Supermicro GPU systems offer a broad portfolio to suit the most demanding workloads requiring GPU co-processing capabilities. From systems optimized for AI learning environments to those for inference systems and HPC applications, Supermicro offers the right system for your needs. Our products combine the flexibility of near limitless configuration options, unlike the fixed configurations of alternatives. We offer support for 2nd generation Intel® Xeon® processors, Intel® Optane™ DC persistent memory, and offer multiple options for NVMe, SAS and SATA drives.
Supermicro GPU systems provide the optimal foundation for maximum performance with the latest in GPU technologies. To provide the highest performance and lowest latency, NVIDIA® NVLink® and/or NVSwitch® is used to connect multiple NVIDIA® Tesla® V100 GPUs in GPU systems optimized for learning, with a 10U HGX-2 system providing up to 16 V100 GPUs, a 4U system providing up to 8 V100 GPUs and a 1U system providing up to 4 V100 GPUs. Systems optimized for inferencing are composed of NVIDIA® Tesla® T4 GPUs, with a 4U system providing support for up to 20 T4 GPUs. Additional models support single or dual PCI-E root complex devices for high-performance optimization. With a broad line of GPU systems supporting 1, 2, 4, 8, 16 or 20 GPUs across form factors ranging from a dense 1U system through the most powerful 10U system, supporting scientific research, 3D rendering, HPC, virtual desktops or AI/deep learning is all possible with Supermicro’s high performance GPU systems.
N0206 ► http://www.supermicro.com
2933 MHz in two DIMMs per channel can be achieved by using memory purchased from Supermicro
For 2nd Generation Intel® Xeon® Scalable processors (Cascade Lake-SP) only, contact your Supermicro sales rep for more info