Quanta Cloud Technology Expands the AI Cloud Universe with NVIDIA HGX-2
30 May. 2018

Top solution provider builds best-in-class AI and HPC system with NVIDIA’s HGX-2 Server Platform


[San Jose, May 30, 2018] Quanta Cloud Technology (QCT), a global data center solution provider, today announced its collaboration with NVIDIA to develop its next accelerated computing platform. Currently among the first to start this development, QCT is incorporating the NVIDIA® HGX-2 cloud server platform, striving for best-in-class performance that is ready for next-generation artificial intelligence (AI) and high performance computing innovation.


The NVIDIA HGX-2 server platform features 16 NVIDIA Tesla® V100 32GB Tensor Core GPUs connected by NVIDIA NVSwitch™ interconnect fabric that enables AI training and inferencing models at an unprecedented 2.4 TB per second. Designed for the ever-increasing complexity and size of AI models, the HGX-2 has sufficient compute power for the toughest workloads.


“As a leading cloud enabler, QCT is committed to developing solutions for the next generation of clouds for a variety of innovative use cases. As we have seen a multitude of AI applications on the rise, QCT works closely with NVIDIA to ensure our clients benefit from the latest and greatest GPU technologies,” said Mike Yang, President of QCT. “We are thrilled to broaden our GPU compute portfolio with this critical enabler for AI clouds as an HGX-2 launch partner.”


HGX-2 will augment QCT's extensive array of accelerated computing platforms. QCT currently offers a full range of solutions in its portfolio that fit a broad spectrum of accelerated computing requirements:


  • QuantaGrid D52G-4U, a part of the NVIDIA SCX-E3 class of GPU-Accelerated Server Platforms, is a powerful accelerated system supporting multiple different topologies optimized for different workloads. Available with 8x SXM2 module support or PCIe version with support up to 20 single-width accelerators.


  • QuantaGrid JG4080 is a companion model to D52G-4U and also part of the SCX-E3 class. It has the same great configurability, excluding the CPU sled. Available with SXM2 support or PCIe configurations, it allows connections to multiple hosts, providing flexibility to fine tune the setup for specific workloads.


  • QuantaGrid D52BV-2U, a SCX-E2 class server, is a compact 4x GPU PCIe server in a 2U form factor designed for workloads that require higher CPU to GPU ratio.


  • QuantaGrid Q72D-2U, a SCX-E1 class server, is a highly versatile PCIe quad-socket server featuring 48 DIMM memory slots and 2 slots for GPUs.


QCT is also fully committed to Open Computing Project (OCP), an open hardware initiative started by Facebook. QCT has contributed multiple designs such as Big Basin and Big Sur optimized for AI/deep learning, inferencing and training.  


Speak to QCT representatives at Computex 2018 in Taipei, Taiwan, starting June 5 in the Taipei International Convention Center (TICC) Room 101C, where the latest HGX-2 based technologies will be on display.


For more information on QCT’s GPU system product line, please visit www.QCT.io.


Follow QCT on Facebook and Twitter to receive the latest news and announcements.