RESEARCH TRIANGLE PARK, N.C.--(BUSINESS WIRE)--
Today, Lenovo (HKSE: 992) (ADR: LNVGY) Data Center Group (DCG) announces new breakthrough high performance computing solutions to help customers of all sizes build a smarter way forward with deep analytics and AI by accelerating and optimizing their data, delivering faster, more cost-efficient insights and improving AI-based decision making.
Unprecedented Data Acceleration
The server is the industry’s first Direct-to-Node (DTN) liquid-cooled server for NVIDIA® A100 Tensor Core GPUs. It includes four board-mounted NVIDIA® A100 GPUs in a 1U system, delivering up to 3PFLOPS of compute performance in a single rack1. Lenovo Neptune™ liquid cooling reduces energy consumption by up to 40%2 while maintaining unprecedented compute power and density.
“Our intense focus on HPC innovation stems from our commitment to our customers, who are working to save human lives and helping solve some of humanity’s greatest challenges,” said Scott Tease, General Manager, HPC and AI at Lenovo Data Center Group. “As a result, Lenovo has been recognized as the #1 supercomputer provider*, reflecting our commitment to groundbreaking innovation in HPC, Analytics and AI for businesses of any size.”
Breakthrough AI innovation
To drive compute-intensive workloads, Lenovo introduces the new , a modular, GPU-rich system that supports up to eight NVIDIA® A100 Tensor Core GPUs or in a single 3U frame, delivering up to 160 TFLOPS of compute performance. Included is a model that leverages ™ liquid-to-air heat exchangers (no need to add plumbing) to cool four board-mounted NVIDIA® A100 GPUs, giving customers breakthrough AI performance with the benefit of reduced power consumption and less noise as compared to air cooled systems.
, a German research university, plans to implement a new 17 PFLOPS system, featuring warm water-cooled NVIDIA A100 Ampere GPUs, with the help of Lenovo and business partner Pro-Com in the coming months. Dr. Jennifer Buchmüller, Head of the Department for Scientific Computing and Simulation at KIT said, “We are looking forward to collaborating with Lenovo for our state-of-the-art HoreKa supercomputer, utilizing their award-winning Neptune Direct Water Cooling (DWC) technology. The eco-credentials of this solution and the performance optimization of the overall system is ideally matched with our objective of developing increasingly efficient and sustainable scientific software. This in turn enables multi-scale simulations, significantly larger than we’ve ever done before, for research in the fields of energy and mobility in engineering, material sciences, earth system sciences, life sciences, and particle & astro-particle physics.”
These new Lenovo servers can also take advantage of the extensive software catalog available from , which includes containerized frameworks, applications, pre-trained models, scripts and Helm charts, all of which are freely downloadable. Today, the NVIDIA data center platform accelerates over 700 HPC applications.
Helping solve humanity’s greatest challenges
Using the (Genomics Optimization and Scalability Tool) system configuration, scientists can leverage a validated, pre-configured bioinformatics solution built on high-performance, high-reliability Lenovo ThinkSystem servers. Traditionally, processing a whole genome (WGS) requires a time scale measured in days, but the Lenovo GOAST (Genomics Optimization and Scalability Tool) system configuration delivers results in just under an hour using standard x86 hardware.
“With the Lenovo GOAST (Genomics Optimization and Scalability Tool) system, I can analyze more data and uncover new insights faster. Our new HPC environment is powering cutting-edge research that will help us to breed more nutritious, more drought and disease-tolerant, high-yield plants to feed the world,” said Dr. Paritosh Kumar, Post-Doctoral Researcher, Centre for Genetic Manipulation of Crop Plants, Department of Genetics, University of Delhi.
To ensure data centers have the required power and cooling capacity for present and future IT plans, offers new interactive Data Center Best Practices Workshops. By partnering with power and cooling experts, Lenovo can help diagnose IT infrastructures through a series of assessments based on customer needs at no charge for customers, resulting in defining proven methodologies to achieve cooling, power and energy efficiency.
To learn more about Lenovo’s award winning solutions visit or sign up for a power and cooling workshop .
1 Based on internal research by Lenovo together with NVIDIA. A single node = 84TF (78TF for the GPU, 5.5TF for the CPU**); therefore, a rack = 3PF (84TF per node x 36 nodes/rack). October 2020.
** NVIDIA has a binary for HPL they are using for their Tensor Cores on the GPU. NVIDIA is getting 18.6TF sustained on the SXM version. Lenovo will use the 19.5 TF Rpeak number for GPU performance.
2Up to 95% heat removal via water and saves up to 40% of energy costs
Substantiation: Based on internal research by Lenovo. 95% heat removal via water – Note: CPU, storage, memory, voltage regulators, and PCI are 95% of the power/heat. Save up to 40% energy costs based on customer testimonials, including Leibniz Supercomputing Centre (LRZ) at 35 – 40% savings and New York University (NYU) at 40% savings. October 2020.
*Based on Lenovo as manufacturer of the largest number of systems ranked on the world’s TOP500; June 22, 2020.
About Lenovo
Lenovo (HKSE: 992) (ADR: LNVGY) is a US$50 billion Fortune Global 500 company, with 63,000 employees and operating in 180 markets around the world. Focused on a bold vision to deliver smarter technology for all, we are developing world-changing technologies that create a more inclusive, trustworthy and sustainable digital society. By designing, engineering and building the world’s most complete portfolio of smart devices and infrastructure, we are also leading an Intelligent Transformation – to create better experiences and opportunities for millions of customers around the world. To find out more visit , follow us on , , , , , and read about the latest news via our .
View source version on businesswire.com:
