PEAP: AIO has a new product 2ru 1.5 pb ai data server, using Dell Hardware, which sends data at 120 Gbit / s.
PEAP: AIO is a British storage startup focused on AI which provides storage defined by software on third -party hardware, which it manages and controls closely to reduce latency and increase the flow. Its 2RU servers have delivered 40 GBPS to medium-sized GPU clusters, using NFS and NVME-OF, and can now go three times faster.
The new equipment incorporates the SSD QLC TB QLC of 61.44 TB of Solidigm is pumped into the NVIDIA CX7 nicks. PEAP: AIO says that specialized areas such as health care and research have roughly the same capabilities of the Genai model at high speed as those produced by massive data infrastructure systems. He says his systems are less expensive but always very fast and energy efficient. Peak says it has a new NVME software stack which “eliminates Linux Legacy strangulation bottlenecks” to provide better performance for AI applications with high data intensity.
The co-founder and director of the strategy Mark Klarzynski said: “Our approach addresses the two ends of the IA infrastructure spectrum. For smaller and innovative projects, this 2U solution offers unrivaled performance and energy efficiency in a compact form. For large -scale deployments, it becomes the foundation of a new generation of scale file systems, specially designed to meet modern GPU energy and performance requests. »»
The CX7 requires a PCIe Gen 5 bus and it is maximum at 50 Gbit / s. We understand that the DELL R7625 Rack Poweredge R7625 server has a 2RU chassis and supports up to eight PCIe Gen5 locations. Three CX7 with single port, each using a PCIe Gen5 x 16 lane location, will provide 120 Gbit / s. The R760 supports up to 24 SSD NVME and filling these slits with SOLIGM SSDs of 61.44 TB produced 1.5 pb of raw capacity.
Klarzynski tells us: “Dell Technologies has validated the NVME discs of 61 TB of Soligm and the cryptographic interfaces of 400 GB of Nvidia within the R7625 exclusively for us, exceeding the solutions of traditional 15TB. … This decision highlights our scale and our global ambitions. Our AI 2U data server (1.5pb, 120 GB / SEC) defines a new industry reference for efficiency. »»
The AI data server can be connected directly to an NVIDIA GPU server for small or new start -up projects, a switch being added to connect it to several GPU servers, up to 10 of them. Data servers can be set up to support more GPU.
Peak says that he has grown 400% in 2023 and 2024, a lot in the United States, and plans to launch a new range of products later this year to complete its new AI data server which, “with the support of Dell Technologies … Disturbing energy, cooling and density challenges”. It also has a next cryptographic unloading capacity at the line of line integrated into the CX7.
