[ad_1]
Michael Feldman |
November 2, 2018 1:43 AM CET
Demand for high-performance space-based computing may be weak at the moment, but for now, Hewlett Packard Enterprise (HPE) has captured the market.
On November 1, the company announced that it would make its space computer available to do real work for astronauts aboard the International Space Station (ISS). The system has been operating on the ISS for one year as an experiment to determine if standard HPC servers could withstand the rigors of space, such as cosmic radiation, power outages, zero-gravities, and temperature fluctuations. . Although the machine was based on standard HPE Apollo 40 technology, it used a "hardening software" to maintain the reliability and robustness of the equipment.
The one-year experience involved the performance of HPC performance tests such as High Performance Linpack (HPL), HPCG suite (High Performance Conjugate Gradients) and NAS parallel performance tests. The results were compared to similar servers on Earth to make sure the Spaceborne computer was working properly. According to HPE, the experiment was a success.
"Our mission is to develop innovative technologies to power the next frontier, on Earth or in space, and to make breakthroughs that we have never imagined before," said Dr. Eng. Lim Goh, Director of Technology and Vice President, HPC and AI, HPE. "After we have learned significant lessons from our first successful experience with Spaceborne Computer, we continue to test its potential by opening HPC capabilities above the cloud to ISS researchers, enabling them to advance the world of space." space exploration. "
This "new level" involves bringing the system into production, at least in a limited sense. The intention is to allow ISS astronauts to conduct onboard "data analysis" related to space research. Presumably, this work would be based on the analysis of high resolution images and video data collected by various ISS experiments.
However, these calculations are somewhat limited, because the space computer can only deliver a teraflop – less than is possible with a single high – end Xeon processor today. Nevertheless, the value here is to provide a verification point to perform a real HPC type work that is not related to Earth-related data centers.
This type of capability becomes a real necessity once a spaceship is moving towards other planets and thus loses the ability to communicate in time with the terrestrial supercomputers. For example, it takes three minutes to send data at the speed of light from Earth to Mars, which is too long for urgent calculations. Even for Earth-orbit scenarios, the real-time analysis work should be done on the spacecraft simply because communication latencies would render remote computations impractical.
If this second phase of the experiment succeeds, the next logical step would be to equip the space station with a more powerful system capable of much more demanding calculations. For now, HPE is focused on using standard technology for these systems. At some point, however, NASA may be forced to explore more personalized solutions because of the power, weight and durability needed.
A timeline of the computer project can be found here.
Source link
