Once inference becomes obsolete this will be incredibly useful for expanding the life of AI GPUS - power down most of the energy hungry cores and use the cards

Once inference becomes obsolete this will be incredibly useful for expanding the life of AI GPUS - power down most of the energy hungry cores and use the cards as ultra fast cache.