Author Topic: Corona Alpha4 Benchmark scene  (Read 306028 times)

2014-10-21, 02:20:16
Reply #390

Asasay

  • Users
  • *
  • Posts: 2
    • View Profile
Why it took me so long?
« Last Edit: 2014-10-21, 02:37:05 by Asasay »

2014-10-21, 03:34:14
Reply #391

juang3d

  • Active Users
  • **
  • Posts: 636
    • View Profile
Ok this became absolutely pointless discussion I don't think you're even reading what I write.

mmmm Maybe I'm not understanding what you mean, I'm assure you that I'm reading what you write with lots of interest, can you explain your point in a different way?

I just don't understand your point because I think you are compairing the 5960 against a 2 Cpu workstation instead of compairing it with it's lower brother, I compare the 5820 vs the 5960 and as an extension with any 2 CPU workstation, I left the 2cpu workstation out of the equation becuase it's cost is way too high in relation with the performance gain, that's why I don't understand why to compare the 5960 vs a 2cpu workstation, keep in mind that I've discarded 2cpu workstations some years ago because it's maintenance is expensive and worse than a single cpu workstation.

I'm not sure if this explains why I answered you what I answered, but maybe I understood you wrong, sorry about that if that is the case.

Cheers.

2014-10-21, 03:49:37
Reply #392

Juraj Talcik

  • Active Users
  • **
  • Posts: 3618
  • Tinkering away
    • View Profile
    • studio website
2p system doesn't take more maintenance than single cpu system these days. I have 2 dual xeons systems, with identical parts as my i7 workstations, except for motheboard and cpus obviously. Same memory, same ssd, same coolers, but that's beside the point.
Same maintenance (none).

The comparison of 5960X to 2p system was that it's almost getting to that sort of performance in single box, with still fraction of price compared to 2p system bringing it to affordable level for people who crave maximum performance in single box.

Of course you can buy 1.5 PCs worth with 5820K, you can also buy 2-3 PCs worth with 4770k, or 5 PCs of i5..or..always find the best value/performance. Except for that isn't the sole factor of choice for everyone. If it was, no one would be buying BMW car if they get almost the same (70perc.) performance for only half the price (Opel,Citroen,whatever there is) and buy two of them, one for wife or sister too. Performance/Value is only single metric, to some the most important, to some completely irrelevant.
Single, powerful workstation will still be valuable to lot of people, either enthusiasts or professionals. And now they can get it "cheap".

5820k is crazy deal from intel (finally), and great choice if you want to build semi-budget renderfarm (because the value/performance will be still better outside of LG2011 platform) or workstation. But that doesn't make it some sort of winner over 5960X. It's different CPU, for different people.
talcikdemovicova.com  Website and blog
be.net/jurajtalcik   Our studio Behance portfolio
Instagram   Our studio Instagram, managed by Veronika

2014-10-21, 21:59:00
Reply #393

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 5724
  • Let's move this topic, shall we?
    • View Profile
    • My Models
Why it took me so long?
You need to render with alpha 4 benchmark utility if you want to compare render times. Look at first post of this topic.
I'm not Corona Team member. Everything i say, is my personal opinion only.

2014-10-21, 22:08:19
Reply #394

krimson

  • Active Users
  • **
  • Posts: 34
    • View Profile
Corona Renderer Alpha 4 benchmark scene
 Living room 100 passes
Intel(R) Xeon(R) CPU E5620 @ 2.40GHz
Time: 0:4:45, Rays/s: 4,462,954

dual processors


2014-10-21, 23:16:39
Reply #395

Asasay

  • Users
  • *
  • Posts: 2
    • View Profile
Corona Renderer Alpha 4 benchmark scene
Living room 100 passes
Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz @ 4.7GHz on air
Time: 0:4:2, Rays/s: 5,240,729

2014-10-22, 11:33:22
Reply #396

SHD

  • Active Users
  • **
  • Posts: 134
    • View Profile
    • https://www.instagram.com/xshd/
Quote
somebody tested 5930k?

In the previous page (25) you have some info about it.

Cheers.

yes, i saw it, but much more interesting for me is corona benchmark scene ;)


SHD look better because i posted the corona benchmark scene too ;)

i saw for 5960x, but not for 5930k

2014-10-22, 20:14:10
Reply #397

Marvey

  • Active Users
  • **
  • Posts: 184
    • View Profile


i saw for 5960x, but not for 5930k

i am sorry SHD my mistake. i didn´t tested 5930k but the 5960x.

