[Nucleophobe]

Hi all,

I am working on configuring a new dedicated FEA rig. The budget is 10-15k USD, but given organizational constraints, we are limited to the Dell Precision 7920 or 7960 lines. We are also license constrained for ABAQUS and can only run on a maximum of 12 threads at a time.

Given these restrictions I have been looking at the Xeon W series processors (e.g., w5-3425 with 30MB Cache, 12 cores, 24 threads, 3.2GHz to 4.6GHz). This thread seems to concur with that conclusion (thanks, flotus1), but it has been a few months. Certainly open to reconsidering Xeon Gold (e.g., 6226R in a 7920) if there are advantages.

Here is an example configuration on the Dell website: link
The above has 256GB DDR5 memory (4x64GB, 4800MHz, RDIMM ECC). We are also considering going all-out on 8x64GB DDR5, although I'm not sure that merits the extra 5k USD.

Any input greatly appreciated!

---

1. Which software do you intend to use? Mostly ABAQUS/Standard
2. Are you limited by license constraints? I.e. does your software license only allow you to run on N threads? Yes; N=12 threads max
3. What type of simulations do you want to run? And what's the maximum cell count? non-linear analyses of 10k to 10M elements
4. If there is a budget, how high is it? 15k max
5. What kind of setting are you in? Hobbyist? Student? Academic research? Engineer? Academic research
6. Where can you source your new computer? Buying a complete package from a large OEM? Assemble it yourself from parts? Are used parts an option? We are limited to the Dell Precision line, 7920 or 7960
7. Which part of the world are you from? It's cool if you don't want to tell, but since prices and availability vary depending on the region, this can sometimes be relevant. Particularly if it's not North America or Europe. USA
8. Anything else that people should know to help you better? Use case is variable; some problems will be relatively small (50k elements), but we may run 1,000s for uncertainty quantification; other cases may be very large (10M elements), e.g. for mesh refinement studies