When SpaceX’s next mission launches on August 14, it won’t just be carrying supplies to the International Space Station (ISS). NASA has partnered with HP to perform a long-term experiment on how high-end, modern computers will perform in space compared with on Earth.
Space, as you may or may not know, is absolutely murder on electronics. Cosmic rays are sufficiently plentiful that astronauts since the 1960s have reported seeing flashes of light when their eyes are closed, and there have been no flashes within the capsule, shuttle, or space station. It’s believed that this is caused by cosmic rays passing through the retina. The ISS isn’t protected at all by Earth’s atmosphere, and while it still receives substantial protection from Earth’s Van Allen belts, the total radiation exposure that an astronaut on the ISS receives in a week is equivalent to what a human on Earth receives in a year. At one point, the ISS passes through the far edge of the inner Van Allen belt, and the crew receives 30x more radiation than they would otherwise during this interval.
Electronics, historically, don’t cope with this well. When NASA builds a probe or a rover, they use CPUs and components that are laughably slow by modern standards. The flip side to this is that the components in question have been extensively over-engineered and hardened against radiation. This is a real problem; L2 cache errors increase measurably when you run code on a server at high altitude compared with at sea level. But while this is a known problem in general, NASA hasn’t tested how conventional, off-the-shelf electronics will behave in space over the long-term, particularly if measures are taken to reduce the computer’s speed as the station passes through higher amounts of radiation.
Here’s how the experiment will work. Two Spaceborne Computers will be sent to the ISS, courtesy of HP and NASA. Two more machines will be configured as control systems and will remain at HP. The Spaceborne Computers will be used to discover whether it’s practical to operate commercial off-the-shelf (COTS) systems in a high-radiation environment. The goals are to run compute and data-intensive applications, monitor power consumption, and dynamically tune it when required.
We don’t know which Apollo systems HP will use, but here’s a representative sample of the product line.
We don’t know much about the specs of the servers themselves. But they’re reportedly based on HP’s Apollo family, and they won’t have any special radiation shielding to protect them. This isn’t just a lark; if NASA is going to send people to Mars, it needs to know what kind of computing horsepower it can send with them. Calculating complex trajectories, orbital insertion maneuvers, and managing a spacecraft over the months it would take to arrive (travel time to Mars is typically estimated at 6-8 months) are difficult tasks. Understanding whether existing conventional computers can function in the relatively shielded orbit of the ISS will tell us important things about how they are likely to function on a long space mission, and what kind of shielding or countermeasures may be required to guarantee proper computer function over the long haul.
We’ve already seen rovers like Curiosity need to switch to its backup computer after several years on Mars. Given that every pound of computer hardware is a pound that can’t be allocated to something else, NASA undoubtedly wants to understand where the limits of safety are so it can plan for appropriate missions – provided, of course, that Congress ever allocates enough money for a manned mission to Mars in the first place.