2014-10-23, 11:09:16
Reply #398

SHD

  • Active Users
  • **
  • Posts: 134
    • View Profile
    • https://www.instagram.com/xshd/


i saw for 5960x, but not for 5930k

i am sorry SHD my mistake. i didn´t tested 5930k but the 5960x.

No problem ;)

so in this moment really interesting for me to know render time in corona benchmark 5820k (@4.2) and 5930k (@4.2)
if somebody has this "stone"  will be very interesting to know

2014-10-23, 15:09:40
Reply #399

juang3d

  • Active Users
  • **
  • Posts: 636
    • View Profile
As promised, here are the results for the 5820k, it's a truly amazing CPU.

OC's at 4.4 just with multiplier, no other thing touched so nothing is beeing stressed, I'm sure I can enable a higher OC but since it's a render node and it's not going to vie overviewed all day I don't want to stress it out too much.
The 5930k it's not worth the money unless you are going to use a lot of GPU's for real time rendering, not for offline rendering (so, not for Octane or iRay, but for videogames and real time engines) and that is because it's extra PCIe lanes AFAIK, but in the end it's not worth the money, IMO you go with the 5820k or the 5960k, this last one it's not worth either IMHO, with the extra money you have to pay for that CPU you can nearly acquire another 5820k :D

Corona Renderer Alpha 4 benchmark scene
 Living room 100 passes
Intel(R) Core(TM) i7-5820K CPU @ 3.30GHz
Time: 0:2:58, Rays/s: 7,126,987

BTW still waiting the license clearance for this node :)

Cheers!

2014-10-23, 15:38:59
Reply #400

SHD

  • Active Users
  • **
  • Posts: 134
    • View Profile
    • https://www.instagram.com/xshd/
very nice!
so in this moment
5960x (@4.4) = 2 min 14 sec
5820k (@4.4) = 2 min 58 sec

2014-10-23, 16:03:13
Reply #401

Jann

  • Active Users
  • **
  • Posts: 142
    • View Profile
very nice!
so in this moment
5960x (@4.4) = 2 min 14 sec
5820k (@4.4) = 2 min 58 sec
Please remember the benchmark scene is not accurate.
I do hope that with 1.0 we finally get a new one :P

2014-10-23, 16:05:28
Reply #402

juang3d

  • Active Users
  • **
  • Posts: 636
    • View Profile
Why is not? We are comparing different CPU in the same circumstances...

2014-10-23, 16:33:49
Reply #403

Juraj Talcik

  • Active Users
  • **
  • Posts: 3618
  • Tinkering away
    • View Profile
    • studio website
Why is not? We are comparing different CPU in the same circumstances...

That doesn't make it accurate if the benchmark itself doesn't scale well, and it doesn't. Not only is it old (A4) but the way the HD Cache is set makes it scale (or basically ignore) badly on high-core CPUs, or double CPUs.
It's extremely confusing and should probably be already addressed because I am getting emails from people asking me why it doesn't work with their dual-xeons for example.

But it just goes on and on and everybody posts the same score all over and over like they would achieve something different. Like how many times for example does 4770k needs to be posted ? 50 times ?

Btw, the SHD's post basically shows nonetheless that 5960X is fantastic performer.
talcikdemovicova.com  Website and blog
be.net/jurajtalcik   Our studio Behance portfolio
Instagram   Our studio Instagram, managed by Veronika

2014-10-23, 17:58:02
Reply #404

juang3d

  • Active Users
  • **
  • Posts: 636
    • View Profile
Indeed the 5960X is a great performer, but for me (and it's important to say that this is from my point of view, others can think differently, like you Juraj :) ) it's not worth the price, for a bit more of the difference of the price you can assemble another 5820k and have double performance, so to speak 14000 in benchmark instead of 9000 that you get with the 5960.
I think the 5960X may be worth if ti costs 600€ instead of 995.

I have the 5820k performing in a current real project, so I have consecutive frames from my farm, I post here what I get from my farm, daily build form the 19th october:

i7-2600k@4.3Ghz : 28min 10secs
i7-3770k@4.2Ghz: 25min 32secs
i7-5820k@4.4Ghz: 15min 06secs

This is from my current real project, similar to the one I have published in the gallery forum.

Cheers!

EDIT: BTW about having different scores of the same CPU, that is great, that give us different samples so we can have a more accurate idea of the medium performance that CPU can give us, in Anandtech they received two 5960X for OC testing, one of them was able to be OC'd up to 4.5Ghz while the second one achieved only 4.2Ghz, so more samples more accuracy :)
« Last Edit: 2014-10-23, 18:03:47 by juang3d »