Author Topic: Test performance xeon 56xx series (Benchmark)  (Read 5449 times)

2013-12-03, 14:45:22

BlessOd

  • Active Users
  • **
  • Posts: 23
    • View Profile
Couple xeons 56xx OC series in cinebench give more than 20 pts
In corona benchmark they speed comparable to 3930K @ 4,7 GHz (which gives in cinebench 14 pts)
Why?
Test this simple scene in A5 (100 passes)
or
build from 17.09 (100 passes)
3dsmax 2011-14
« Last Edit: 2013-12-03, 23:22:43 by BlessOd »
xeon X5650@4.5GHz, 16Gb, EVGA GTX780

2013-12-03, 14:51:48
Reply #1

BlessOd

  • Active Users
  • **
  • Posts: 23
    • View Profile
build from 17.09
i7-2600K@4.9
time 11.29 min
ray/s 4.565.224
« Last Edit: 2013-12-03, 21:11:58 by BlessOd »
xeon X5650@4.5GHz, 16Gb, EVGA GTX780

2013-12-03, 15:51:57
Reply #2

kregred

  • Active Users
  • **
  • Posts: 58
    • View Profile
i7 3930K@4.5
8:16

2013-12-03, 19:55:45
Reply #3

Levian

  • Active Users
  • **
  • Posts: 21
    • View Profile
i7-4930@4.2
build 16.07.13
time 8.34 min
ray/s 6.079.840
« Last Edit: 2013-12-03, 21:16:27 by Levian »

2013-12-03, 23:23:56
Reply #4

Stan_But

  • Active Users
  • **
  • Posts: 526
    • View Profile
    • https://www.behance.net/archdizs
build 17/09
i7-2600K@4.6
time 12.45 min
ray/s 4.116.911

alpha 5
i7-2600K@4.6
time 06.25 min
ray/s 4.126.281

As you see both of tests have a same noise. So maybe Fobus is right about their speed - looks like alpha 5 was couple times faster than 17/09



2013-12-04, 00:21:05
Reply #5

Ondra

  • Administrator
  • Active Users
  • *****
  • Posts: 9048
  • Turning coffee to features since 2009
    • View Profile
the performance is the same, but 17/9 has a bug that causes the real number of passes to be 2 times the selected one.
Rendering is magic.How to get minidumps for crashed/frozen 3ds Max | Sorry for short replies, brief responses = more time to develop Corona ;)

2013-12-04, 01:36:57
Reply #6

Stan_But

  • Active Users
  • **
  • Posts: 526
    • View Profile
    • https://www.behance.net/archdizs
the performance is the same, but 17/9 has a bug that causes the real number of passes to be 2 times the selected one.

Yest. But image rendered in 17/09 don't looks like in two times better than image in alpha.
It's more looks like the same quality images

Below the test in 17/09 with equal time as in alpha and It looks much more noisy than test in alpha

2013-12-04, 04:07:32
Reply #7

fobus

  • Active Users
  • **
  • Posts: 388
    • View Profile
And again it is very strange number of materials and objects in 2013-09-17 build