NVIDIA and HP Inc. have teamed up for the integration of NVIDIA CUDA-X data processing libraries with HP (artificial intelligence) AI workstation solutions, hoping to enhance the efficiency of data preparation and processing crucial for generative AI (GenAI) development.
Leveraging the NVIDIA CUDA compute platform, CUDA-X libraries will be able to expedite data processing across diverse data types, including tables, text, images, and video. Among these libraries is the NVIDIA RAPIDS cuDF, which significantly accelerates data processing tasks, empowering nearly 10 million data scientists who rely on pandas software, achieving up to a 110x acceleration when utilizing an NVIDIA RTX 6000 Ada Generation GPU instead of a CPU-only system, all without necessitating any code modifications.
The integration of RAPIDS cuDF and other NVIDIA software into Z by HP AI Studio on HP AI workstations provides a comprehensive development solution, expediting data science workflows.
READ:
NVIDIA builds Eos supercomputer to optimize AI workload
VMware, NVIDIA develop integrated generative AI platform
“Pandas is the essential tool of millions of data scientists processing and preparing data for generative AI,” said 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.”
Pandas offers a potent data structure, known as DataFrames, facilitating developers in effortlessly manipulating, cleaning, and analyzing tabular data.
Data science
“Data science provides the foundation for AI, and developers need fast access to software and systems to power this critical work,” said Enrique Lores, president and CEO of HP Inc. “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 accelerates pandas operations, enabling it to run seamlessly on GPUs without any code modifications, eliminating the reliance on CPUs, which can impede workflows as dataset sizes expand. RAPIDS cuDF seamlessly integrates with third-party libraries, harmonizing GPU and CPU workflows, enabling data scientists to develop, test, and deploy models seamlessly across different environments.
As datasets continue to burgeon, RTX 6000 Ada Generation GPUs, boasting 48GB of memory per GPU, cater to processing large-scale data science and AI workloads on HP Z by workstations. With the potential for up to four RTX 6000 GPUs, the HP Z8 Fury emerges as one of the world’s most potent workstations for AI development. The close collaboration between HP and NVIDIA empowers data scientists to streamline development by working on local systems, even when handling extensive generative AI workloads.