i7 8700k vs E5-2690 v2
 

Hello,

I am going to get new computer to run Ansys fluent.

There are two configuration which are nearly same price

Configuration one: New Collection

* Processor : i7-7800k (6 Cores/12 Thread - 3.7GHz Base Frequency/ 4.7GHz Turbo Frequency 12 MB cache dual channel with Memory Bandwidth:41.6 GB/s)
* Ram: 2x16GB DDR4-2666 (or might DDR4-3200 MHz with overclocking)
* Storage:250 SSD + 2TB HDD
* No graphics card (gpu with processor will be used)


Configuration two: Used Dell T3610

* Processor: E5-2690 v2 (10 Cores/20 Thread - 3.0 GHz Base Frequency/ 3.6 GHz Turbo Frequency- 25 MB cache quad channel with Memory Bandwidth:59.7 GB/s)
* Ram: 4x8GB ECC DDR3-1866
* Storage:250 SSD + 2TB HDD
* Graphics card: GTX 1030

Which one will be better in simulations?
Thanks

I would go with Xeon.
1) Quad Channel, more bendwidth
2) Graphics card for postprocessing.

EDIT:
flotus is more competent when it comes to hardware and I'd go with his advice

Thank you Redo

Is Bandwidth more important than core speed?

They should perform about equally in parallel CFD, at least if you use fast memory for the I7.
The latter has the benefit of being much faster for lightly-threaded applications.
I would only consider the older Xeon if it is significantly cheaper than the new platform.
Just to clarify: we are talking about I7-8700k, not I7-7800x?

thank you flotus

One option is to go with a cheaper Coffee Lake CPU since you will be memory limited anyway. 8600k is a very reasonable choice (and with some mild overclocking it will match the 8700k in games as well). In fact, even the quad core 8350k is a reasonable choice, although it may be bad in many non-memory limited situations.



If you go cheaper CPU then you will be able to spend more on faster RAM. The scaling seems almost linear up to 3200 MHz. We have not had any reported results in the benchmarking thread with higher clocked memory, but I expect the scaling to continue since Coffee Lake @ 4.7 GHz is way too fast for dual channel memory.