NVLink unlocks the full potential of GPU accelerators by enabling CPUs and GPUs to transfer data at unprecedented speeds,” said Sumit Gupta, general manager of the Accelerated Computing business at NVIDIA. “QCT extends the benefits of NVLink to the x86 ecosystem and gives our customers an easy path to higher levels of enterprise and HPC application performance.”
QCT will work with NVIDIA on development of next-generation x86 systems around the latest components of the NVIDIA Tesla Accelerated Computing Platform. The program allows QCT to fully leverage GPU acceleration for a diverse set of applications, such as supercomputing, deep learning and data analytics.
This video provides an overview of NVLink. Download the whitepaper for more details.
This week at the GTC Conference, the company also showcased two new GPU-accelerated servers:
- QuantaGrid D51BV-2U: 2U server based on Intel Xeon Processor E5-2600v3 and DDR4 technology with support for up to two GPU accelerators.
- STRATOS S210-X2A2J: a high-density server designed to support up to four GPUs in a compact 2U chassis.
As a pioneer in VDI on 3D applications, QCT has been working with Nvidia on its “VDIBlock 3D Advance” solution. This hyper-converged solution includes compute and storage nodes with Nvidia GRID GPUs fully utilizing the power of VMware Horizon 6.0 Advanced Edition. A technology preview at GTC will show how the solution targets virtual desktop users with intensive graphics, engineering and rich media needs. This well-tuned and balanced solution also provides flexible options to allow high IOPS and high capacity on demand.
Additionally at GTC, QCT will show its new storage innovation product, the QuantaPlex T21P-4U. The 4U, two-node storage server features two dual-socket server nodes and up to 70 top-loaded hot-swappable 3.5-inch HDD bays. It provides unprecedented features like a screw-less and tool-less HDD tray, enabling HDD drive replacement in seconds. Both motherboards and HDD backplanes can be quickly serviced.