I'm planning my first PC build, mainly for gaming, programming, and running AI models locally. My current parts list is a Ryzen 9 9900X3D, Gigabyte RTX 5070 Ti Aero, 32GB of DDR5-6000 memory, and an ASUS ROG Strix B850-A motherboard. Does this combination make sense, or would you recommend different parts for better gaming performance, AI capability, value, or reliability?
4 Answers
I’d also compare the motherboard pricing and reliability before settling on that ASUS model. MSI or Gigabyte boards may offer similar features for less money. The rest of the parts look sensible, but make sure you choose a quality power supply and adequate cooling for whichever Ryzen 9000 processor you select.
Nvidia is generally the safer choice for local AI because CUDA support is broader and setup tends to be easier, especially on Windows. AMD cards can work and may offer better value for gaming, but ROCm compatibility and software support can be more troublesome. If AI is a serious priority, keeping the Nvidia card is reasonable even if another GPU offers better gaming value.
For gaming, the Ryzen 7 9800X3D may be a better fit. It uses one chiplet with 3D V-Cache, while the 9900X3D has two six-core chiplets and only one of them has the extra cache. The 9900X3D can make sense if you have heavier productivity workloads, but for a gaming-focused build the 9800X3D is usually the simpler choice.
The RTX 5070 Ti’s 16GB of VRAM could become the main limitation for local AI. Models that fit in VRAM should run well, but once you exceed that capacity, performance can drop sharply. Consider 64GB of system RAM if AI work is important, since larger models may need to spill into regular memory.
That makes sense. I’m leaning toward Nvidia because I’d rather have easier AI software support, but I’ll compare the 9800X3D and 9900X3D based on how much programming and other CPU-heavy work I actually do.