PEAK:AIO storage collaboration with Scan and Micron improves power efficiency, cooling, and space utilization in AI workloads.
PEAK:AIO, a provider of a transformative software defined storage platform designed to meet the needs of artificial intelligence, today announced that it has collaborated with Scan and Micron Technology, Inc. (Nasdaq: MU) to deliver unparalleled artificial intelligence (AI) performance and scalability with PEAK:AIO’s introduction of Micron’s ground-breaking 30TB drives.
Micron turbocharges PEAK:AIO’s AI Data Server, unlocking new levels of performance and scale, delivering up to 650TBs of storage capacity in a single 2U server with enough performance to supercharge the hardiest of AI projects. With over 171GB/sec internal performance, PEAK:AIO is futureproofing compatibility while maximising NVIDIA’s network connectivity such as the Connect-X range and BlueField DPUs.
“The AI market is evolving rapidly, driven by startups and research institutions seeking cutting-edge capabilities. Strategic allocation of funds is crucial, particularly towards GPU resources,” said Elan Raja, CEO of Scan International, an NVIDIA Elite Partner who ran and oversaw the tests. “As resellers, our responsibility is to help drive client innovation and advise on technology. With PEAK:AIO’s latest solution coupled with Micron’s newest drives, capabilities that were previously only accessible to those with much larger budgets are now available to everyone.”
Beyond ground-breaking performance, Micron and PEAK:AIO together address the critical aspects of power efficiency, cooling, and space utilization in AI installations. The combination of Micron 9400 NVMe SSDs and the PEAK:AIO AI Data Server with 650TBs of capacity and up to and up to 170GB/sec of performance in a 2U form factor is available at 1/6th of the footprint and cost of systems that would typically provide that level of power.
PEAK:AIO achieves its exceptional performance without resorting to historic storage benchmarking practices such as using multiple clients. PEAK:AIO’s full performance can be delivered to a single GPU client server. No other solution in the market matches this level of performance delivered to a single client.
“Maximising internal RAID performance is important, but in real life AI solutions, the true value of high-performance storage capabilities lies in the ability to be shared with other AI/GPU systems through newer and innovative RDMA protocols, such as GPUDirect Storage by NVIDIA and RDMA NFS,” said Eyal Lemberger, CTO, PEAK:AIO. “Developing, managing, and supporting this complex ecosystem is what truly takes focus. With the help of Micron, what PEAK:AIO has achieved is a realistic, large scale storage appliance in a small 2U footprint, designed for AI teams, not storage for Linux administrators, and priced to align with GPU-centric budgets.”
Visit AITechPark for cutting-edge Tech Trends around AI, ML, Cybersecurity, along with AITech News, and timely updates from industry professionals!
