Hey everyone! I'm diving into my first build and could use some guidance. I've typically had machines built for me at work, but now I'm ready to take the plunge on my own. My main goals are to achieve 4K gaming at 120Hz and support machine learning development. My budget tops out at $3,500, but I would prefer to spend less if possible. Here are my specific questions: 1. Should I go for the 9800X3D or the 9950X3D? I'm leaning towards the 9800X3D since I think I'll be GPU bottlenecked for my tasks. Is that a good assumption? Saving money could allow me to boost my GPU spending. 2. I'm also torn between the 5080 and the 5090. I have a spare 3080 Ti, but I've read it struggles to hit 4K at 120fps in many games. Is a 5080 sufficient for future-proofing and current-gen gaming? I noted that some games rely on DLSS with the 5080 for those frame rates, and the extra VRAM in the 5090 would definitely aid in ML tasks. 3. Am I picking solid components here? Is there anything that stands out as a poor choice or anything I could optimize? 4. I plan to use the Fractal Design North XL case without the glass panel, since I'm not into RGB lighting. Did I still select high-quality components? 5. I'm also considering building a home server with Proxmox for media streaming, file serving, and running a few VMs. I believe this should be a separate machine for performance. Is it feasible to add this to my current specs? I could upgrade to 128GB RAM and the 9950X3D for more cores for VMs. Can I realistically achieve 4K at 120fps in modern games using a Windows VM with GPU passthrough? I'm open to increasing my budget if necessary. Here's my current part list for reference: [PCPartPicker Part List](https://pcpartpicker.com/list/rmrhC8).
1 Answer
1. You know your workload best, but generally, most machine learning tasks are going to lean more towards being GPU-bound. So, choosing the 9800X3D makes sense. 2. If you can snag a 5090 at MSRP, it should definitely fit into your budget. Just make sure you're mindful of costs across the board. 3. I think you might be overspending on some components like the liquid cooler and power supply. But hey, cases and coolers come down to personal preference. 4. There are cheaper alternatives if you're just looking for functionality rather than aesthetic. You might consider exploring those options!
You could consider cheaper RAM with lower latency—there's a good chance you could still fit in a 5090.