Does your HPC/AI cluster need better storage performance for converged workloads of mixed and random I/O requirements for ML/AI applications while preserving the cost-effectiveness and scalability of traditional parallel HPC storage?
The HPE Cray Storage Systems C500 utilizes an open-source, parallel file system architecture for impressive performance and is ideal for entry and mid-range HPC/AI clusters with limited storage budgets that want to improve the utilization of their CPU/GPU investments. This unique, hybrid architecture exploits the strengths of different storage media (NVMe, SSD/SAS, HDD) while avoiding their respective weaknesses, allowing superb performance speeds with fewer rack units.
Utilize the HPE Cray Storage Systems C500 to experience the same price/performance benefits for storage as the largest supercomputing leadership sites.
HPE Cray Storage Systems C500 uses the same cost-effective open source architecture that was selected for 100% of the United States DOE exascale supercomputers.123
Avoid capacity-based file system licensing either by terabyte storage capacity or storage drive as the embedded file systems are owned by vibrant communities and not by a single company.
Leverage a hybrid storage architecture that exploits the strengths of different storage media (solid state and spinning disk) while avoiding their weaknesses.
Achieve on-budget and balanced HPC/AI systems to navigate unprecedented data explosions.
HPE Cray Storage Systems C500 features a shared file storage that can read and write (r/w) data fast enough to keep your CPU/GPU compute node investment fully-utilized.
Remove job pipeline congestion and prevent missed deadlines with just three rack units to leverage up to 80 gigabyte per second read and 60 gigabyte per second write performance.
Compress CPU/GPU compute node wait times during the critical checkpointing and restart process for long-running HPC/AI jobs.
Leverage AI acceleration technologies like NVIDIA GPUDirect Storage (GS) to bypass the CPU and achieve superb bandwidth.
As part of your strategic infrastructure plan, the HPE Cray Storage Systems C500 provides unified support accountability for your full infrastructure stack.
Get solutions faster with customized software updates from the Hewlett Packard Enterprise R&D team for your specific issues–solutions, which later get embedded into the open-source file system for the benefit of the whole community.
Avoid production disruptions when migrating to your next HPC/AI storage infrastructure with HPE Accelerated Migration Services which helps generate value from your ageing infrastructure while your modern infrastructure is prepared for deployment.
Exascale supercomputer "Frontier" at Oak Ridge National Laboratory https://www.olcf.ornl.gov/frontier/
Exascale supercomputer "Aurora" at Oak Argonne National Laboratory https://www.anl.gov/aurora
Exascale supercomputer "El Capitan" at Lawrence Livermore National Laboratory https://asc.llnl.gov/exascale/el-capitan
GPUDirect and NVIDIA are trademarks and/or registered trademarks of NVIDIA Corporation in the U.S. and other countries. All third-party marks are property of their respective owners.
* Prices may vary based on local reseller.
