Intel Xeon E5-2660v2 or Intel Core i9-7920X for the parallel RANS

cfdhelp · December 3, 2017, 13:32

Dear friends,

I would like to ask you what processor is better for the calculations.
Intel Xeon E5-2660v2
Intel Core i9-7920X
Both CPU have 10 cores ...
What is important parameter ?
Is it possible to find some benchmarks for matrix solvers ?
I want to use for some simulations.

Regards,
Ivan

ztdep · December 4, 2017, 05:28

find intel linkpack to benchmark the cpu.
the bandwith between each cores and memories are the most important parameter to have better performance.

ztdep · December 4, 2017, 05:36

find intel linkpack to benchmark the cpu.
the bandwith between each cores and memories are the most important parameter to have better performance.

FMDenaro · December 4, 2017, 08:29

I am not sure if still updated but in netlib.org there was a performance database file based on linpack test.

naffrancois · December 4, 2017, 10:16

I9-7920X is a 12 core cpu not 10. These CPUs are from a very different generation, four years between them is quite a lot (xeon v2 is already quite "old"). Intel released the new xeon family (Skylake) a few months ago which would correspond to xeon v5. For a given number of cores, you may look at the frequency, the higher the better (here the I9 is 2.9Ghz whereas the xeon is 2.2Ghz), the L3 cache memory (the higher the better), the number of memory channels, the memory bandwidth.

Honestly I expect the I9 to outperform the xeon by far, so if I had to choose I would go for the machine with I9

flotus1 · December 5, 2017, 05:11

Both of these can be a good choice if you play your cards right.
Intels old Socket 2011 processors come cheap and use inexpensive DDR3 reg ECC. So you can pick up two of these CPUs, a dual-socket motherboard and 128GB of RAM for less than 1000$. It will probably be even faster than The I9 in highly parallelized workloads.
The I9 on the other hand needs expensive DDR4 memory but will outperform the older Xeon CPUs in single-threaded and lightly-threaded applications. E.g. grid generation and post-processing.
Both are possible choices for a CFD workstation, depending on the use-case ("I want to use for some simulations" is not accurate enough) and the budget. I am just wondering why you are asking specifically about these two CPUs...

December 3, 2017, 13:32	Intel Xeon E5-2660v2 or Intel Core i9-7920X for the parallel RANS	#1
cfdhelp New Member Ivan Join Date: Nov 2012 Location: Czech Republic Posts: 22 Rep Power: 13	Dear friends, I would like to ask you what processor is better for the calculations. Intel Xeon E5-2660v2 Intel Core i9-7920X Both CPU have 10 cores ... What is important parameter ? Is it possible to find some benchmarks for matrix solvers ? I want to use for some simulations. Regards, Ivan

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
simpleFoam parallel	AndrewMortimer	OpenFOAM Running, Solving & CFD	12	August 7, 2015 18:45
Superlinear speedup in OpenFOAM 13	msrinath80	OpenFOAM Running, Solving & CFD	18	March 3, 2015 05:36
Performance of Xeon X56-- 6 core	hvem10	Hardware	4	August 3, 2010 10:55
OpenFOAM 13 Intel quadcore parallel results	msrinath80	OpenFOAM Running, Solving & CFD	13	February 5, 2008 05:26
AMD X2 & INTEL core 2 are compatible for parallel?	nikolas	FLUENT	0	October 5, 2006 06:49

December 4, 2017, 05:28		#2
ztdep Senior Member p ding Join Date: Mar 2009 Posts: 427 Rep Power: 19	find intel linkpack to benchmark the cpu. the bandwith between each cores and memories are the most important parameter to have better performance.

December 4, 2017, 05:36		#3
ztdep Senior Member p ding Join Date: Mar 2009 Posts: 427 Rep Power: 19	find intel linkpack to benchmark the cpu. the bandwith between each cores and memories are the most important parameter to have better performance.

December 4, 2017, 08:29		#4
FMDenaro Senior Member Filippo Maria Denaro Join Date: Jul 2010 Posts: 6,768 Rep Power: 71	I am not sure if still updated but in netlib.org there was a performance database file based on linpack test.

December 4, 2017, 10:16		#5
naffrancois Senior Member Join Date: Oct 2011 Posts: 239 Rep Power: 16	I9-7920X is a 12 core cpu not 10. These CPUs are from a very different generation, four years between them is quite a lot (xeon v2 is already quite "old"). Intel released the new xeon family (Skylake) a few months ago which would correspond to xeon v5. For a given number of cores, you may look at the frequency, the higher the better (here the I9 is 2.9Ghz whereas the xeon is 2.2Ghz), the L3 cache memory (the higher the better), the number of memory channels, the memory bandwidth. Honestly I expect the I9 to outperform the xeon by far, so if I had to choose I would go for the machine with I9

December 5, 2017, 05:11		#6
flotus1 Super Moderator Alex Join Date: Jun 2012 Location: Germany Posts: 3,399 Rep Power: 46	Both of these can be a good choice if you play your cards right. Intels old Socket 2011 processors come cheap and use inexpensive DDR3 reg ECC. So you can pick up two of these CPUs, a dual-socket motherboard and 128GB of RAM for less than 1000$. It will probably be even faster than The I9 in highly parallelized workloads. The I9 on the other hand needs expensive DDR4 memory but will outperform the older Xeon CPUs in single-threaded and lightly-threaded applications. E.g. grid generation and post-processing. Both are possible choices for a CFD workstation, depending on the use-case ("I want to use for some simulations" is not accurate enough) and the budget. I am just wondering why you are asking specifically about these two CPUs...