Subtitle Integration , Development and Performance of the 500 Tflops Heterogeneous

The Air Force Research Laboratory Information Directorate Advanced Computing Division (AFRL/RIT) High Performance Computing Affiliated Resource Center (HPC-ARC) is the host to a very large scale interactive computing cluster consisting of about 1800 nodes. Condor, the largest interactive Cell cluster in the world, consists of integrated heterogeneous processors of IBM Cell Broadband Engine (Cell BE) multicore CPUs, NVIDIA General Purpose Graphic Processing Units (GPGPUs) and Intel x86 server nodes in a 10Gb Ethernet Star Hub network and 20Gb/s Infiniband Mesh, with a combined capability of 500 trillion floating operations per second (TFLOPS). Applications developed and running on CONDOR include large-scale computational intelligence models, video synthetic aperture radar (SAR) back-projection, Space Situational Awareness (SSA), video target tracking, linear algebra and others. This presentation will discuss the design and integration of the system. It will also show progress on performance optimization efforts and lessons learned on algorithm scalability on a heterogeneous architecture. INTRODUCTION The Affiliated Resource Centers (ARCs) are Department of Defense (DoD) Laboratories and Test Centers that acquire and manage High Performance Computing (HPC) resources as a part of their local infrastructure, but share their HPC resources with the broader DoD HPC user community via the High Performance Computing Modernization Program (HPCMP) which coordinates allocation of their HPC resources. In order to provide tomorrow’s Air Force with massively parallel and scalable HPC applications, the software must be developed on large clusters. Unlike typical HPC clusters, all AFRL/RI clusters allow for interactive development and testing. In 2010, the AFRL Information Directorate won a two-million-dollar project, sponsored by the HPCMP, and built the Condor Cluster, which is DoD’s largest interactive super computer as of November 2011. The Condor cluster consists of 84 Servers (2U Dual six-core Intel Westmere 5660, 24 or 48 GB RAM) each with 2 GPGPUs (NVIDIA C1060, C2050 or C2070s) [2]. The heterogeneous cluster has 22 Play Station 3s (PS3s) connected to each of the 78 server nodes (1716 PS3s in total). Figure 1. Condor Cluster: DoD’s largest interactive HPC. The long-term goal of AFRL/RI’s high performance computing research is to provide the warfighters with Secure Embedded HPC (SEHPC) of the highest computing performance, under the Size-Weight-and-Power (SWaP) constraints. At the time when it was built, Condor was the largest, fastest and most energy-efficient interactive HPC in the Department of Defense.