Stable Diffusion, a state of art AI art generation tool. The results produced by it are fast, intuitive, and pretty impressive results. You can also run your Stable Diffusion locally on your PC. Before that, you should check if your PC is up to spec that is capable enough to run Stable Diffusion without any problem.
The basic requirement to run Stable Diffusion locally on your PC is
- 16 GB of RAM
- Nvidia Graphic with at least 10GB of VRAM
The single most needed requirement for running Stable Diffusion is a graphic card (GPU). Stable Diffusion runs exclusively on GPU.
There are several forks and iteration has lowered the Stable Diffusion requirements with newer options. The best result is possible on powerful graphic cards like Nvidia RTX 4080 AND RTX 4090 with their 16GB and 24GB of VRAM each.
Table of Contents
Stable Diffusion Nvidia Requirement?
All the AI code was built on Nvidia CUDA. Get the biggest VRAM you can afford. 12GB 3060ti will get you farther than a 10GB 3080.
16 GB will help you carry your most image-generating and art workflow and 32GB is a better comfortable starting place. 8GB is not enough.
If you are getting in to training and developing models, you are gonna find things work better with more VRAM.
- RTX 3060, RTX 3060 Ti, RTX 3070 Ti , RTX 3080 (12GB), RTX 3080 Ti, RTX 3090 or RTX 3090 Ti
- RTX 4090, RTX 4080
If you are buying a PC with Stable Diffusion in mind. Always go with the latest GPU with more than 12 GB of memory. Never compromise on this, if you are working on training and developing stable diffusion models.
Does Stable Diffusion run on AMD GPU?
Stable Diffusion run on all AMD GPUs from RX470 and above are working fine. You will see better performance on AMG GPU with at least 8GB of VRAM
Working Working AMD GPUs
- Radeon RX 66XX/67XX/68XX (XT and non-XT) GPUs
- VEGA 56/54
- Radeon VII
- Radeon RX 6600/6650 (XT and non-XT) and RX6700S Mobile GPU
- Radeon 5500/5600/5700(XT)
- 8GB Models of Radeon RX 470/480/570/580/590 ( have to reduce the batch size to 1 or lower resolutions)
You can check out this Reddit thread for the complete information:
Does Stable Diffusion work on Intel Graphics Cards (GPU) or Intel CPUs?
Intel Graphics cards are not officially supported by Stable Diffusion. But there are forks that support Intel Graphic cards.
- stable_diffusion.openvino supports text-to-image generation using Stable Diffusion on Intel CPU or GPU
Does Stable Diffusion work on Apple M1 Processors?
Yes, you can run Stable Diffusion on Apple M1 Processor but you need to use a specific fork that supports Apple M1 Processors.
- InvokeAI is currently one of the most popular Apple M1 Chips. It requires at least 12GB of memory and 12GB of installation space.
- DiffusionBee works well with Apple. It requires macOS 12.5.1 and Machine with Apple silicon is recommended. Diffusion Bee works with Intel-based machines and it is very slow compared with M1 machines.
- CHARL-E is a Stable Diffusion on your mac with 1 Click. M1 or M2 required.
How to run Stable Diffusion without a GPU?
You can run Stable Diffusion through DreamStudio, without any need for hardware requirements.
To run Stable Diffusion without a dedicated graphics card (GPU). You have to use one of the forks that especially runs Stable Diffusion on the CPU.
- stable-diffusion-cpuonly doesn’t require high-end graphics and runs exclusively on your CPU.
- Stable Diffusion on OpenVino gives an indication of the speed that can be achieved on the CPU. Supported CPUs are AMD Ryzen, Intel Core 17, Intel Xeon Gold 6154 CPU
These are options who really want to try Stable Diffusion without a GPU.
Conclusion
You will need a PC with a modern AMD or Intel processor, 16 gigabytes of RAM, and an NVIDIA RTX GPU with 8 gigabytes of memory. Minimum 10 GB of free space available.
- CPU: A modern AMD or Intel CPU
- RAM: A minimum of 16 gigabytes of RAM
- Storage: Minimum 10 GB of free space. 1TB of space is better
- GPU. GeForce RTX GPU with a minimum of 10 GB of memory.