Multi-GPU in ComfyUI (local, remote and cloud GPUs)

ComfyUI Distributed Extension

I've been working on this extension to solve a problem that's frustrated me for months - having multiple GPUs but only being able to use one at a time in ComfyUI AND being user-friendly.

---

What it does:

• Local workers: Use multiple GPUs in the same machine

• Remote workers: Harness GPU power from other computers on your network

• Cloud Workers: GPUs hosted on a cloud service like Runpod, accessible via secure tunnels

• Parallel processing: Generate multiple variations simultaneously

• Distributed upscaling: Split large upscale jobs across multiple GPUs

---

Using Cloud Workers?

Join Runpod with this link and unlock a special bonus: https://get.runpod.io/0bw29uf3ug0p

---

Real-world performance:

• Ultimate SD Upscaler with 4 GPUs: before 23s -> after 7s

Easily convert any workflow:

1. Add Distributed Seed node → connect to sampler

2. Add Distributed Collector → after VAE decode

3. Enable workers in the panel

4. Watch all your GPUs finally work together!

Upscaling

• Just replace the Ultimate SD Upscaler node with the Ultimate SD Upscaler Distributed node.

I've been using it across 2 machines (7 GPUs total) and it's been rock solid.

---

GitHub: https://github.com/robertvoy/ComfyUI-Distributed

---

📚 Resources:

🔗 WAN 2.2 Workflow

🔗 GitHub

📺 Watch "Deploy Cloud Worker on RunPod" Tutorial

📺 Watch the most recent update video

---

Happy to answer questions about setup or share more technical details!