If you’re looking to use AI-powered image, video, sound or text generators such as Stable Diffusion, you’re going to need a PC that can deliver speed and performance. The primary focus should be on the graphics card as this is where most of the work is done in generative AI applications. Your choice of graphics card will help to determine the other components you are going to need, as you will need a CPU which can support your GPU and enough RAM to optimise your graphics cards performance.

For generative AI and running generative models, the CPU is not the most important component, however, it will become more important if you’re also going to use your system for data science, large language modelling or manipulating large data sets. If this is the case, you may want to consider our Threadripper or Xeon workstations.  

When it comes to Generative AI, almost any 4 core, 8 thread CPU will be sufficient to run the necessary software, but depending on the graphics card you opt for, you may want to go for something a little more powerful. Any modern AMD or Intel CPU should work well for your requirements, but in particular we recommend:

Intel Core i5-14600K

With a boost clock speed of 5.3 GHz, the i5-14600K is a fast processor that supports PCIe 5.0, making for faster communication with your GPU which is extremely important for generative AI tasks. The 14600K can also support multi-GPU setups if you need to run more powerful accelerated AI training and inference.

AMD Ryzen 7 7700X

The Ryzen 7 7700X also offers fast clock speeds (up to 5.4GHz) as well as support for PCIe 5.0 and support for DDR5 memory. It also offers a great balance of single-threaded and multi-threaded performance.

When it comes to generative AI applications, your graphics card is the most important component to consider to get the best out of your system. We recommend a graphics card with at least 16GB of VRAM to ensure quick data processing and reduced latency. You should also consider memory bandwidth (Interface width and clock speeds) which will help with managing and processing large datasets.

It is worth noting that NVIDIA’s CUDA cores are better supported than AMD or Intel graphics cards currently, especially if you’re using the native Stable Diffusion and not an optimised community fork.

RAM is not a huge consideration for generative AI, with it being possible to run Stable Diffusion on just 8GB RAM, however we would recommend a minimum of at least 16GB. For optimal performance and stability, you would ideally have double the amount of RAM as you do VRAM.