The article discusses the increasing presence of AI and LLMs in technology, with companies like Nvidia and Qualcomm promoting the use of accelerators like GPUs and NPUs for machine-learning work. Arm argues that its CPU cores are suitable for running AI software without the need for a separate accelerator. The article also delves into the debate around running AI workloads on CPUs, NPUs, and GPUs, as well as the power and infrastructure required for such tasks. It mentions Palantir’s AI targeting system being integrated into the US military. The article includes a video discussion on the topic, featuring hosts Iain Thomson, Chris Williams, Tobias Mann, and Brandon Vigliarolo. The Kettle series is available as a podcast for those who prefer audio content. The article emphasizes the growing importance of AI in various technologies and the different approaches to handling AI workloads.
Source link
Source link: https://www.theregister.com/2024/06/01/kettle_ai/
Do you need GPU or NPU for AI apps? #HardwareDecision
