The next-generation Trinity supercomputing platform project has recently achieved two important milestones. Trinity is an advanced computing platform for stockpile stewardship that is being architected and procured by the Alliance for Computing at Extreme Scale (ACES), a partnership between Sandia and Los Alamos National Laboratories, under the NNSA Advanced Simulation and Computing (ASC) Program. The first phase of Trinity was delivered and installed at Los Alamos in Fall 2015, and at the end of November 2015, the ASC Level 2 Milestone for Trinity System Integration Readiness of the Phase 1 system was successfully completed. The Phase 1 system consists of compute nodes based on Intel Haswell processors, whereas the Phase 2 system will be based around Intel Phi (Knights Landing) many-core processors. Following the delivery and integration of Trinity Phase 1, acceptance testing and performance assessment was completed in December 2015, leading to formal acceptance of the Phase 1 system before the end of 2015. Importantly, the system performance exceeded the procurement requirement by more than 25%, as measured by a sustained system performance metric of 516 (relative to 400 required) based on representative proxy applications. The next steps for Trinity will be further user testing of Phase 1 during an open science period, followed by initiation of stockpile computing campaign work, and delivery of the Phase 2 system later in fiscal year 2016.
January 1, 2016