By Bright Staff | November 15, 2021 | Composable Heterogeneous
Join us on November 17th virtually or in person, as Bright Computing, GigaIO, and San Diego Supercomputing Center present a BOF (Birds of a Feather) at Supercomputing Conference 2021. They will foster a technical discussion with the audience around the challenge and promise of composable infrastructures for HPC/AI, and then discuss different approaches for realizing easy-to-use composable systems. This highly relevant HPC industry topic will increase resource utilization, enable resource sharing, boost system performance, and drive down OpEx and CapEx expenditures.
The need to leverage new types of processors and accelerators, implement servers from different manufacturers, integrate with the cloud, extend to the edge, and host machine learning and data analytics applications has placed a demand on organizations to find a more flexible approach to delivery of HPC resources. A paradigm shift of this magnitude presents several challenges that must be overcome to automate a composable heterogeneous computing solution. Composable nodes do not exist until they are composed, but some workload management systems do not submit jobs unless the resources exist. Other workload managers will allow a job to be submitted, but it will never actually be run. Another challenge is how to prevent the workload manager from running jobs on all the composable nodes to the extent that nodes cannot be composed.
Bright's new product will explore DIY options as well as a fully automated approach. Along with GigaIO's FabreX technology, the solution can compose nodes manually using the Cluster Manager Shell (CMSH) or composed on the fly by Bright's Auto Scaler.
By breaking the barrier of the server box, GigaIO's FabreX technology enables the entire rack to be treated as the compute unit. All resources usually located inside the server (GPUs, FPGAs, storage) can now be pooled in accelerator or storage appliances where they are available to all the servers in the rack. Bright Cluster Manager enables users to quickly build and manage heterogeneous high-performance Linux clusters that host HPC, machine learning, and analytics applications that span from edge-to-core-to-cloud.
End-users have a decision to make regarding how they will approach composable infrastructures. They can strike out on their own and piece together their solution, taking a do-it-yourself approach, or they can leverage Bright Computing and GigaIO together, enabling full automation of composable heterogeneous compute.
To learn more about Composable Heterogeneous Computing Environment, join us on November 17th at 12:15 pm CST.