The Heterogenous Advanced Architectures Platforms (HAAPs) Team explores pre-production high performance computing architectures in collaboration with vendors, researchers, and code developers to assess their suitability to become the next extreme-scale computer architecture for the NNSA Tri-labs.
The HAAPs team collaborates to deliver functional platforms for multiple pre-production systems running research software in exploratory configurations. Each system has unique architectural features relating to their CPU’s, GPU’s (or other accelerators), high-speed networking, memory subsystems, or storage … or all of the above! Working with prototype hardware and software stacks provides a unique opportunity for our team to develop system management methodologies and to work with vendors to drive these to production. In addition, we explore new technologies and solutions in areas such as machine learning, containerized workflows, and HPC in the cloud.
Our team also manages small instances of extreme-scale systems sited at other national laboratories, enabling tool development, feature exploration, and code preparation for those systems.
The HAAPs team and the LDMS team are collaborative partners in innovating HPC management and operational methodologies to improve the efficiency and performance of systems. The exploratory nature of our platforms makes them innately suitable for designing, developing, and testing new approaches for HPC.
