AMD from the first Imaging AI with Stable Diffusion 3.0 Medium Nervous processing units XDNA 2 NPUS Unveiled that Local On laptops AI Consulta It is implemeed.
According to reports, this model is offline without the need for iernet or cloud services and is capable of producing quality images and in collaboration with the company. Stability ai Developed.
AMD Unveiling of Stable Diffusion 3.0 Medium Special XDNA 2
This model Artificial Ielligence Convert Text to Image Based on the Stable Diffusion 3.0 Medium that is specifically for the format BF16 Optimized and designed to run devices with XDNA 2 neural processing units. The model first produces 1024 × 1024 resolution and then uses a Inner route in NPU Upgrades these images to 2048 x 2048 resolution to provide a 4 -megapixel final output. AMD emphasizes that the quality of the images produced for Pri And professional uses are good.
To run this model, users to a computer with the AMD Ryzen AI 300 or Ryzen AI Max+, XDNA 2 Neural Processing Unit with Minimum Power 50 Traflas (Tops), And at least 24 GB of RAM. The model itself consumes about 9 GB of RAM when producing the image, indicating its heavy processing. The most importa advaage of this model is local implemeation on the device that guaraees high production speed and operation without the need for iernet or cloud service.
This model is mainly for Coe manufacturers And Designers Designed that require customizable images and supports advanced capabilities to accurately corol the image combination.
To use this model, users must last the latest drivers AMD Adrenalin Edition And software Amuse 3.1 Beta Install from TensSorstack. After installing, by unlocking the Amuse software, you must go to EZ Mode, place the slider on HQ and the option “XDNA 2 Stable Diffusion Offload” Activate.
Use this model under authorized Stability ai Community License It will be free for individuals and small businesses with less than $ 1 million annually, but it is possible to change the license in the future. It should also be noted that the AMUSE software is still in the beta stage, so it may have problems with stability and performance.
