Hello PCMR,

I’m currently dabbling in a side project and want to be as economical as possible, both in money and space, on getting an NVidia setup. My use cases, sorted in order of importance:

  1. AI side project dabbling (yes, Transformers, but also some old school stuff like ConvNets, GANs, basically to catch up on some domain knowledge)
  2. CUDA / C++ Torch dabbling: uncovering some lower layers below PyTorch
  3. Distributed Systems learning: it would be nice to run a Kubernetes cluster without wiring physical Ethernet cables to a switch :')
  4. General dev: Being able to run an emulator and 10 tabs would be nice
  5. Gaming: nothing special, maybe Microsoft Flight Sim?

I’ve been out of the PC game since the age of GTX 9xx, and no longer consider myself an informed shopper, would love to get some recommendations, and even some guidance via Socratic method so that I can be more clear about my actual requirements. Thank you so much for your time, even if you just spent it reading :)

  • @[email protected]English
    2•1 year ago

    For a small setup, are you talking about doing an ITX build?

    Since you have a wider range of requirements or purpose for the build, I’m concerned that you might have to decide between economical and productive.

    In the cheaper side, I would suggest a Ryzen 7 5800x3D, at least 32GB of RAM, and RTX 3080 from last gen. If you can spend more money for new hardward features, I would say 7800x3D and RTX 4070.

    If you were doing heavy development work and naking money from your computer, I would suggest the 16-core Ryzen 9 7950x and 64GB of memory/RAM.