Accounts and Allocations
PPPL scientists and engineers and their collaborators can apply here for accounts on three different computer systems.
The Traverse cluster is composed of 46 IBM POWER9 nodes with four NVIDIA V100 GPUs per node. It is primarily for users who are ready to efficiently utilize many GPUs for whole-device modeling (WDM) of plasma systems, or for developers who wish to have access to a platform they can use to build community codes for this purpose.
Stellar is a heterogeneous cluster with more than 50,000 Intel and AMD cores together with specialized visualization and computing nodes with Nvidia A100 and V100 GPUs. It is primarily for users who are ready to efficiently utilize many CPUs for whole-device modeling (WDM) of plasma systems, or for developers who wish to have access to a platform they can use to build community codes for Perlmutter.
Portal is the entryway to a heterogeneous cluster with more than 6,000 Intel and AMD cores. Like Stellar, there are additional special services available through portal, including high-memory nodes, direct access to experimental databases, and an on-demand TRANSP server for WDM of tokamaks. Portal provides a wide range of development and "pre-flight" services (including continuous integration and host services for software repositories), as well as single- and multi-core applications that require teraflop-scale resources.