High-end workstation

Drunken_SheeP · March 30, 2017, 08:40

Hello,

I am still looking to build a powerful dual CPU workstation to use for my icing simulations. I will be working on mainly 2D problems but also 3D with up to max 3million cells. I will be using ANSYS software (FENSAP-ICE) which scales very similarly to FLUENT. I have a total of 80 licenses available but they are limited to 16 cores for each task. We will probably go for 128GB (8x16GB) 2400MHz DDR4 RDIMM ECC working memory.

But I am still wondering what CPU I should prefer. E5-2687W v4 (12C @ 3GHz) seems to be the economically best solution but E5-2697A v4 (16C @ 2.6GHz) would offer a bit more performance on paper. Any thoughts on how much performance increase I could expect and if it's really worth the additional money?

flotus1 · March 30, 2017, 18:24

It is probably not worth the extra money. You might want to have a look at this: Socket 2011-3 processors - an overwiew
With 3 million cells maximum, scaling on large numbers of cores will be less than ideal. So go for the CPU with less faster cores. Better spend the extra money on more fast storage. Btw: I hardly think that you will have any use for that amount of RAM with such small simulation models.
Just don't overthink the purchase. You will never make up for the time lost delaying the decision for a few more weeks, no matter how fast the workstation is.

evcelica · March 31, 2017, 16:38

Well said flotus1.
The 2687W will be faster if you are limited to 16 cores per computation. This is due to the higher frequency: 3GHz vs 2.6GHz.

March 30, 2017, 08:40	High-end workstation	#1
Drunken_SheeP New Member asdfad Join Date: Jul 2012 Location: Norway/Germany Posts: 13 Rep Power: 13	Hello, I am still looking to build a powerful dual CPU workstation to use for my icing simulations. I will be working on mainly 2D problems but also 3D with up to max 3million cells. I will be using ANSYS software (FENSAP-ICE) which scales very similarly to FLUENT. I have a total of 80 licenses available but they are limited to 16 cores for each task. We will probably go for 128GB (8x16GB) 2400MHz DDR4 RDIMM ECC working memory. But I am still wondering what CPU I should prefer. E5-2687W v4 (12C @ 3GHz) seems to be the economically best solution but E5-2697A v4 (16C @ 2.6GHz) would offer a bit more performance on paper. Any thoughts on how much performance increase I could expect and if it's really worth the additional money?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Setting the height of the stream in the free channel	kevinmccartin	CFX	12	October 13, 2022 21:43
CFD analaysis of Pelton turbine	amodpanthee	CFX	31	April 19, 2018 18:02
mass flow in is not equal to mass flow out	saii	CFX	12	March 19, 2018 05:21
Problem with an old Simulation	FrankW	CFX	3	February 8, 2016 04:28
RPM in Wind Turbine	Pankaj	CFX	9	November 23, 2009 04:05

March 30, 2017, 18:24		#2
flotus1 Super Moderator Alex Join Date: Jun 2012 Location: Germany Posts: 3,399 Rep Power: 46	It is probably not worth the extra money. You might want to have a look at this: Socket 2011-3 processors - an overwiew With 3 million cells maximum, scaling on large numbers of cores will be less than ideal. So go for the CPU with less faster cores. Better spend the extra money on more fast storage. Btw: I hardly think that you will have any use for that amount of RAM with such small simulation models. Just don't overthink the purchase. You will never make up for the time lost delaying the decision for a few more weeks, no matter how fast the workstation is.

March 31, 2017, 16:38		#3
evcelica Senior Member Erik Join Date: Feb 2011 Location: Earth (Land portion) Posts: 1,167 Rep Power: 23	Well said flotus1. The 2687W will be faster if you are limited to 16 cores per computation. This is due to the higher frequency: 3GHz vs 2.6GHz.