Ho, Hum. Just another day in space with HPE's Spaceborne Computer
February 20, 2018 • Blog Post • Dr. Mark Fernandez, Americas HPC Technology Officer at Hewlett Packard Enterprise
IN THIS ARTICLE
- February 14th marked the six-month anniversary of the launch of the Spaceborne Computer
- During the first six months in space, the supercomputer has passed numerous benchmarking tests, ranking as one of the fastest supercomputers in the world ... and out of this world
Project lead Dr. Mark Fernandez shares updates from the Spaceborne Computer's first 6-month in space
Boring is good.
That may be a surprising statement from a science enthusiast like me. But its an especially satisfying one to make when you're monitoring the performance of HPE's Spaceborne Computer on the International Space Station (ISS).
Last week, marked the six-month anniversary of the launch of the SpaceX CRS-12 rocket, which carried the first high performance commercial off-the-shelf (COTS) computer system into space. And March 14 – Pi Day, a true scientist's holiday will mark six months since the system was installed by ISS astronauts and powered on.
Since then ... nothing but yawns, from back here on Earth (well, except for one unplanned shutdown I'll discuss later). Our Spaceborne Computer is running like a dream, and has even achieved one teraFLOP status – meaning that it can calculate over one trillion calculations per second.
Spaceborne is passing its space test with flying colors. And that's vitally important, because our mission is to operate seamlessly in the harsh conditions of space for one year which is roughly the amount of time it will take to travel to Mars.
A sophisticated onboard computing system capable of extended periods of uptime will be an essential component for any super-long space missions. Think about it: The near-real-time communications astronauts maintain with Earth – from the space station, or even the moon – is not possible from Mars. If a problem occurred, the latency in delivering a message to Earth and having a response arrive back to the spacecraft could be as much as 40 minutes.
That's just too long when a mission could hang in the balance. Our collaboration with NASA to test the efficacy of the Spaceborne Computer in space is a critical step to ensuring the viability and success of a long-range mission. Before we go, we have to know.
In the first six months, the system has passed both the multi-node High Performance LINPACK (HPL) and High Performance Conjugate Gradients (HPCG) benchmark tests, as well as NASAs own benchmark. That confirms that the Spaceborne Computer ranks as one of the fastest supercomputers in the world ... and out of this world, too.
The only down time thus far was a planned, two-hour shutdown while astronauts were replacing an electrical component on another system and a 16-hour unplanned shutdown that occurred in late January due to a smoke detector false alarm. These things happen, even on the ISS – and the Spaceborne Supercomputer performed just as we'd hoped it would.
As part of standard safety precautions, the electrical power to the computer's rack had to be immediately cut off. That meant, of course, that it was not powered down in the recommended manner due to the possibility of this being an emergency. However, once NASA determined no emergency had occurred, its staff at the Huntsville Operations Support Center followed HPE's power-on procedures and returned Spaceborne Computer to its previous state without any issues and no apparent harm.
For each shutdown, the computer system powered back on just as it was designed to do.
So, the ISS keeps on hurtling through space. And the Spaceborne Computer continues to perform majestically, having completed over 2,700 trips around the world. It has been over six months since all teraFLOP supercomputers have been on Earth at the same time.
Nice and boring. Just the way we like it.