Buying a custom PC for running machine learning algorithms - Ubuntu compatibility

ScoFot

New member
Hi all,

New to the hardware space (filthy Mac user) but looking to purchase a high-spec PC to run some Python-based machine learning algorithms on. I'm just a little worried about driver compatibility with the configuration I'm looking at and I was hoping someone more expert with Linux could confirm if we are likely to run into any trouble with the following configuration. The partition we will migrate over is currently running Ubuntu 20 but I'm happy to upgrade to 22 or higher if that's the least troublesome..

I have tried Googling compatibility for the components below but I'm not getting anywhere (certainly no info that I fully understand). I'm guessing the issues mainly will be with: motherboard, GPU and CPU. Sound card isn't important as we won't be using speakers but I'll note it below anyway.

Anyway, it's a high-spec configuration of the Fusion Goliath with the following:

* An AMD Rizen 7950X3D stuck to an ASUS ROG Crosshair X760E Hero
• CPU cooling with a Corsair iCUE H100i cooler
• An extra 120mm Thermaltake ToughFan
• 64GB Corsair Vengeance DDR5 RGB 5600 (I'm assuming RAM isn't usually an issue with Linux)
• A 24GB RTX 4090
• 2x 2TB Samsung 990 Pro M.2 NVMe (also assuming M.2 aren't usually an issue)
• Corsair 1000W RMx Series PSU (probably overkill)
• Sound card is just the standard onboard 6 channel (I'm assuming this is just part of the motherboard?)

I'd be really grateful for any advice the Linux experts here can give.

Side note: I know we'd be better off with a ThreadRipper and compatible motherboard for full 16x2 PCIE lanes and sticking 2 of the 4090 cards on it but I'm restricted by a budget of £3k (+VAT). Again, happy to be advised of an alternative configuration if we are heading down the wrong path.

Thanks in advance.
 
Top