I'm an academic researcher and part-time gamer planning my first custom PC build. I'm upgrading from an MSI GF75 Thin 9SC purchased in 2020, which still works but is beginning to feel dated. The system will mainly be used for machine learning, AI workloads, coding, and general research, with gaming as a secondary use. I'm willing to give up some gaming performance in favor of stronger overall productivity and multi-use performance.
The proposed parts are:
- CPU: Ryzen 9 9950X — €295
- GPU: Acer Predator BiFrost RX 9070 XT White — €799.90
- Motherboard: ROG Strix X870E-H Gaming WiFi 7 — €247.99
- Memory: Crucial Pro 32 GB DDR5-6000 — €389.99
- SSD: Crucial E100 2 TB — €199.99
- Cooler: Thermalright Aqua Elite 360 V6 White — €47.07
- Power supply: MSI MAG A850GLS PCIe 5 — €94.95
- Case: MSI MAG PANO 130R PZ White — €64.96
Does this look balanced for my workloads, and are there any parts I should change?
3 Answers
You may not need a 16-core processor for ordinary coding, ML development, and AI work unless you’ll be running several virtual machines, containers, or heavily threaded workloads at once. A high-end Intel Core Ultra 7 270K-class processor could be a better fit. That said, €295 for the Ryzen 9 9950X sounds like an unusually good deal, so keeping it is reasonable if the price is legitimate and you value the extra cores.
For local AI and machine learning, I’d seriously consider Nvidia instead of the RX 9070 XT. AMD cards are excellent for gaming, but Nvidia’s CUDA ecosystem and software support are still much better for many AI tools and models. The Radeon would be a great gaming choice, but Nvidia is likely the smoother option for your specific work.
I agree. ROCm support has improved, especially on Linux, but compatible models are often still slower or require more setup than their CUDA equivalents.
That makes sense. My main hesitation is the price, since decent Nvidia cards seem considerably more expensive.
The X870E motherboard seems excessive unless you specifically need its extra connectivity, expansion slots, or storage support. You could save money with a less expensive board and put the difference toward a better SSD. The Crucial E100 is a very budget-oriented drive with low endurance, so it’s not a great choice for heavy workstation workloads or large amounts of data.
I chose the board mainly for its additional NVMe and SATA options, since I work with a lot of data and expect to add more storage later. I’ll look for a higher-quality SSD, though.
Several people have warned me about the SSD, so I’ll replace that first. The 9950X is discounted heavily, which is why I’m comfortable choosing the higher-end CPU even if I don’t strictly need all of its cores.