i7 5960X 8-core beast cinebench test

[Jiri.Matys]

Hi,

I have this new procesor and made some tests and one test in Cinebench R15, so anyone can test and compare.

I used this pc:
Core i7 5960X 8core, 3GHz (it overclock to 3,3GHz when all 16 threads are used actually). Overclocked on 4,2GHz, motherboard ASUS X99 Deluxe, 32GB DDR4 2400MHz, CPU Fan Noctua NH-D15, SSD drive and thats it.

Cinebench R15
At 3,3GHz: 1307 points
And at 4,2 GHz: 1630 points

I put Cinebench R15 to dropbox, so you can download it and test it. Put here your rig and score.

download

And this is a test of my old pc:

Core i7 2600K 3,4 GHz (overclocked at 4,4GHz), motherboard Gigabyte P67A, 32GB DDR3 2133 MHz, CPU FAN Noctua something, SSD drive.

3,4GHz: 614 points
4,4GHz: 772 points

[vlado]

Keep in mind that it is best to test with the particular application(s) that you will be using. Different softwares scale differently with number of CPUs/cores, and while the results should be in agreement generally, sometimes they can be misleading. What I mean is, if you are looking to benchmark rendering with V-Ray or other software, you should be testing with that software.

Best regards,
Vlado

[baastyan]

Keep in mind that it is best to test with the particular application(s) that you will be using. Different softwares scale differently with number of CPUs/cores, and while the results should be in agreement generally, sometimes they can be misleading. What I mean is, if you are looking to benchmark rendering with V-Ray or other software, you should be testing with that software.

Best regards,
Vlado

Jiri.Matys,
That also means, go to this thread :
http://forums.chaosgroup.com/showthr...OR-SCENE/page2

download the scene (post#1) and then let us now what is the score. CPU scaling in Cinebench and Vray can differ a lot

From the other hand if you compare single CPUs (as you did) then Vray and Cinebench performance should be quite similar.

My score in Cinebnech 11,5: 944
3930K @ 3,2 Ghz (no OC)

But in general you are right - 5960x is a beast

[Jiri.Matys]

But its not! I made a couple of test in max + vray and its the same. From 40 minutes with overclocked i7 2600K at 4,4GHz to 16 minutes with 5960X 4,2GHz! But I cannot provide these scenes, so this is why I wrote here cinebench tests.

[Jiri.Matys]

I made a render and put results here:

http://forums.chaosgroup.com/showthr...OR-SCENE/page2

[baastyan]

But its not! I made a couple of test in max + vray and its the same. From 40 minutes with overclocked i7 2600K at 4,4GHz to 16 minutes with 5960X 4,2GHz! But I cannot provide these scenes, so this is why I wrote here cinebench tests.

As I said: If you compare two machines with single CPU then the score might be very similar (in Cinebench and Vray).

It is getting more complex when you include dual CPU computer in this benchmark.

[Jiri.Matys]

Yes, its possible. I compare only single CPU computer, so I think that too, its similar And 8-core is really fast!

[alexyork]

Maybe it's time for Chaos to release a "VrayBench"?

[vlado]

Maybe it's time for Chaos to release a "VrayBench"?

Yeah, we actually had this done, but we ended up not releasing it. Why? Because every time newer CPUs came out with more cores, we found bottlenecks in our code which we had to go and fix or find ways around. We also constantly update the core code of V-Ray. This would mean that scores between different versions of the benchmark would not be comparable. If you benchmarked a new machine with an older version of the benchmark software, it could show poorer results than what the machine is really capable of.

Best regards,
Vlado

[alexyork]

Yeah, we actually had this done, but we ended up not releasing it. Why? Because every time newer CPUs came out with more cores, we found bottlenecks in our code which we had to go and fix or find ways around. We also constantly update the core code of V-Ray. This would mean that scores between different versions of the benchmark would not be comparable. If you benchmarked a new machine with an older version of the benchmark software, it could show poorer results than what the machine is really capable of.

Best regards,
Vlado

Interesting, and I guess it makes sense.

[Morbid Angel]

Man, if it renders faster - be happy !