[ad_1]
One year ago, Hewlett Packard Enterprise sent an off-the-shelf supercomputer up to the International Space Station, to see if its mass-produced hardware could survive, basically unmodified, in the harsh environment of space. Now NASA and the computer company are declaring the experiment a success—even though nearly half of its hard disks failed after getting fried by solar radiation.
The experiment was designed to test the performance of the HPE Spaceborne Computer, which is comprised of 32 separate cores that work together and can run operations 30 to 100 times faster than a standard iPhone or tablet. Despite the radiation blast, the computer kept working and none of the test data was lost. “They learned a few things and they demonstrated that the system works in space,” says NASA’s David Hornyak, who manages technology demonstration projects on ISS. ”It gives NASA the option to purchase a capability we may need in the future rather than developing it ourselves.”
The computer also remained operational even thought it lost its primary upload and download connection to NASA up to eight times a day for periods ranging from 3 seconds to 20 minutes. That kind of independence might prove useful on an eventual moon base or mission to Mars, to work around communications delays back to Earth. NASA’s IT experts are also preparing for a potential overload of signals coming into NASA’s Deep Space Network, a system of satellite dishes across the globe that sends and receives signals from various spacecraft. The Spaceborne computer could help by cutting down on the amount of traffic generated by the ISS.
HP’s Mark Fernandez, who oversaw the project, says the idea behind putting a supercomputer in space was to help the astronauts do more science on their own. “There’s a lot of 4K cameras and videos on ISS looking for something,” he says, for example hunting through reams of data for a type of cloud, lightning strike or temperature gradient in the ocean. “We should do that kind of general-purpose image processing on board and save that bandwidth for other things.”
The computer can also compress data from science experiments onboard the station and then send it down to NASA computers below. Eventually, astronauts and explorers on the Moon or Mars might want a supercomputer to figure out problems without waiting for NASA to respond.
To protect the computer’s internals, engineers at HPE developed software that could serve as a kind of shield. Because they couldn’t install radiation detectors inside the computer, they identified components that were most vulnerable to radiation and set limits for their operating conditions. If things got too hot, for example, whether from radiation or a problem with the power supply, the component automatically went into a protected safe mode.
The HPE software worked pretty well, but it wasn’t perfect. Of the 20 solid state disk drives, nine failed, a rate that Fernandez calls alarming. The computer also experienced seven “bit flips,” which is when the digital ones turn to zeros, unprompted. Fernandez says the computer was able to identify what happened, throw out the corrupted data, and get a new batch of the same information. But if that wasn’t enough for the IT folks to worry about, the device also experienced four power interruptions, either from radiation or from problems with the station’s solar panels, Fernandez says.
Any future long-duration space mission might need to have tweaks to the software as well as some kind of additional shielding made from lead, water or new kinds of ceramic or composite material. But the year-long experiment gave the software and hardware designers a boatload of data to work on. “We’re going to look into that and consider alternatives,” Fernandez said.
Now that the year-long test is complete, the computer’s first task will be to test some NASA descent and landing software that requires a large number of computations. It will remain on the station until its scheduled return next March.
More Great WIRED Stories
Source link