Discussion Groups

Simulation CFD

New Member
1 Posts
0 Kudos
Registered: ‎10-24-2012
Accepted Solution


284 Views, 2 Replies
10-24-2012 12:59 AM

Hi all


I am looking to upgrade my PC to increase the speed I can process CFD models.  I currently have an old 32 bit dual core Intel Xenon with 4 GB RAM.  I am unsure as to what aspect of a new machine will give me the best improvement – i.e. processor speed or amount of RAM.


A possible machine I am looking at is a new quad core or 64 bit Dual Six Core model with 32-128 GB RAM.  Can anyone advise on how much faster this new machine would perform when compared with the old machine, or advise on a better option?  Would more RAM speed the simulation or would that just only limit the number of cells that can be simulated and not make any appreciable difference to the simulation time?


A typical model would be 4 meshes with a total of 2 million cells, although using more meshes would be possible to make better use of multi-core processors.


Any response would be much appreciated.



Product Support
1,062 Posts
149 Kudos
Registered: ‎08-25-2011


10-24-2012 02:55 AM in reply to: Mark1987

Hi Mark,


Here is a link to the recommended spec, although if you can afford more, go for it.


The RAM will allow you to mesh larger models rather than have a large impact on runtimes. Bear in mind that CFD uses 2^n cores so you will see the largest speedups at 4, or 8 cores. If you have 6, 2 will be left unused. If you can aim for more than 8, brilliant.

Jon Wilde
Product Support
310 Posts
54 Kudos
Registered: ‎08-31-2011


11-01-2012 07:04 PM in reply to: Mark1987

For reference


RAM will dictate the size of the model that you can run

the number of cores and clock speed of those cores will dictate how fast your model solves


At this time CFD uses cores in 2^n fashion (2, 4, 8, 16, etc)

With that being said, I would advise stepping down to a dual-quad core (at a higher clock speed) or now with the latest Intel chips purchasing a single 8core (or even a dual 8Core) machine.


For proper comparisons keep the clock speed at about the same as what you are coming from.

For example its harder to compare a 4core @ 3.6Ghz to 16cores @ 2Ghz, than comparing 4cores @2.8Ghz to 16cores @2.8Ghz.



For ram based on the meshed element count (not the Approximate Count on the Mesh task dialog) we use ~2-2.5GB for every 1Million elements.

So if you have a 1M element model whether you have 8GB or 128GB, it will run at the same pace. Ensure taht you have enough ram to run the solver, interface and any other applications you might use at the same time (CAD, email, other daily work apps).