LLMs conquer laptops: NVIDIA and HP CEOs hail AI-enabled PCs

In an informal discussion, the CEOs of NVIDIA and HP have now predicted that 2024 will be the year when generative AI becomes personal. They unveiled new laptops capable of creating, testing and executing large-scale language models. At HP Amplify, a gathering in Las Vegas of some 1,500 resellers and distributors, NVIDIA founder and CEO Jensen Huang hailed the new era, calling it the “renaissance of personal computing”. He emphasized that the work of creatives, designers and data scientists will be revolutionized by these new workstations. HP’s CEO Enrique Lores emphasized that AI is a breakthrough development for the PC and announced that HP will introduce the industry’s largest portfolio of AI PCs and workstations.

Speed and security are of great importance

At the event, Lores said in his keynote that the new systems will offer increased speed and security compared to running their AI work in the cloud, while reducing costs and energy. The new HP ZBooks offer a range of mobile AI workstations powered by a full range of NVIDIA RTX Ada generation GPUs. Entry-level models with the NVIDIA RTX 500 Ada Generation laptop GPU enable users to run generative AI apps and tools wherever they go. High-end models pack the RTX 5000 to deliver up to 682 TOPS to create and run LLMs locally, using retrieval-augmented generation (RAG) to connect to their content for results that are both personalized and private.

Access to accelerated software

The new workstations can access NVIDIA’s full AI platform, including software that accelerates the data analysis that underpins generative AI. The systems’ Z by HP AI Studio platform – developed in collaboration with NVIDIA – is connected to the NVIDIA NGC, a catalog of GPU-accelerated software for AI and data analysis. NGC also includes NVIDIA NeMo, a framework for creating, customizing and deploying generative AI models. In addition, HP and NVIDIA have announced that the NVIDIA CUDA-X libraries will be integrated into the systems to accelerate data preparation and processing, which is fundamental to generative AI.

Accelerations for data scientists

The libraries include NVIDIA RAPIDS cuDF, which accelerates pandas, a software used by nearly 10 million data scientists. “It used to take hours and sometimes days to process data that they can now do in minutes,” Huang said. “This pandas library is insanely complex,” he added, noting that NVIDIA engineers spent more than five years reformulating the code so it could be accelerated with GPUs.

The dawn of a new era

Along with the new systems, HP announced a partner training program developed in collaboration with NVIDIA. It will equip computer vendors to advise customers on the right AI products and solutions to meet their needs. Such programs are paving the way for an industry entering an era where AI enables software to write software. “We have reinvented the computer. We have reinvented how software is written, and now we need to reinvent how software is used,” Huang said. “Large language models connected to other LLMs will help solve application problems – that’s the future.”

Source: NVIDIA