a good workflow in the office :) Thanks

2017-12-09, 23:43:53
Reply #1

Juraj

  • Active Users
  • **
  • Posts: 4762
    • View Profile
    • studio website
When looking at the benchmarks i can see for example that Intel(R) Xeon(R) CPU E5-2686 v3 @ 2.00GHz (×2) performs better then a setup with x4 of the same CPU's. Seems odd.

What particular benchmark is this ? E5-2xxx is only dual-socket architecture, quad requires either E5-4xxx or E7-8xxx (the latter can be used up to octa-socket). So it's not possible to find 2683v3 in x4 setup.

Nonetheless, for nodes, it's best to find the best ratio in price/performance for the whole node (so not just CPUs), unless you have unlimited budget or space constraints.

Corona scales very linearly and generally also with distributed rendering of single-frame. I use personal render-farm to only render single frames :- ).
Please follow my new Instagram for latest projects, tips&tricks, short video tutorials and free models
Behance  Probably best updated portfolio of my work
lysfaere.com Please check the new stuff!

2017-12-10, 16:45:18
Reply #2

Aldoren

  • Active Users
  • **
  • Posts: 25
    • View Profile
About the benchmark, i just looked at different CPU's at Coronas own benchmark list.

2017-12-11, 09:10:19
Reply #3

hrvojezg00

  • Active Users
  • **
  • Posts: 273
    • View Profile
    • www.as-soba.com
And note that 4 unit pc`s work only on Windows Server and are pretty loud due to redundant power supply and specific air cooling (passive cpu cooler combined with jet sound vents)

2017-12-11, 15:03:47
Reply #4

Juraj

  • Active Users
  • **
  • Posts: 4762
    • View Profile
    • studio website
And note that 4 unit pc`s work only on Windows Server and are pretty loud due to redundant power supply and specific air cooling (passive cpu cooler combined with jet sound vents)

Hrvoje, what is the final benchmarks of your monster build :- ) ?
Please follow my new Instagram for latest projects, tips&tricks, short video tutorials and free models
Behance  Probably best updated portfolio of my work
lysfaere.com Please check the new stuff!

2017-12-11, 15:26:44
Reply #5

Ondra

  • Administrator
  • Active Users
  • *****
  • Posts: 9048
  • Turning coffee to features since 2009
    • View Profile
btw: there are some options in BIOS to report each CPU as more CPUs, so some of the results in corona benchmark have wrong CPU description (there is no dual ryzen or 8way TR). WHen in doubt, look at the number of cores, that one is always detected correctly
Rendering is magic.How to get minidumps for crashed/frozen 3ds Max | Sorry for short replies, brief responses = more time to develop Corona ;)

2017-12-12, 15:16:32
Reply #6

hrvojezg00

  • Active Users
  • **
  • Posts: 273
    • View Profile
    • www.as-soba.com
And note that 4 unit pc`s work only on Windows Server and are pretty loud due to redundant power supply and specific air cooling (passive cpu cooler combined with jet sound vents)

Hrvoje, what is the final benchmarks of your monster build :- ) ?

00:00:22.82 13th on the list. But had to make a server room for it :)