Ebook

MLPerf and FIO Storage Benchmarking with Sycomp Intelligent Data Storage Platform

MLPerf and FIO Storage Benchmarking with Sycomp Intelligent Data Storage Platform

Pages 12 Pages

The document presents the results of MLPerf and FIO benchmarking for the Sycomp Intelligent Data Storage Platform on Google Cloud. Using A3 Ultra VMs, the system achieved 97% GPU utilization, 23 GiB/s MLPerf throughput, 157,000 IOPS for 4KiB random reads, and 28.5 GiB/s for 64KiB sequential reads. The tests confirmed scalability beyond 1 TiB/s, supporting demanding AI and ML workloads. Built on IBM Storage Scale, the solution demonstrates near-linear scalability and performance efficiency. Sycomp’s software-defined platform enables fast deployment, auto-data placement, and 24/7 support, optimizing hybrid and cloud storage performance for high-performance computing and AI environments.

Join for free to read