HPE Cray SC User Services Software helps gauge the power utilization and distribution of your system by enabling measurement by processor (e.g. CPU or GPU) and by job.
Power data is collected from out-of-band (OOB) sensors accessible by BMC on the compute nodes through workload managers and system management tools.
Customers can leverage the collected data to form insights that influence their decisions to achieve better balance between power and performance.

HPE Cray SC User Services Software focuses compute resources on running HPC/AI workloads rather than executing background activities with low noise mode, which confines OS operations to designated Linux CPUs or hardware threads, freeing up remaining cores to uninterruptedly run user applications.
By providing the ability to set performance thresholds on selected nodes, turbo boost limiting can deliver predictable, consistent performance and reduce power consumption for workloads, consequently protecting uptime.

HPE Cray SC User Services Software enables systems to seamlessly run HPC/AI tasks involving data that resides on different nodes with Data Virtualization Service (DVS).
DVS is an I/O forwarding technology for storage subsystems that can flexibly scale with your needs. It utilizes the bandwidth and capacity of the cluster’s high-speed network to deliver enhanced performance and global access to parallel file systems residing on nodes with DVS servers.
Lustre and DVS facilitate system performance at scale as a more efficient alternative to having physical storage network interface cards on every node in the system.
In the case that a node fails, DVS prevents you from losing your job progress by providing continued access to parallel file systems. More specifically, DVS provides load balancing to evenly distribute file system projections, which in turn, enables active-active failover and failback capabilities.

HPE Cray SC User Services Software includes documentation and foundational functionalities to help you better leverage HPC systems and create environments conducive for both HPC and AI workloads (e.g. set up user mode container support with Podman or Singularity).
Receive modules, libraries, and software from Hewlett Packard Enterprise to customize your managed node environment, enhance your system performance, and get more out of your HPC systems.
Selectively install features for power measurement, workload management (WLM) integration and orchestration, compute at scale, and file system integration to meet your organization's unique needs.
Establish the foundation for HPC managed nodes to run diverse workflows, including compiling applications, scheduling batch jobs, and more.