feature article
Subscribe Now

The Moving Picture Show

These guys should meet. The other day we talked about Ceva’s MM3000 cell-phone processor (Feb 16, “Viddy This, O My Brothers”). This week we’ve got an equally interesting device from startup company Movidius that makes handheld video even cooler. Brace yourselves for Myriad, the video-editing deck in your hand.

In some ways, the two companies are related, so it’s not surprising they would approach the same market. Ireland-based Movidius is staffed, in part, with exiles from Parthus, the company that eventually became the Irish/Israeli firm Ceva. Apart from their shared accents, though, the two companies and their chips are very different.

For starters, Ceva licenses IP cores while Movidius makes real chips. They call it Myriad, probably because it can do countless different things. To an engineer, Myriad is a massively parallel, superscalar, 8-way VLIW machine. To a cell phone user, it’s a way-cool method to capture, edit, and massage high-definition video taken with your cell-phone camera.

Myriad is designed to be a coprocessor, an add-on to an existing cell-phone design that works alongside the phone’s main processor (or processors), handling video-related tasks. That makes Myriad fairly simple to integrate because it doesn’t require a complete overhaul of the cell phone design or a rethink of its hardware architecture. Assuming handset makers can find room for the chip’s modest 8×8-millimeter footprint, they’re mostly home free.

Once it’s in and integrated, the Myriad chip takes over image-capture, -editing, -viewing, and –sharing tasks. On the capture side, Myriad can do real-time image stabilization and zoom (by ignoring outer pixels and oversampling the inner ones), reduce noise in low light conditions, and even do a kind of hyper-speed 120 frames/sec capture that makes for slow-motion playback.

Once the video is captured, the chip can clean up audio and video artifacts. Users can edit the video, join or split clips, add transitions, fool around with color effects, and generally play micro auteur. All this is done directly by the Myriad chip, with no help or interference from the cell phone’s main processor. The handset maker “merely” adds a user interface to Myriad’s features and lets the chip do all the heavy lifting. For Android or other open operating systems, that’s a pretty straightforward task.

Under the Hood

Inside, Movidius’s Myriad is as complex as its outside is simple. The chip is sort of like a miniature Cell processor, with eight identical VLIW processors overseen by a ninth master processor. In the case of Myriad, the eight VLIW cores are Movidius’s own design: a 128-bit VLIW machine that handles both integer and floating-point data. All eight are geared toward the kind of SIMD vector operations that are typical of video processing. Each individual core can handle one to eight parallel operations, for a total of 64 operations across all the cores. Altogether, the chip delivers 20 GFLOPS, according to the company.

Typically, all 64 operations would be the same, but not necessarily. Relatively lightweight tasks (like processing JPEG still images) might use only one or two of the eight VLIW processors while the others sleep to save power.

The ninth processor is unlike the others; it’s a SPARC knock-off that distributes tasks among its eight minions and oversees the chip’s many peripherals. Camera and LCD interfaces, memory controller, serial interfaces, and more make up the chip’s connections to the outside world.

As far as clock speed, Movidius is a bit shy about mentioning numbers except to say that it’s “a couple hundred” MHz. That’s pretty slow by modern CPU standards, but that’s partly the point. The chip is so wide (in terms of data path) that it can broadside a lot of video data at once without resorting to power-sapping clock frequencies. It’s currently fabricated in TSMC’s 65-nm process, which also doesn’t push the envelope.

Programming this beast must be a chore, but we’ll never know because Movidius does all of that in-house. The Myriad chip is provided with one of two software libraries (your choice) and a written API specification that tells you how to call functions from your host operating system. Apart from that, the magic all happens behind the curtain. Myriad is the proverbial black box.

The first chips are sampling now, with production coming around Q3 and end-user products in early 2011. Movidius offers the chip with two different software loadouts, depending on whether you want HD video or standard definition, and whether or not the image sensor is connected directly to the Myriad chip. Future versions will support two image sensors for 3D image capture, at which point the processing – and the applications – get a lot more complex and interesting. Because the differences are all in software, Movidius could add, remove, or tweak features over time to keep up with customer requests, market changes, or just basic feature creep. The hardware provides a base platform for the programmers to go nuts. And all without a lot of design work from the handset maker. Check out the spiffy video at www.youtube.com/movidiuscorp.

Leave a Reply

featured blogs
Oct 23, 2020
The Covid-19 pandemic continues to impact our lives in both expected and unexpected ways. Unfortunately, one of the expected ways is a drop in charitable donations. Analysts predict anywhere from a 6% decrease '€“ with many planning for a bigger decline than that. Also, mor...
Oct 23, 2020
[From the last episode: We noted that some inventions, like in-memory compute, aren'€™t intuitive, being driven instead by the math.] We have one more addition to add to our in-memory compute system. Remember that, when we use a regular memory, what goes in is an address '...
Oct 23, 2020
Any suggestions for a 4x4 keypad in which the keys aren'€™t wobbly and you don'€™t have to strike a key dead center for it to make contact?...
Oct 23, 2020
At 11:10am Korean time this morning, Cadence's Elias Fallon delivered one of the keynotes at ISOCC (International System On Chip Conference). It was titled EDA and Machine Learning: The Next Leap... [[ Click on the title to access the full blog on the Cadence Community ...

featured video

Demo: Inuitive NU4000 SoC with ARC EV Processor Running SLAM and CNN

Sponsored by Synopsys

Autonomous vehicles, robotics, augmented and virtual reality all require simultaneous localization and mapping (SLAM) to build a map of the surroundings. Combining SLAM with a neural network engine adds intelligence, allowing the system to identify objects and make decisions. In this demo, Synopsys ARC EV processor’s vision engine (VPU) accelerates KudanSLAM algorithms by up to 40% while running object detection on its CNN engine.

Click here for more information about DesignWare ARC EV Processors for Embedded Vision

featured paper

An engineer’s guide to autonomous and collaborative industrial robots

Sponsored by Texas Instruments

As robots are becoming more commonplace in factories, it is important that they become more intelligent, autonomous, safer and efficient. All of this is enabled with precise motor control, advanced sensing technologies and processing at the edge, all with robust real-time communication. In our e-book, an engineer’s guide to industrial robots, we take an in-depth look at the key technologies used in various robotic applications.

Click here to download the e-book

Featured Chalk Talk

Bluetooth Overview

Sponsored by Mouser Electronics and Silicon Labs

Bluetooth has come a long way in recent years, and adding the latest Bluetooth features to your next design is easier than ever. It’s time to ditch the cables and go wireless. In this episode of Chalk Talk, Amelia Dalton chats with Mark Beecham of Silicon labs about the latest Bluetooth capabilities including lower power, higher bandwidth, mesh, and more, as well as solutions that will make adding Bluetooth to your next design a snap.

Click here for more information about Silicon Labs EFR32BG Blue Gecko Wireless SoCs