High Performance Dependable Multiprocessor II

Abstract

With the ever-increasing demand for higher bandwidth and processing capacity of today's space exploration, space science, and defense missions, the ability to efficiently apply Commercial-Off-The-Shelf (COTS) processors for on-board computing has become a critical need. In response to this need, NASA's New Millennium Program (NMP) office commissioned the development of Dependable Multiprocessor (DM) technology for use in science and autonomy missions, but the technology is also applicable to a wide variety of DoD missions. The goal of the DM project is to provide spacecraft/payload processing capability 10x -100x what is available today, enabling heretofore unrealizable levels of science and autonomy. DM technology is being developed as part of the NMP ST8 (Space Technology 8) project. The objective of this NMP ST8 effort is to combine high-performance, fault tolerant, COTS-based cluster processing and fault tolerant middleware in an architecture and software framework capable of supporting a wide variety of mission applications. Dependable Multiprocessor development is continuing as one of the four selected ST8 flight experiments planned to be flown in 2009. 1, 2, 3There are three key problems that need to be overcome in order to fly COTS in space: 1) an effective approach for handling SEUs (Single Event Upsets) in high performance cluster processors, 2) handling thermal issues associated with state-of-the-art COTS components, and 3) achieving high power efficiency (throughput per watt). DM technology solves all three problems. DM solves the SEU problem by combining cluster management software with SEU tolerance-enhancing software in a flexible, efficient, integrated DM middleware suite. DM solves the thermal 1 1-4244-The project formerly was known as the Environmentally-Adaptive FaultTolerant Computing (EAFTC) project. issue by mining the ruggedized, conductive-cooled, COTS airborne embedded processing domain. DM solves the power efficiency problem by mining the high performance, low power mobile computing processing domain.Recently, the DM project successfully passed several key NMP ST8 project milestones: the TRL5 (Technology Readiness Level 5) technology validation demonstration, the Experiment Preliminary Design Review, and the NASA Non Advocate Review. Passing the TRL5 milestone qualified the DM project for advancement to flight system development status. The ST8 project has passed its Preliminary Design and Confirmation Reviews and is now moving into the Implementation Phase.This paper describes the DM technology, the technology validation experiments and demonstrations achieved to date, the plans for the TRL6 technology validation effort, and the plans for the TRL7 flight validation, and provides an update on the status of the project TABLE OF CONTENTS

Model-Based Reconfiguration Planning for a Distributed On-board Computer

“…There are more complex systems involving COTS processors, such as [14] and NASA's Dependable Multiprocessor [15,16], where a cluster of COTS processing nodes is controlled by a reliable computer. In these systems the configurations are not precalculated, but rather every job is scheduled online to a currently available node.…”

Fault‐tolerant on‐board computing for robotic space missions

“…The Remote Exploration and Experimentation project conducted at NASA was among the first to consider putting a COTS‐based parallel machine into space and address the resulting problems related to application‐adaptive fault tolerance . More recently, NASA's Millennium ST‐8 project developed a ‘ Dependable Multiprocessor ’ around a COTS‐based cluster using the IBM PowerPC 750FX (IBM, Somers, NY, USA) as a data processor, with a Xilinx VirtexII 6000 FPGA coprocessor (Xilinx, San Jose, CA, USA) for the support of application‐specific modules for digital signal processing, data compression, and vector processing. A centralized system controller for the cluster is implemented using a redundant configuration of radiation‐hardened Motorola processors.…”

Small, light-weight, low-power, low-cost, high performance computing for CubeSats

“…DM and DM CubeSat technology development has been regularly documented in open literature [1], [2], [3], [4], [5], [6], [7], [8], [9]. A brief overview of DM and DM CubeSat technology is provided in this introductory section to provide a basis for what is discussed in the remainder of the paper.…”

“…The DM TRL6 technology validation effort included the demonstration of low overhead and ease-of-use for MPI (Message Passing Interface )-based parallel applications. Summaries of the TRL6 technology validation effort can be found in [5], [6], [7], [8], and [9]. A comprehensive discussion of DM technology development from TRL4 through TRL6 is provided in [6].…”