For example, she works at Stable Diffusion
At the moment, artificial intelligence is practically not involved in Windows, except for the integration of the GPT language model. However, firstly, there are already AMD processors with a dedicated unit for working with AI, and similar ones are being prepared by Intel, secondly, according to rumors, Windows 12 will be built around AI. And now it is reported that ordinary users will get their hands on accelerators for AI from Nvidia, only in software form.
The power of Nvidia AI accelerators is now on Windows.
The company announced that its TensorRT-LLM library, which improves inference performance for the latest large AI language models, is now available for Windows. TensorRT-LLM allows you to significantly speed up your work with AI.
In particular, in Stable Diffusion in the popular web interface of the Auto1111 distribution, Nvidia’s development speeds up AI by almost half. As a result, with this improvement, the RTX 4090 is seven times faster than the Apple M2 Ultra.
This will only work on a GeForce RTX video card, but for Nvidia, this is quite logical. The company is thus “transforming” its gaming adapters into accelerators for AI.
