- Budget: $3000 (I can go over if its worth it…)
- Usage
- School:I plan to starting a masters in data science in the upcoming autumn or spring semesters
- Random number crunching: I like to dabble in PowerBI, R, Excel. It’s just fun for me.
- Photo editing: Currently using ON1 Photo Raw Max 2024, but may consider going back to Lightroom/Photoshop
- Basic gaming: Stuff like Rimworld, Cities Skyline 1 & 2, and Dota 2.
- I mainly do not want to have to redo my whole build again in 5 years. Ideally, it’ll keep for 10+. My current PC is 5 years old with a 1060-3gb that’s doing it’s very best, which is almost never good enough these days.
I’m really torn between the 7900 XTX, and the RTX 4070 Ti Super, 4080, and 4080 Super. Not sure which is going to remain a strong contender for the next 5 years. In that time, though, I’d be okay with upgrading my GPU, hence why I’d like the MOBO to have PCIe 5 compatibility.
Oh, and this is also a present to me for getting a really good job. If you have any suggestions to further personalize it, I’d appreciate them.
| Type | Item | Price |
|---|---|---|
| CPU | AMD Ryzen 9 7950X 4.5 GHz 16-Core Processor | $516.72 @ Amazon |
| CPU Cooler | ARCTIC Liquid Freezer II 420 72.8 CFM Liquid CPU Cooler | $124.99 @ Amazon |
| Motherboard | Gigabyte X670E AORUS PRO X ATX AM5 Motherboard | $299.99 @ Amazon |
| Memory | TEAMGROUP T-Create Expert 96 GB (2 x 48 GB) DDR5-6800 CL36 Memory | $374.99 @ Amazon |
| Video Card | PowerColor Hellhound Spectral Radeon RX 7900 XTX 24 GB Video Card | $999.99 @ Amazon |
| Case | Lian Li O11 Dynamic EVO XL ATX Full Tower Case | $244.99 @ Newegg |
| Power Supply | be quiet! Dark Power 13 1000 W 80+ Titanium Certified Fully Modular ATX Power Supply | $219.90 @ Amazon |
| Prices include shipping, taxes, rebates, and discounts | ||
| Total | $2781.57 | |
| Generated by PCPartPicker 2024-02-02 19:54 |
How’s this look?
I would suggest the Aorus Master board for long term UEFI updates for future CPU releases.
I don’t have specific recommendations, but IMHO, “future proofing” is not really a thing so much as upgrade paths are. You can keep some wiggle room with your PSU, try to anticipate/sync upgrades with sockets or RAM generations, or keep RAM sockets free… but old hardware gonna get old regardless. Can’t think of many CPUs that held up for that long. And DDR5 will probably be considered rather slow by then, who knows.
If you’re going into data science though, you may want to think about ever needing CUDA stuff, therefore have stick with Nvidia…?
FYI unless you’re planning to manually tune that memory to a lower speed, DDR5@6800 isn’t really feasible on AM5 at the moment. 6000MHz is the current sweet spot, and even getting that stable is much easier to achieve with 2x16GB modules.
You could likely get better average performance by rethinking your budget. Instead of $3000 and stretching it to 5-10 years, you could spend $1500 now, $500 for a gpu and/or cpu upgrade in a few years, then a complete rebuild a few years after that.
The idea is that by getting less than the latest and greatest, you’ll spend less and be more willing to upgrade when the component starts showing its age. You’ll also spend less time with components out of warranty.
One last note, I don’t see storage on your list. I recommend a 2 TB SSD at minimum, 4 TB if the prices make sense.
That’s a pretty compelling argument for going for a mid-range build. I’ll definitely consider it.
Also, I have a 2 TB nvme currently that I’ll just reuse.
You should be learning how to do data science on cloud instances. Start an aws account!
