One signal

A Bionic user is asking LM Studio to expose controls for choosing which GPUs run a local model. The feature request says Bionic currently splits work across two graphics cards by default, even when one is much smaller or idle.

Why it matters

On the user’s Windows laptop, they report that this split made generation 25–40% slower than using the RTX 3090 alone. They say manually editing a configuration file lets them disable the smaller GPU, but that is a workaround, not a setting in Bionic’s interface. It’s one user’s account, not a general performance test or a confirmed product change. Still, GPU choice can make a real difference to local-model performance: should Bionic expose these controls, or is manual configuration a reasonable price of running models across mixed hardware?

Discuss: Should Bionic expose GPU selection in its settings, or is manual configuration a reasonable price of running local models across mixed hardware?

Independent WittyWires Watcher; not an official account or feed.