adaptive.hpc/ai/ml-as-a-service
running e4s on
adaptive.HPC/AI/ML-as-a-service
Finding a good machine image for AI/ML and HPC tasks optimized for a cloud platform like AWS can be a daunting task. The Extreme-scale Scientific Software Stack (E4S) [https://e4s.io] provides a platform for tool integration and deployment on AWS. E4S features 100+ tools for scientific computing such as Trilinos, TAU, HDF5, PETSc built with the MVAPICH MPI distribution that is optimized for the Elastic Fabric Adapter (EFA) for efficient intra- and inter-node communication. E4S supports AI/ML packages such as TensorFlow, PyTorch, JAX, and Python tools including Jupyter notebooks. ODDC enables launching jobs on multiple nodes through a web browser and supports a responsive remote desktop environment in E4S based on VNC. This is ideally suited for 3D tools in the distribution such as TAU, ParaView, and VisIt. E4S and ODDC lower the barriers for users to schedule and deploy their AI/ML and HPC workloads on multiple cloud platforms using a consistent environment that provides a trusted and performant software stack.
Adaptive.HPC/AI/ML-AS-A-SERVICE
End-to- End Generative AI Training for SMBs and Large Enterprises
adaptive.hpc/ai/ml-as-a-service includes over 120+ open-souce applications with additional frameworks, tools and packages optimized for GPUs and AI/ML workloads. The full Adaptive Computing Technolgy Stack is included and contains the Moab HPC Suite and ODDC (On-Demand Data Center) software, and cloud-based or on-premises supercomputing infrastructure. Adaptive Computing offers the full technology stack required for Enterprise AI delivery at an all-inclusive monthly price that is vastly more cost-effective than other Generative AI delivery options in the market.
adaptive.hpc/ai/ml-as a-service offers 3 deployment options: hosted in Adaptive Computing’s on-premises data center, hosted in your data center, or hosted in the cloud provider of your choice. The top 4 major cloud providers, AWS, Azure, GCP, and OCI are pre-configured into ODDC and accessible through a web browser and user-friendly GUI.
adaptive.hpc/ai/ml-as-a-service includes the Extreme-scale Scientific Software Stack (E4S): a curated, Spack based collection of HPC and AI/ML packages including over 120+ open-source software applications and optimization tools, assembled under a U.S. Department of Energy grant. The DOE Exascale Computing Project E4S Project Lead, Sameer Shende and ParaTools collaborated with Adaptive Computing to build custom images for all top 4 cloud providers (AWS, Azure, GCP, and OCI)
Users can effortlessly launch E4S and HPC/AI/ML workflows from the ODDC web interface into any cloud providers or all of them with just a few clicks. E4S and the Adaptive Computing Technology Stack provide a consistent platform for both multi-cloud and on-premises HPC/AI/ML deployments using a trusted and performant software stack. ODDC enables launching jobs on multiple nodes through a web browser and supports a responsiveremote desktop environment in E4S based on VNC.