Page 1 of 1

Planning to invest in 2xCPU model

Posted: Tue Feb 14, 2017 10:00 pm
by RuiManuel
Hello.

Right now I am using an Intel i7 7700k CPU and a geforce nvidia 1080.

Rendering simple scenes (an object and a splash for instance) are too slow, about 5/6 hours for rendering 50 frames (no dyverso used).

I am thinking about buying a 2 CPU capable motherboard and use 2x Intel i7 7700k CPU instead of just one.

My question is : am I spending the money well doing that? Will I get real differences in my rendering times?

Thanks.

Re: Planning to invest in 2xCPU model

Posted: Wed Feb 15, 2017 8:22 am
by ChristianZ
Hi

yes, there is some common wisdom, at least what I have heard on different forums:
1) For rendering images in a 3D software the best is to have many cores and multiple CPUs, because this task can be split between processors.
2) If you do a lot of physics simulations such as fluids, fire, breaking things, cloth etc. It is preferable to have a high clock speed, and this usually comes with few cores because of the heat produced. The reason is, that many software packages use only one core alone for such tasks and these calculations are more difficult to split between many cores without loosing speed.

Re: Planning to invest in 2xCPU model

Posted: Wed Feb 15, 2017 9:08 am
by Evan.Qiu
I just received my new PC (two E5 2683 V3 CPUs (totally 28 cores, 56 threads, 2.0GHz), but seems the simulation is much slower than my another I7 4790K CPU (4 cores, 8 threads, 4.0GHz), maybe because of the low CPU clock speed
And as I know there's no motherboard which supports multi I7 series CPUs´╝îthe multi CPU capable motherboard only supports server CPU like E3, E5

Re: Planning to invest in 2xCPU model

Posted: Wed Feb 15, 2017 9:47 am
by ChristianZ
Your new machine certainly is still a very good machine for rendering all sorts of graphics.

I don't know what would be the "perfect" simulation PC. Maybe something involving overclocking the CPU.
However, the future is with the GPU hardware, it seems. The capability of the GPU hardware might be easier to predict by the number of parallel cores on the card. Some benchmarks on the Nvidia page http://www.nvidia.com/object/computatio ... amics.html