NVIDIA discussed the issue of artificial intelligence in modern computers in a meeting with the press. discussed. Judging by the discussion, it may contain a clue that Jensen is preparing for the Computex keynote speech to be held next month.
While the AI revolution has begun with the growing popularity of AI chatbots and productive multimedia, according to NVIDIA, the company ushered in the AI revolution by introducing AI-powered DLSS upscaling in 2018. pioneering did. NVIDIA would like to remind you that artificial intelligence has always been the domain of graphics processors and still offers superior performance.
Emphasizing that Premium AI PCs are powered by GPU and used instead of NPU, NVIDIA states that Basic AI PCs offer 10 to 45 TOPS for AI workloads, while Premium AI PCs offer 100 TOPS. It states that it offers up to 1300 TOPS. They may be right about this, because sales of Premium AI PCs have now exceeded 100 million units, but NPU-based systems have not reached 1 million sales even by the end of 2023.
NVIDIA offers AI acceleration support for GPUs via PyTorch, ONNX and TensorFlow. NVIDIA’s technologies such as DLSS, RTX Remix and NVIDIA ACE are currently used or will be used in future games. Going forward, AI workloads will still rely on GPUs.
Intel’s Core Ultra 200V Lunar Lake processors also offer 100 TOPS with the combined power of the CPU, NPU and GPU and fall into the Premium AI PC category according to NVIDIA’s nomenclature. GPU empowerment of Premium AI PCs is preferred because it provides higher performance. Basic AI PCs are only sufficient to run basic tools.
