Corona Alpha4 Benchmark scene

[Jann]

Could someone please update this to Alpha 5?
Seems like A4 on which this is based doesn't work correctly on dual Xeon systems.
With A4 base, it doesn't show the speed increase you get in real life comparing to a quad i7 system.

[michaltimko]

Bought new pc today. Did some OC. Currently running on 4,4ghz stable on 1,255vcore. Temps are around 65-70stable on Noctua NDH14.
Im quite satisfied. It performs +- same as 3930k on stock clock.

For comparison, my another pc with 2600k is doing 4:33 on 4,4ghz. Those new Haswells are great at stock clock in terms of power consumption/performance ratio but overclocking is meh.
As you can see, 2years old SB performs very well at relatively "low" clock. I can push it to 4,6 easily.

So for those, who have SB and wants to upgrade to Haswell - just decide. I think its not worth. Wait on new generation or go to 3930k if you want maximum performance. But note that 4770k can perform +- same at 4,4ghz as 3930k on stock clock BUT if you push 3930k higher, power consumption is around 500W which is insane.

For example, 4770k at 4,4 eats +180W at plug. 3930k eats same at stock speed but pushing it higher can raise your electricity bills. Especially with one or two powerful GPU in rack.

BTW. here is the result :

Corona Renderer Alpha 4 benchmark scene
 Living room 100 passes
Intel(R) Core(TM) i7-4770K CPU @ 3.50GHz
Time: 0:4:14, Rays/s: 5,003,488

[Juraj]

Could someone please update this to Alpha 5?
Seems like A4 on which this is based doesn't work correctly on dual Xeon systems.
With A4 base, it doesn't show the speed increase you get in real life comparing to a quad i7 system.

Jann, what xeon system do you have ? I really need to buy few new systems, and I originally wanted 2x2687, but I see many people have some trouble with performance scaling with double socket, at which case, I would just buy few new 4930k, but I would not be so happy about that...

What performance gain do you get in "real scene" ? Thank you much in case of answer :- )

Nakama's result is utterly depressing to see double top xeons perform only as good as single 3930k, while the result should be 2.5x higher. Maybe it is only the benchmark setup, but I would like to see the performance scale decently equal in all situations..

[Ondra]

If anyone has some top of the line dualsocket machine and some time, please add me on skype, I'd like to do some testing ;)

[michaltimko]

If anyone has some top of the line dualsocket machine and some time, please add me on skype, I'd like to do some testing ;)

Throw message to Kryzstof, he have dual sys.

[Jann]

Jann, what xeon system do you have ? I really need to buy few new systems, and I originally wanted 2x2687, but I see many people have some trouble with performance scaling with double socket, at which case, I would just buy few new 4930k, but I would not be so happy about that...

What performance gain do you get in "real scene" ? Thank you much in case of answer :- )

Nakama's result is utterly depressing to see double top xeons perform only as good as single 3930k, while the result should be 2.5x higher. Maybe it is only the benchmark setup, but I would like to see the performance scale decently equal in all situations..

I have dual E5-2660 ES xeons in Z9PE-D8 WS board with 64Gb ram. On A4 benchmark they get 3.59 - about the same as overclocked 2700k@4.2Ghz we tested. In real scenes using A5 the xeons got 2-3 times better results than OC'd 2700k. So I suspect that A4 code just didn't scale as well as A5 does :)
If you're looking at retail xeons, 3930k or 4930k is definitely a better price/performance option. If you feel adventurous to get some ES cpus, then 2687w C0 could be a good investment. Or wait for the new Ivy gen xeon range with 10 and 12 cores and get ES of those :D

If anyone has some top of the line dualsocket machine and some time, please add me on skype, I'd like to do some testing ;)

You can find me on skype by ragsjv :)

[Ondra]

ok, if it scales, then there is no problem ;)

[Juraj]

Jann,

does it scale under both PT/PT and PT/HD conditions ? I've seen some conflicting info on this.

3930k is arguably better deal, but I have already 3 of those and one 2600k. I am not interested in building farms and taking care of so many boxes... also I like the idea of ECC ram and many other features of xeon based systems. Definitely won't be able to afford the new ivy bridge 12 cores, impossible price :- ) I have my limits too.

[agentdark45]

Clocking in at 3.17 on an overclocked 3930k over here! Still can't quite believe how quick Corona is!

Just wondering if anyone has a copy of the 3D studio max scene available for a closer look? I've PM'd Rawalanche too.

[foxid]

Corona Renderer Alpha 4 benchmark scene
 Living room 100 passes
Intel(R) Core(TM) i7 CPU 950 @ 3.07GHz
Time: 0:6:55, Rays/s: 3,065,401

Can I get this scene for 3ds max somewhere?
And nice render BTH.

[cerber13]

Corona Renderer Alpha 4 benchmark scene
 Living room 100 passes
Intel(R) Core(TM) i5-3570K CPU @ 3.40GHz
Time: 0:6:50, Rays/s: 3,095,308

(cpu overclock - 4.4 GHz)

[foxid]

Corona Renderer Alpha 4 benchmark scene
 Living room 100 passes
Intel(R) Core(TM) i7 CPU 970 @ 3.20GHz
Time: 0:4:40, Rays/s: 4,537,144

[johnmarston]

Hello, is there any way to run  this with an older CPU, I have Quad Core Q6600. Going upgrade soon , but I'm eager to test this renderer :)

[maru]

Corona Renderer Alpha 4 benchmark scene
 Living room 100 passes
Intel(R) Core(TM)2 Duo CPU E8400 @ 3.00GHz
Time: 0:24:10, Rays/s: 878,062

:D

It was ~19min on a laptop i3@2.4GHz.

[Aragon Speed]

Corona Renderer Alpha 4 benchmark scene
 Living room 100 passes
Intel(R) Core(TM) i7-3770K CPU @ 4.50GHz
Time: 0:4:21, Rays/s: 4,858,956