Author Topic: Corona Benchmark Analysis  (Read 1919 times)

2018-05-31, 19:10:43

antoro

  • Users
  • *
  • Posts: 2
    • View Profile
I analyzed the Corona Benchmark Database to look at possible CPU choices for someone like myself, an individual freelancer. Unsurprisingly, Threadrippers and Core i9's performed very well. Something unexpected that I noticed is that having two Threadripper 1920X's or 1950X's is slower than just one, but not the 1900X. Can anyone provide insight as to why this is?
I then looked at value, by getting the (sometimes approximate) cost from Newegg/Amazon/Ebay (in CAD) and then dividing it into (5 min - render time). Having a dual socket Xeon E5-26X0 v2 had the most value, followed by the new Ryzen 2700/X's.

2018-05-31, 19:47:10
Reply #1

iancamarillo

  • Active Users
  • **
  • Posts: 282
    • View Profile
I get extremely slow 3ds max UI menu refresh performance and was curious if users with higher frequency CPU's experience this? I wish I could just disable all of max's ui features to and revert to basic.

Dual Xeon 1660v3 @3GHz
128gb ram
quadro k4200
win10

and

Dual Xeon 2683v3 @2-2.6GHz
128gb ram
1080ti
win10

2018-05-31, 21:01:46
Reply #2

Nejc Kilar

  • Corona Team
  • Active Users
  • ****
  • Posts: 1334
    • View Profile
    • My personal website
@antoro,

You can't have two 1950x Threadrippers in one system. It appears to be a glitch :) Epyc is dual socket but not Threadripper.
Nejc Kilar | chaos-corona.com
Educational Content Creator | contact us

2018-05-31, 21:05:44
Reply #3

Nejc Kilar

  • Corona Team
  • Active Users
  • ****
  • Posts: 1334
    • View Profile
    • My personal website
I get extremely slow 3ds max UI menu refresh performance and was curious if users with higher frequency CPU's experience this? I wish I could just disable all of max's ui features to and revert to basic.

Dual Xeon 1660v3 @3GHz
128gb ram
quadro k4200
win10

and

Dual Xeon 2683v3 @2-2.6GHz
128gb ram
1080ti
win10

You mean the menus popping up and such? If thats the case then that appears to be random here... Obviously most stuff in 3ds Max is single threaded (with good reason, not everything can or should be multithreaded) and so it benefits from a higher clocked CPU.

It shouldn't be too big of a difference but I think it is noticeable if your cores are busy running at their all core boost (around 2-2.6ghz for most Xeon CPUs) compared to say a 4790k that runs at full load at 4ghz.

With that being said, you'll probably gain more from the fast IPR than slightly slower menus...

Oh and sometimes it seems like 3ds Max gets weird too and it gets slow in all of the weird areas possible. Just my impression :)
Nejc Kilar | chaos-corona.com
Educational Content Creator | contact us

2018-05-31, 22:10:07
Reply #4

antoro

  • Users
  • *
  • Posts: 2
    • View Profile
@antoro,

You can't have two 1950x Threadrippers in one system. It appears to be a glitch :) Epyc is dual socket but not Threadripper.

Yes, but you can have 2+ render nodes on a network, so I thought it mean two systems. Does the benchmark only run locally? If so, then it's a naming glitch.

2018-05-31, 22:49:47
Reply #5

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 9261
  • Let's move this topic, shall we?
    • View Profile
    • My Models
I think the benchmark would be pointless if you could run it in distributed rendering mode.
I'm not Corona Team member. Everything i say, is my personal opinion only.
My Models | My Videos | My Pictures