Multi-processors by the numbers: mathematical foundations of spaceflight grid computing

The mathematics of connectivity elucidates predictive insights about multi-processor architectures. Space missions, can (and should) benefit at a level where avionics designers have the most leverage. Combining classical results with fresh research unveils new methods for maximizing fault tolerance and throughput, and for minimizing latency and cost. We illustrate and explain how to exploit objective functions corresponding to feasible regions of design and operation. The payoff: platforms that optimize grid computing among spaceborn processors. The grid may span hundreds - or even thousands - of nodes. To put a face on the mathematics of connectivity, we cite key software and hardware enablers for grids of spaceborn processors. Two such technologies stand perched on the brink of operational readiness: i) tunable multi-processor topologies, and ii) the vertical cavity surface emitting laser (VCSEL). Other enablers, such as iii) multi-processor partitioning of flight software, await the fruits of efforts by investigators