Supercomputer to undergo rigorous exam after returning to Earth from the space station

A prototype of the Spaceborne Computer in an ISS mockup
Image: HPE

After spending a year and a half on the International Space Station, Hewlett Packard Enterprise’s experimental space supercomputer is now back on Earth — and the company is poised to do an autopsy on the machine. Ultimately, HPE wants to know just how much wear and tear the machine sustained while in orbit. That way, the company can potentially build supercomputers for use on deep-space missions to the Moon and Mars someday.

HPE had a clear goal for this experiment, known as the Spaceborne Computer: could a regular commercial supercomputer manufactured for use on Earth operate the same in space? A supercomputer can be a valuable asset on a spacecraft, but the space environment can also be an unforgiving place for machinery. Outside of Earth’s thick protective atmosphere, equipment encounters high levels of radiation from deep space cosmic rays and particles from the Sun. Machinery on the ISS can also experience fluctuations in power and temperature that can damage electronics.

Because of these environmental extremes, computers and other electronics bound for space are often built with extra protective hardware. But that route can be expensive and time-consuming, which is why HPE decided to strengthen the Spaceborne Computer with software instead. The supercomputer is equipped with software that can detect if the machine’s various parameters — such as its power, temperature, voltage, etc. — go out of normal bounds. Then the computer would take necessary precautions, like shutting down, to avoid any catastrophic damage. “There were times when we weren’t able to operate and we’d say, ‘Okay, I’d rather you not operate for a short period of time, rather than fail and be destroyed,’” Mark Fernandez, the high-performance computing technology officer at HPE, tells The Verge. “And then when we come back online, we can continue our science.”

Originally launched to the ISS in August 2017, HPE mostly meant to check out a number of standards for the Spaceborne Computer while it was in space, to see if all its various components were working properly. And Fernandez says that goal has been achieved. “We’ve blown away all the naysayers,” he says. “Everyone in the community is really impressed that we went well over a year with successful completion of our benchmarks,” he says. But some scheduling changes on the ISS forced the supercomputer to stay longer than expected on the station, allowing HPE to push the limits of what it could do. During its extended stay, the computer was able to run a program for NASA Langley, testing out the same software a spacecraft would use for guidance and navigation when it enters Earth’s atmosphere and descends down to the surface.

The Spaceborne Computer returned to Earth on Monday aboard a SpaceX Dragon cargo capsule. Now, a boat is carrying that spacecraft and all of its contents to shore. The supercomputer will be the last thing taken out of the Dragon when it gets back, and HPE will receive it on July 8th in Houston, Texas. There, the company will analyze the computer and take note of all the components that have failed and how “old” they look. “Because they’ve been in space for a year, do they appear to be a year old like their Earthbound counterparts? Or do they appear to be older?” says Fernandez.

Based on what the engineers find, HPE will use that information for the next iteration of the Spaceborne Computer. Fernandez says the company hasn’t spoken to anyone about using the supercomputer in actual missions yet, but he hopes HPE can get the technology to a place where it could be a viable tool in the future. Being able to run software and crunch numbers in real time on a supercomputer is going to be particularly useful the farther away from Earth that astronauts travel, when communication with ground control can be super delayed. And if a supercomputer doesn’t need to be outfitted with too much heavy equipment, taking one on a trip to the Moon or Mars becomes more of a possibility. “Any experiments that you design today in an Earthbound laboratory are probably going to use relatively new computer systems,” Fernandez. “And that’s what we want to be able to take with us on these missions to Mars.”

Comments

Hopefully they built this thing to be fully modular so astronauts can easily repair it themselves. It would suck to have your supercomputer blue screen when you’re millions of miles from home.

If it was a production box, they’d probably just take a second and even third copy, and have earth-computers be their backup. At least until we start heading for other planets

The spares would be getting hit with radiation while not in use though so are almost as likely to fail in the same time period. Obviously it decreases the chance of all 3 failing but it’s not by anywhere near the amount it would on earth with one live system and two switched off backups.

You could say the same of the spare modules:

Hopefully they built this thing to be fully modular

I hope it studied. Surprise exams suck.

It is the Phoenix Force now, beware.

I don’t really get what there is to learn here. Just because you call a computer "super", it’s still just a computer, which have been going in and out of space for very long time.

But not Commercial Off The Shelf [COTS] systems, which this basically is. Per the article:

Because of these environmental extremes, computers and other electronics bound for space are often built with extra protective hardware. But that route can be expensive and time-consuming…

The whole point of the experiment is that this is NOT the type of purpose-built hardware that is normally sent into space. Now that they’ve done it, they’ll get to see how space treated it.

Still cheaper than the new Mac Pro.

View All Comments
Back to top ↑