Lenovo ThinkSystem SR780A V3
March 18, 2024 — Today, GTC (GPU Technology Conference), a global AI event, officially opened in 2024. Lenovo Group and NVIDIA announced a partnership to launch a new hybrid artificial intelligence solution to help enterprises and cloud providers obtain the critical accelerated computing capabilities needed to succeed in the AI era, turning artificial intelligence from concept to reality.
At the same time, in terms of efficiently enhancing AI workloads on a large scale, Lenovo announced the expansion of the ThinkSystem AI product portfolio, which is an industry-changing milestone, including two new powerful 8-channel NVIDIA GPU systems, which are designed to provide massive computing power and unparalleled energy efficiency to accelerate the implementation of artificial intelligence. Designed for generative artificial intelligence, natural language processing (NLP), and large language model (LLM) development to support the market entry of the NVIDIA HGX artificial intelligence computing platform, including NVIDIA H100 and H200 Tensor Core GPUs and the new NVIDIA Grace Blackwell GB200 superchip, as well as advanced next-generation NVIDIA Quantum-X800 InfiniBand and Spectrum-x800 Ethernet networking platform.
The Lenovo ThinkSystem AI server is equipped with an NVIDIA B200 Tensor Core GPU, which will power a new era of generative AI. The NVIDIA Blackwell architecture's generative AI engine, along with faster NVIDIA NVLink interconnects and enhanced security features, has propelled data centers into a new era. The B200 GPU is 25 times more capable of real-time inference, accelerates trillion-parameter language models, and is designed for the most demanding artificial intelligence, data analysis, and high performance computing workloads.
The new Lenovo ThinkSystem SR780a V3 is a 5U system using Lenovo Neptune liquid cooling technology, with a power usage efficiency (PUE) of up to 1.1. By using direct water cooling of the CPU and GPU and NVIDIA NVSwitch technology, the system can maintain maximum performance without reaching any thermal limitations. For more than a decade, Lenovo's industry-leading Neptune direct water cooling solution uses warm water to cool data center systems, enabling customers to reduce power consumption by up to 40% and increase cooling performance by 3.5 times. Compared with traditional air cooling systems, it is more efficient. According to the “Uptime Institute 2021 Global Data Center Annual Survey,” PUE and power consumption are among the most watched sustainable development methods as industry metrics used to determine data center energy efficiency. Because liquid cooling provides a more energy efficient alternative to air, the system can achieve higher sustained performance while consuming less energy. It also allows the ThinkSystem SR780a to be installed in a dense 5U package, helping to save valuable data center space.
Additionally, the new Lenovo ThinkSystem SR680a V3 is an air-cooled two-socket system that provides maximum acceleration for complex artificial intelligence equipped with an Intel processor and NVIDIA GPU. The highly accelerated system provides tremendous computing power and uses an industry-standard 19-inch server rack, allowing intensive hardware configurations to maximize efficiency while not taking up too much space.
Lenovo also unveiled the new PG8A0N, an ultimate 1U node designed for AI and supporting the NVIDIA GB200 Grace Blackwell Superchip. The new GB200 chip will power a new era of computing. The superchip's real-time large language model (LLM) inference speed was 45 times faster, the total cost of ownership (TCO) was reduced by 40 times, and energy consumption was reduced by 40 times. Lenovo will work closely with NVIDIA to provide the GB200 rack system to enhance AI training, data processing, engineering, and simulation capabilities.
Customers can run NVIDIA AI Enterprise, an end-to-end cloud-native software platform for developing and deploying production-grade AI applications on Lenovo's NVIDIA certified systems portfolio. Customers can run the NVIDIA NIM inference microservice included in NVIDIA AI Enterprise on Lenovo's enterprise-grade infrastructure to infer popular AI models in NVIDIA and its ecosystem with high performance.
The product portfolio includes Lenovo's XClarity systems management software to improve the efficiency of infrastructure deployments for enterprises, and Lenovo's Intelligent Computing Platform (LiCO), which streamlines the use of cluster computing resources for artificial intelligence model development and training, and HPC workloads. The product portfolio also supports 4th and 5th generation Intel Xeon scalable processors and provides heat headroom for future higher-power GPUs.