NVIDIA and HP have announced that NVIDIA's CUDA-X data processing libraries will be incorporated into HP AI workstation solutions. This integration aims to enhance the data preparation and processing activity that underpins generative AI development.
NVIDIA's CUDA-X libraries are designed to hasten data processing for a diverse spectrum of data types, which include tables, text, images, and videos. The libraries are underpinned by the NVIDIA CUDA computing platform and incorporate the NVIDIA RAPIDS cuDF library. The latter is designed to accelerate the tasks of the roughly 10 million data scientists who use pandas software by up to 110x using an NVIDIA RTX 6000 Ada Generation GPU instead of a CPU-only system. It does so without necessitating any changes to the code.
The cuDF and other NVIDIA software will be accessible as part of the Z by HP AI Studio on HP AI workstations, which will furnish a comprehensive, full-stack solution to expedite data science workflows.
"Pandas is the essential tool of millions of data scientists processing and preparing data for generative AI," stated Jensen Huang, Founder and CEO at NVIDIA. "Accelerating pandas with zero code changes will be a massive step forward. Data scientists can process data in minutes rather than hours, and wrangle orders of magnitude more data to train generative AI models."
"Data science provides the foundation for AI, and developers need fast access to software and systems to power this critical work," expressed Enrique Lores, President and CEO of HP "With the integration of NVIDIA AI software and accelerated GPU compute, HP AI workstations provide a powerful solution for our customers."
The NVIDIA RAPIDS cuDF library speeds up pandas to work on GPUs with zero code changes. This replaces dependence on CPUs, which can decelerate workloads as data size expands. RAPIDS cuDF is compatible with third-party libraries and synchronises GPU and CPU workflows, allowing data scientists to develop, test, and run models in production without interruption.
As larger quantities of data continue to accumulate, RTX 6000 Ada Generation GPUs offer 48GB of memory per GPU to process expansive data science and AI workloads on Z by HP workstations. With up to four RTX 6000 GPUs, the HP Z8 Fury is recognised as one of the world's most potent workstations for AI creation. The robust collaboration between HP and NVIDIA aids data scientists in streamlining development by processing even vast generative AI workloads on local systems.
NVIDIA RAPIDS cuDF, for accelerated pandas with zero code alterations, is expected to be available on HP AI workstation solutions with NVIDIA RTX and GeForce RTX GPUs this month. It will be available on HP AI Studio later this year.