Solidigm recently announced the opening of its AI Central Lab featuring what it claims are the industry's highest-performing and most dense storage test clusters built for artificial intelligence workloads. Located in Rancho Cordova at the FarmGPU facility near Solidigm headquarters, the AI Central Lab lets Solidigm run and study real-world workloads on NVIDIA B200 and H200 GPUs. This gives the company insights into how storage can boost cluster efficiency and allows key AI ecosystem partners to run experiments on the lab's hardware. The AI Central Lab features several notable configurations. The performance cluster using Solidigm D7-PS1010 SSDs achieved 116 GB/s per-node throughput in MLPerf Storage testing—the highest ever measured in the AI model training test. The density cluster packs 192 Solidigm D5-P5336 SSDs with 122 TB each, fitting 23.6 PB into just 16U of rack space. Additional hardware includes 800 Gbps Ethernet networking and storage servers. Workloads available for testing include AI model training and inference, KV cache offload, VectorDB tuning, and power consumption benchmarking.

The lab was built with common reference architectures in mind, letting customers test their solutions with Solidigm SSDs in an environment using the same standards as large-scale data centers worldwide. The facility will also speed up storage research beyond conventional component-level testing to better understand real-world usage at the server, rack, and cluster levels. Work in the lab is already underway with multiple organizations.
Read full story

More...