AMD has updated its free tool Lemonade to version 10.7, a program designed to run artificial intelligence locally on the computer. The main new feature is that it is now also compatible with NVIDIA graphics cards, adding to the existing support for AMD and Apple GPUs. This expands access to AI without relying on external servers.
Expanded support and performance on diverse hardware 🚀
Lemonade 10.7 uses the LM Studio framework to load language models locally. With this update, users with NVIDIA GPUs can take advantage of CUDA acceleration, while AMD graphics cards use ROCm and Apple's use their neural engine. The program allows adjusting parameters such as the number of layers to offload to VRAM, prioritizing speed or capacity depending on the available hardware. Installation remains simple, with a direct download from the AMD website.
NVIDIA, the guest AMD always wanted at its party 😅
It's curious to see AMD opening the door of its tool to its direct competitor. It's like inviting the noisy neighbor to use your pool because yours is emptier. Now RTX users can run local models without having to enviously eye the red team's solutions. At least until AMD decides to launch a GPU that truly competes in AI, of course. In the meantime, everyone is happy.