NVIDIA Releases Major Updates to Quantum-2, cuQuantum, CUDA, and BlueField DOCA Acceleration Libraries

Accelerating Scientific Research: NVIDIA Releases Major Updates to Quantum-2, cuQuantum, CUDA and BlueField DOCA Acceleration Libraries

DALLAS - SC22 - November 14, 2022 - NVIDIA today announced broad adoption of Quantum-2 InfiniBand, with multiple customers, including Microsoft Azure Cloud, using the new product to accelerate scientific research.

NVIDIA partners introduced these new products at SC22, where NVIDIA also announced major updates to cuQuantum, CUDA®, and the BlueField® DOCA™ acceleration library.

Both Quantum-2 and the library updates are part of the NVIDIA HPC platform, a complete technology stack that includes CPUs, GPUs, DPUs, systems, networking, and a wide range of AI and HPC software, allowing researchers to significantly accelerate their work based on powerful systems both locally and in the cloud.

“AI is reshaping the way science is done,” said Jensen Huang, founder and CEO of NVIDIA. “By learning from data, AI can predict the highly complex workings of nature — from the behavior of plasma particles in a nuclear fusion reactor to the impact of humans on regional climates over the coming decades. We are providing scientists with a universal scientific computing platform that accelerates principled numerical and AI methods, a tool that will enable scientists to conduct research that will benefit humanity.”

Azure First to Offer NVIDIA Quantum-2 for HPC Workloads

After NVIDIA announced the general availability of Quantum-2 at GTC in March this year, Microsoft Azure was the first to adopt the NVIDIA Quantum-2 InfiniBand network platform.

“Next-generation cloud platforms like Microsoft Azure are the future of revolutionary enterprise technologies like AI and HPC, enabling innovators to achieve breakthroughs that will revolutionize the way we do business,” said Nidhi Chappell, general manager of Azure AI Infrastructure at Microsoft. “NVIDIA Quantum-2 InfiniBand networking gives Azure the throughput of a world-class computing center, available on demand at cloud scale, enabling researchers and scientists using Azure to achieve their life’s work.”

The acceleration library has received a major update

To help advance scientific research, NVIDIA released major updates to the CUDA, cuQuantum, and DOCA acceleration libraries:

● The NVIDIA CUDA library now includes a multi-node, multi-GPU Eigensolver, enabling unprecedented scale and performance for leading HPC applications such as VASP, a software package for first-principles quantum mechanics calculations.

● The NVIDIA cuQuantum software development kit for accelerating quantum computing workflows now supports methods for approximating tensor networks. This enables researchers to simulate tens of thousands of qubits and automatically enables multi-node, multi-GPU support for ultra-high-performance quantum simulations using the cuQuantum Appliance.

● Open cloud SDK and acceleration framework for NVIDIA BlueField DPUs – NVIDIA DOCA now includes advanced programmability, security, and capabilities to enable new storage use cases.

These libraries enable researchers to scale across multiple servers and achieve massive performance gains to advance scientific research. NVIDIA HPC acceleration libraries are already in use on leading cloud platforms AWS, Microsoft Azure, and Oracle Cloud Infrastructure.

To learn more about NVIDIA's high-performance computing platform, watch the SC22 special presentation.