Sign up for our newsletter and get the latest HPC news and analysis.
Send me information from insideHPC:


OSS Introduces World’s First PCIe Gen 4 Backplane at GTC

Today One Stop Systems introduced the world’s first PCIe Gen 4 backplane, the OSS 5 Slot Gen 4 Backplane (OSS-PCIe4-BP-5×16), at the GPU Technology Conference in Silicon Valley.

We have. been consistently first-to-market with each new generation of PCIe technology, and Gen 4 is no exception,” said Steve Cooper, CEO of OSS. “OSS PCIe backplanes form the core of our high-performance computing products, and our new Gen 4 backplane offers double the IO performance. Our customers in the HPC, military, media and entertainment markets can now take advantage of the additional bandwidth available in Gen 4-equipped servers, and be ready when Gen 4 IO cards hit the market.”

PCIe Gen 4 offers cutting-edge system performance that doubles the interconnect bandwidth over Gen 3. It provides 16 GT/s per lane, or 256 GT/s in a x16 PCIe Gen 4 slot, alleviating system bottlenecks between CPU complexes and accelerators, storage and IO devices.

The OSS Gen 4 backplane features one Gen 4 x16 target slot and five Gen 4 x16 downstream slots for GPUs, NVMe drives or network interfaces. Two of these backplanes are installed into the OSS 4U Value Chassis, along with OSS Gen 4 cable adapters and eight NVIDIA® V100 Tensor Core GPUs, to create the next-generation HPC acceleration platform.

The value chassis can be connected to any Gen 4-equipped server with two 256Gbps connections. The V100 Tensor Core GPUs currently have Gen 3 interfaces; however, the OSS 4U Value Chassis aggregates the GPU traffic from several Gen 3 V100s and provides connection to the server at Gen 4 speeds. It will support all Gen 4 IO and accelerator cards as they become available throughout the year.

Delivering the high performance required by edge applications necessitates PCIe interconnectivity traveling on the fast data highway between high-speed processors, NVMe storage and compute accelerators using GPUs or application specific FPGAs,” continued Cooper. “‘AI on the Fly’ applications naturally demand this capability, like the government mobile shelter application we announced earlier this year.”

Visit the OSS booth #640 at GTC.

Sign up for our insideHPC Newsletter

Leave a Comment

*

Resource Links: