Today Penguin Computing announced the Relion 1904GT server, which packs four GPU accelerators in a 1U form factor. As the company’s densest 1U GPU server, Relion is an exceptional platform for running scientific and engineering applications that support GPU technology.
We redesigned this system with a completely new layout of the motherboard,” said William Wu, senior product manager, Penguin Computing. “The result not only paid dividends with a fourth GPU into a 1U server, but also an improved thermal-optimized architecture to support even higher core count Intel Xeon E5-2600 v3 processors.”
Penguin Computing provides turnkey, ready-to-run HPC clusters combining this new GPU platform with storage and cluster interconnects. The Relion 1904GT server implements NVIDIA Tesla K80 dual-GPU accelerators, providing extremely high levels of computational performance and energy efficiency. The Tesla K80 dual-GPU is the flagship offering of the Tesla Accelerated Computing Platform, the leading platform for discovery and insight at scale. The Tesla K80 delivers nearly two times higher performance and double the memory bandwidth of its predecessor, and 10 times higher performance than today’s fastest CPU on hundreds of applications.
Relion 1904GT server features include:
- Innovative motherboard layout design allows support for higher core count Intel Xeon E5-2600 v3 processors
- Exceptional 2 CPU:8 GPU ratio implementing NVIDIA Tesla K80 dual-GPU accelerators
- Faster Local Machine – SAS/SATA and NVMe
- Management and monitoring tools for server and installed GPUs
- Dual 2000W redundant high efficiency power supplies
- Improved sensors for environmental measurement
HPC users are increasingly turning to higher density computing solutions to power their deep learning, engineering and scientific computing workloads,” said Roy Kim, group manager of Accelerated Computing at NVIDIA. “Packing four Tesla K80 GPU accelerators into a compact 1U form factor, Penguin’s new Relion system provides HPC customers with new levels of energy-efficient performance for their most pressing computing challenges.”