I've never built a PC before, so I'm trying to understand whether this idea is technically possible. I want to build a SteamOS gaming PC that can also do some lightweight AI animation, such as turning a drawing of a dragon into a short clip of it breathing fire for an RPG project.
I was considering a high-end Ryzen 9 X3D processor without buying a dedicated graphics card immediately. My thought was that one part of the processor could handle normal CPU tasks while another part, perhaps one of its 3D V-Cache sections, handled graphics or AI work. The plan would be to use the integrated graphics for basic gaming at around 50 FPS, then add a powerful GPU later when prices improve.
Would a system like this be usable for basic gaming in the short term and then become a strong gaming and AI machine after adding a dedicated GPU?
4 Answers
The sensible upgrade path is to buy the CPU you actually need and pair it with an affordable dedicated GPU now, unless you’re comfortable with very basic graphics temporarily. When you upgrade later, you can replace the GPU without needing to replace the processor.
A high-end X3D CPU can be useful for gaming, but buying one specifically to avoid a GPU won’t achieve the result you’re hoping for. If local AI is only occasional, using a cloud service may also be cheaper and easier than building around a large amount of VRAM.
The CPU can technically perform many of the calculations a GPU does, but that doesn’t make it a practical substitute. Modern games and AI tools are optimized for highly parallel GPU hardware, so CPU-based rendering or generation can be orders of magnitude slower. The integrated GPU may handle basic games, but not the combination of modern gaming and AI animation at useful speeds.
Both gaming and AI generation depend heavily on GPU compute performance and VRAM. 3D V-Cache helps certain CPU workloads and some games, but it does not replace graphics memory or GPU cores.
You could technically run some AI software on the CPU, but it would usually be dramatically slower. For local image or animation generation, a dedicated GPU with plenty of VRAM is much more practical. The exact amount depends on the model and resolution, but 16 GB or more is a safer target for heavier workloads.
You can’t dedicate one CPU core complex or cache section to act as a GPU. CPUs and GPUs are built very differently: a CPU is designed for a small number of flexible, complex tasks, while a GPU has thousands of simpler parallel processors for graphics and AI workloads.
The processor’s integrated graphics can provide display output and run very lightweight games, but it isn’t intended for modern gaming or local video and image generation. A dedicated GPU would be needed for a good experience.
Even 16 GB can become limiting with larger models or high-resolution generation, so VRAM requirements can rise quickly. Lower resolutions and optimized settings may work, but CPU-only generation would still be very slow.