New innovations meet the generative AI needs of modern-day workloadsSan Jose, Calif., March 18, 2024—Quanta Cloud Technology (QCT), a leading provider of data center solutions, today demonstrated at NVIDIA GTC its NVIDIA MGX-based server systems, AI use cases, and support of the upcoming NVIDIA GB200 Grace Blackwell Superchip and NVIDIA GB200 NVL72, a multi-node, liquid-cooled, rack-scale platform designed for the new era of generative AI. These were showcased under the theme “Step into the AI Era with QCT NVIDIA MGX-powered Systems” at Booth 716.
QCT’s latest NVIDIA MGX-based systems, powered by the NVIDIA GH200 Grace Hopper Superchip, leverage a modular reference architecture. The MGX architecture allows system makers to create server variations tailored for a wide range of applications, including generative AI, high-performance computing (HPC), and edge deployments. For example, the QCT QuantaGrid S74G-2U based on the MGX architecture and powered by the NVIDIA GH200 Superchip helps enable scientific research breakthroughs at Oregon State University and San Jose State University, accelerating simulations and HPC modeling.
The QuantaEdge EGX77GE-2U is QCT’s first short-depth edge server to be powered by the NVIDIA GH200 Superchip and NVIDIA MGX architecture to support smart city, smart manufacturing, and smart healthcare workloads. QCT is leveraging NVIDIA’s latest technologies to create a series of accelerated servers in both air-cooled and liquid-cooled variants to handle diverse simultaneous workloads across various verticals or disciplines.
The just-announced NVIDIA GB200 Grace Blackwell Superchip is set to power a new era of computing, delivering 30X faster real-time large language model (LLM) inference, 25X lower TCO, and 25X less energy consumption compared to the NVIDIA Hopper architecture generation of GPUs. It combines two NVIDIA Blackwell GPUs and an NVIDIA Grace CPU and can scale up to the GB200 NVL72, a 72-GPU fifth-generation NVIDIA NVLink-connected GPU system with 36 Grace CPUs and NVIDIA BlueField-3 DPUs, that acts as a single massive GPU. Paired with the NVIDIA Quantum-X800 InfiniBand and Spectrum-X800 Ethernet solutions, also announced at GTC, the GB200 Superchip delivers unprecedented levels of performance for generative AI, HPC, and data processing.
“With generative AI weaving its way into all businesses and industry verticals, building systems with the latest NVIDIA AI architectures and innovations helps our customers reach insights faster,” said Mike Yang, President of QCT. “With NVIDIA MGX architecture implemented into our QCT QuantaGrid and QuantaEdge systems and groundbreaking Blackwell advancements including the GB200 NVL72, as well as BlueField-3 DPUs, NVIDIA Quantum-X800, and Spectrum-X800, all working in tandem, we’re poised to offer unprecedented levels of performance, efficiency, and security.”
“Generative AI is booming, with companies and researchers looking for ways they can use this groundbreaking technology to deliver better performance and faster time to discovery,” said Kaustubh Sanghani, vice president of GPU product management at NVIDIA. “Using our latest technologies — from NVIDIA GPU-accelerated platform servers featuring the new Blackwell architecture to NVIDIA MGX servers powered by the GH200 Superchip — QCT is developing advanced systems that push the boundaries of AI.”
Additionally, QCT is showcasing its QCT Platform on Demand, NVIDIA AI Enterprise software-based solutions, accelerated computing with NVIDIA HGX hardware, edge AI use cases, its QoolRack solution, and the aforementioned NVIDIA MGX systems and GB200 NVL72 platform at Booth 716 during NVIDIA GTC, taking place at the San Jose Convention Center from March 18-21.
To learn more about QCT accelerated systems, register for NVIDIA GTC to attend QCT’s free online sessions, or visit the QCT at GTC event page.
   
About Quanta Cloud Technology (QCT) All other brands, names, and trademarks are the property of their respective owners.
Media Contact Jean Ko, [email protected]
|