Intel’s latest chip is designed for computing at the edge

Intel’s latest chip is designed for computing at the edge

As we develop increasingly sophisticated technologies like self-driving cars and industrial internet of things sensors, it’s going to require that we move computing to the edge. Essentially this means that instead of sending data to the cloud for processing, it needs to be done right on the device itself because even a little bit of latency is too much.

Intel announced a new chip today, called the Intel Xeon D-2100 processor, to help customers who want to move computing to the edge. It’s part of an effort by the chip giant to stay ahead of emerging technology trends like edge computing and the Internet of Things.

Computing at the edge has some unique space and power requirements that Intel has tried to address with this announcement. For starters, it provides a stand-alone system on a chip (SoC). This means everything you need is built into into the chip including compute, networking and storage. It’s also low power, which might be necessary to run an edge computing device without the benefit of a data center power structure.

In a blog post announcing the new chip, Jennifer Huffstetler, vice president and general manager of data center product management in the Data Center Group at Intel, pointed out the growing need for this type of architecture. “By expanding the capabilities of the data center outward to the network edge, solution providers can process more data closer to endpoint devices, reducing application latency and opening up a whole new world of potential services and experience,” Huffstetler wrote.

She added that the SoC gives customers an “integrated, hardware-enhanced network, security and acceleration capabilities in a single package.” The chip achieves all of this by packing a lot into a small package including up to 18 ‘Skylake-server’ generation Intel Xeon processor cores integrated with up to 100 Gbps of built-in cryptography, decryption and encryption acceleration. Intel calls this ‘QuickAssist Technology.’

The company sees this being particularly useful for new 5G technologies being developed for smartphones like augmented and virtual reality applications and autonomous driving. They also see it being useful for communications network use cases like virtual private networks and software-defined wide area networks and also certain cloud workloads that require processing close to the edge such as content delivery networks.

Intel has already built a partner network for the new chip with a variety of companies including including Dell EMC, Ericsson, NEC, NetApp and Palo Alto Networks.

And if you’re concerned about the impact of the Spectre and Meltdown exploits, Intel reports that it has built in the latest patches into the new chip. (Although it’s worth noting that Intel had trouble with its initial attempts at patching the exploit.)

SpaceX confirms it lost the center core of the Falcon Heavy

SpaceX confirms it lost the center core of the Falcon Heavy

SpaceX pulled off quite the feat today when it launched the Falcon Heavy rocket.|

What’s more, it landed the two flanking boosters in perfect synchronized formation. But the fate of the core booster was unclear; now it appears that the center booster, which was supposed to land on a drone ship, was lost.

Elon Musk said on a conference call with reporters that the launch “seems to have gone as well as one could have hoped with the exception of center core. The center core obviously didn’t land on the drone ship” and he said that “we’re looking at the issue.” Musk says that the core ran out of propellant, which kept the core from being able to slow down as much as it needed for landing. Because of that, the core apparently hit the water at 300MPH, and it was about 100 meters from the ship. “It was enough to take out two thrusters and shower the deck with shrapnel,” Musk said. That should be worth seeing on video: “We have the video,” Musk confirmed, “it sounds like some pretty fun footage… if the cameras didn’t get blown up as well.”

It’s been suspected that the core was lost since just after the other two rockets landed successfully. The video feed from the center core cut out, and we didn’t hear anything official in the intervening three-plus hours. Audio from an unpublished YouTube video also seemed to confirm the loss of the center core; “we lost the center core” can be clearly heard on the recording. Without any context, though, it was hard to know for certain if they were talking about communications or the rocket itself.

This shouldn’t take away from SpaceX’s spectacular achievement today. The goal of this demo launch was to prove that the Falcon Heavy was ready for flight, and it more than accomplished that. Landing the first stages is always a secondary objective, and no one should overlook the fact that the company was able to land two of the three boosters at once. SpaceX will likely learn quite a bit from this failure, and be better prepared to land all three boosters next time, which will probably be in three to four months.

Nathan Ingraham contributed to this report.

Vision Research Phantom camera captures 1080p video at 11,750 frames per second

Vision Research Phantom camera captures 1080p video at 11,750 frames per second

In the never-ending pursuit of video capture at higher frame rates than ever before, the team at Vision Research has outdone itself with the latest camera in its Phantom lineup, the Phantom v2640.

The CMOS sensor inside the Phantom v2640 is only four megapixels. But what it lacks in resolution, it makes up for in just about every other category. At full resolution (2,048 x 1,952 pixels) it captures video at 6,600 frames per second (fps) and almost doubles that to 11,750fps at 1080p resolution. The dynamic range of the Phantom v2640 is also the highest in any Phantom camera to date, which is a critical feature considering how fast the minimum shutter speed of the Phantom v2640 is, at 142 nanoseconds. Yes, you read that right — nanoseconds.

Made with researchers, scientists, and engineers in mind, the Phantom v2640 comes in both color and monochrome models. The color model maxes out at 11,750fps, while the monochrome version can use a specialized mode to bump up the frame rate to 25,030fps, although that comes at the cost of resolution, dropping it down to just one megapixel.

As you would expect with so many frames, storage is a major factor in recording limits. It’s for this reason Vision Research is offering the Phantom v2640 with up to 288GB of onboard memory. If more space is needed, it’s also compatible with Vision Research’s 1TB and 2TB CineMags. Offloading data is done using a 10Gb Ethernet connection.

As for design, the camera looks like most other Phantom cameras released by Vision Research. It features a block-like design with a basic handle and massive vents on both sides to keep the internals cool.

Pricing for the Phantom v2640 hasn’t yet been announced, but it’s safe to assume this beast won’t come cheap. A recent predecessor of the Phantom v2640, the Phantom v2512, retailed for $110,000 at launch.

That said, if you have a hundred grand burning a hole in your pocket — or you just want to find out more specs of this thing — head on over to Vision Research’s website where you can find out more information and detailed instructions on how to pre-order the Phantom v2640.