Western Digital has announced the results of its MLPerf® Storage V2 submission, confirming the real-world performance of its OpenFlex™ Data24 4000 Series NVMe-oF™ Storage Platform. The results validate the platform’s ability to meet the demanding requirements of modern AI workloads with high performance, efficiency, and scalability.
As AI workloads become more complex, the need for storage systems that can keep pace with accelerated computing infrastructure is crucial. The OpenFlex Data24 platform is designed to extend the high performance of NVMe flash over Ethernet fabric, providing low-latency shared storage for scalable, disaggregated AI infrastructure. This design allows for independent scaling of storage and compute, simplifying deployment and reducing costs.
To ensure the results reflected realistic deployment scenarios, Western Digital partnered with PEAK:AIO, a high-performance software-defined storage (SDS) provider. The tests utilized KIOXIA CM7-V Series NVMe SSDs, which were chosen for their performance in demanding AI workloads.
The MLPerf Storage V2 benchmark, considered the industry standard for AI benchmarking, uses GPU client nodes to simulate real-world AI training and inference workloads. Western Digital’s results show that its architecture not only delivers performance at scale but does so with a focus on efficiency and practical deployment economics.
The benchmark included two key workloads:
- 3D U-Net Workload: This model, used in medical imaging, places a heavy load on storage systems due to its large datasets and intensive data-streaming patterns. In this test, the OpenFlex Data24 achieved a sustained read throughput of 106.5 GB/s, saturating 36 simulated H100 GPUs.
- ResNet50 Workload: This image classification model tests a system’s ability to handle high-frequency access to smaller files. Here, the OpenFlex Data24 delivered optimal performance across 186 simulated H100 GPUs with a highly efficient GPU-to-drive ratio.
“These results validate Western Digital’s disaggregated architecture as a powerful enabler and cornerstone of next-generation AI infrastructure, maximizing GPU utilization while minimizing footprint, complexity and overall total cost of ownership,” said Kurt Chan, vice president and general manager, Western Digital Platforms Business. “The OpenFlex Data24 4000 Series NVMe-oF Storage Platform delivers near-saturation performance across demanding AI benchmarks, both standalone and with a single PEAK:AIO AI Data Server appliance, translating to faster time-to-results and reduced infrastructure sprawl.”
“These MLPerf results spotlight the breakthrough efficiency achieved by combining PEAK:AIO’s software-defined AI Data Server with the scalability of Western Digital’s OpenFlex Data24 and the performance density of KIOXIA’s CM7-V Series SSDs,” said Roger Cummings, President and CEO at PEAK:AIO. “Together, we’re delivering high-performance AI infrastructure that’s faster to deploy, more efficient to operate, and easier to scale. It’s a compelling proof point that high performance no longer requires high complexity.”
The OpenFlex Data24 platform, with its RapidFlex™ network adapters, allows for up to 12 hosts to be attached without a switch. This makes it a scalable and cost-effective solution for organizations at any stage of their AI journey.
