Fizgig 8.9.0: Multiple GPUs for previews, shared training, and pod combos
retained text
Fizgig 8.9.0 is out. It lets your other GPUs work for your run: a second card in your PC, a laptop, another PC on your network, or a RunPod / Vast.ai pod. - **Train on two GPUs** (experimental). Each GPU trains its own copy of the LoRA on its part of the images, and they merge every 3 epochs, so you get one LoRA that learned from every image. Its likeness matched a one-GPU run in testing, and a pod with two RTX 4090s took about 40% off each epoch. On a fast network, merge every epoch for the closest result. It pays most on long runs. - **Previews that never pause your run.** Another computer renders them while your GPU keeps training. - **Need your PC back? Move the run** to a laptop, another PC or a pod, and bring it back when you want it. - **Your home GPU can help a pod.** Its checkpoints come home as they're saved, and the pod can delete itself once they're all here. - **Renting:** the Pods tab lists the RunPod cards free right now, cheapest per run first. [Vast.ai] works too. - **Also:** Krea 2 trains with about 2 GB of system RAM instead of 25, and the Look Filter finds furry, anime and drawn faces (thanks [@Murf-Code](https://github.com/Murf-Code)). Free and open source: [GitHub](https://github.com/shootthesound/Fizgig) · [Release notes](https://github.com/shootthesound/Fizgig/releases/tag/v8.9.0) · [Helpers guide](https://github.com/shootthesound/Fizgig/blob/master/docs/HELPERS.md) Update Fizgig on every computer you want to use as a helper, and restart your pods.
document fetched 2026-10-23 00:34 · canonical https://reddit.com/r/StableDiffusion/comments/1xcx77q/fizgig_890_multiple_gpus_for_previews_shared