GPU Slider

GPU Slider is the Flux Continuum pack's control for how many GPUs a task should use - a 1–4 slider that outputs an integer. It's part of the pack's slider family, and it exists because the workflow's roadmap explicitly lists multi-GPU support as coming soon. Right now, for a single-GPU setup, this node mostly sits there returning 1. That's not a bug; it's a control surface waiting for its feature.

How it works

Like every slider in the pack, the mechanism is a rounded pass-through: FLOAT slider in, INT out, source is int(round(value)). The value ranges 1–4 with a default of 1. What consumes it depends on the workflow's state - the distributed/multi-GPU path (which touches ComfyUI_NetDist, given the pack's BooleanToEnabled node) is where a GPU count actually matters. Until that lands, feeding the value anywhere will just... pass the number along, and a single-GPU ComfyUI ignores it.

Inputs and outputs

• value - FLOAT, 1 to 4, default 1, step 1.

Output is one INT. The count of GPUs to target.

Installing it

Ships in robertvoy/ComfyUI-Flux-Continuum:

cd ComfyUI/custom_nodes
git clone https://github.com/robertvoy/ComfyUI-Flux-Continuum

or ComfyUI Manager ("Flux Continuum"), restart. No dependencies.

Honest take

Here's the truth: on the current stable workflow, this slider is a placeholder. It's the front panel control for a distributed-processing feature that's announced but not shipped. If you're on one GPU - which is most people - leave it at 1 and forget it's there. If you're already running a multi-GPU or multi-machine setup and the pack has activated that path, this is where you'd set the count, but expect to configure the actual distribution elsewhere (NetDist or your own multi-GPU nodes). Don't buy a second GPU expecting this slider to do anything today.