Aitech, a maker of rugged computers for military, aerospace and space applications, has tapped Nvidia's Jetson TX2i system-on-module (SoM) for a new radiation-characterized system, it announced recently. The Aitech S-A1760 Venus is a commercial off-the-shelf (COTS) system that can be used for spacecraft and small satellites and takes advantage of around 1 FP32 TFLOPS of "AI performance," as Nvidia puts it.
There is a growing need for advanced imaging and data processing in various space applications, but equipping a small satellite with a high-performance, rad-hardened computer is extremely expensive, since tiny satellites are supposed to be light and tiny. This is where Aitech’s S-A1760 Venus system comes into play.
According to the Aitech, the S-A1760 Venus is targets "short duration spaceflight" as well as near earth orbit (NEO) and low earth orbit (LEO) satellite applications. Its best use is "video and signal processing in distributed systems."
At the heart of Nvidia's Jetson TX2i SoM sits the company's Tegra X2 system-on-chip (SoC) that integrates two or four Cortex-A57 general purpose CPU cores. It also uses the GP10B GPU, which is based on the Nvidia's Pascal architecture featuring 256 CUDA cores that offer up to 1.26 FP32 TFLOPS performance (around 1TFLOPS in case of the S-A1760 Venus) for AI or image processing. The Tegra X2 can also connect up to six cameras (12 via virtual channels) and encode/decode up to 1/2 4Kp60 or 4/20 1080p60 HEVC video streams concurrently.
Nvidia's Tegra X2 SoC is not radiation-hardened, but with proper protection it can still be used for some space applications. Aitech's S-A1760 Venus small-form-factor system has passed the Series 300 level qualification standard that identifies the rad-tolerant needs of space components and systems not used in deep space or long-haul applications.
The Jetson TX2i module for industrial applications and harsh environments carries 8GB of 128-bit LPDDR4 memory and 32GB of eMMC 5.1 storage, which appears to be good enough for space applications that tend to be custom made and spend resources economically. Meanwhile, the Aitech’s S-A1760 Venus has GbE, UART Serial, USB 2.0, CANbus, and a DVI/HDMI output. Video capture capabilities include an HD-SDI input with a dedicated H.264 encoder and eight RS-170A (NTSC)/PAL composite channels.
"With the growing need for advanced imaging and data processing throughout space-rated applications, transitioning our powerful GPGPU-based AI supercomputers to this industry was a logical choice," said Dan Mor, Director, Video and GPGPU Product Line at Aitech, in a statement. "By validating these space-rated, COTS-based systems with a clearly defined and recognized qualification level, we're helping lead the charge in the development of commercial space applications and small sat cluster innovations."
Nvidia in Space
One interesting thing to note about Aitech’s S-A1760 Venus system is that it will be the first Nvidia SoC-based solution that will power devices like satellites. But it's not the first time we've seen Nvidia technology that's space-ready.
Select Lenovo's ThinkPads are certified for use on the International Space Station, and historically these PCs have used graphics processors from ATI Technologies (now AMD) and Nvidia. Of course, displaying graphics and perhaps doing some simulations is a different than powering a satellite or a unit within a spacecraft.