Skip to main content


DM-7-E.jpgThe DM Cube is an extremely miniaturized, high performance, parallel computer processor system that is applicable to a wide variety of technologies. DM cube is a scalable cluster of high performance COTS processors with a high speed interconnect with operation under the control of a reliable, radiation-hardened, system controller and platform, and application-independent fault tolerant middleware.  DM Cube takes advantage of the availability of small, light-weight, low-power, COTS Computer-on-Module Gumstix™, technologies allowing significant reductions in size, weight, and power consumption while increasing processing performance. DM has the following features:
  • Based on small, light-weight, low-power, low-cost, Gumstix™ COM (Computer-On-Module)-based, high-performance onboard payload processing--1600 DMIPS, 5.6 grams, ~ 2 watts
  • Operated under the control of a reliable system controller
  • Platform and application-independent
  • Highly scalable (to >24 processors)
  • Requires low overhead (<10% throughput & memory)
  • Extremely flexible--user-configurable fault tolerance includes hybrid replication, temporal and spatial self-checking and Triple Modular Redundancy  for critical functions and Algorithm-Based Fault Tolerance
  • Peak throughput density of 300 million (MOPS)/watt
  • Optional fault tolerant DM Middleware (DMM)

        DM Cube Specifications Per Processor

Processor Architecture Clock Rate Memory Max Power Usage Digital Interface Types Mode Manufacturer
Cluster Nodes:
ARM Cortex A8
600 MHz-1.4 GHz 512 MB RAM
512 MB Flash
1.5 Watt UART, I2C, USB, SPI, CAN, PHY Network Node Gumstix™
Cluster Backbone 24 MHz 1 MB 2.0 Watts PHY, UART, SPI Network Backbone & Management MSU

Middleware (Honeywell International)

DM software is an architecture and software framework that enables COTS-based, high performance, scalable, cluster processing systems to operate in inhospitable environments by providing software-based SEE-tolerance enhancement.  DM is available with user-configurable fault tolerant options spanning the mission level to the application level.  Fault tolerant execution includes replication, i.e., temporal and spatial self-checking (SC) and triple modular redundancy (TMR), combined with more computationally-efficient Algorithm-Based Fault Tolerance (ABFT). DM can execute multiple missions sequentially or concurrently based on resource availabilityA predicted 5-year probability of incorrect computation less than or equal to 0.005 in a Low Earth Orbit (LEO) environment is possible.

DM-7 Flight Validation Experiment

DM-7 (NASA Technology Readiness Level-7) experiment will be installed on the Nanoracks Exterior Platform (NREP) on the International Space Station to test DM in the space radiation environment.  DM-7 launched on HVT-6 from Japan on December 9, 2016 and has operated successfully on orbit.