For scientific computing, 128GB is rarely overkill.

The short answer: Go for it. Your Dual Xeon workstation can handle it, and if you’re doing any kind of real science — simulations, big datasets, ML training — 64GB fills up fast.

I’ve seen people hit the wall at 64GB with a single decent-sized NumPy array or a modest molecular dynamics run. Once you start swapping to disk your performance tanks. Two Xeons mean you can feed them data, but they need RAM to hold it. If your workload fits in 64GB today, fine. But you’re asking, which means you’re probably close to the edge.

Buying RAM for a workstation isn’t cheap, but it’s cheaper than wasting hours waiting on swap. Pop the extra 64GB in and don’t look back. Future you will thank you when that simulation doesn’t crash at hour 23.