At CES Intel had three things to show off, wearables, a development platform, and gesture recognition with one having much more prospect than the others. While all the technologies shown were all over the rest of CES too, Intel did manage to stand out from the pack in one respect.
The biggest news was of course a platform Intel called Edison, an SD card form factor PC that had an unfortunately confused message. Edison is not a device, it is a dev platform for embedded and wearable devices built around a 22nm Quark core, the SD form factor grabbed headlines but detracted from the real purpose. Just looking at Edison, it looks like an SD card until you turn it over and see I2S, I2C, and SIM pinouts on the bottom.
The SD card of the future
What does the device bring to the table? Word has it that the SoC could be sporting two cores but that wasn’t officially stated. Edison would need a new uncore for that which was officially stated, and is quite welcome. LPDDR2 is the memory of choice, NAND storage, and BT-LE plus Wi-Fi for connectivity. Unfortunately, Intel went out of their way to take what is an interesting device and message it badly enough to make you wonder what they are hiding. Just like their 32nm predecessors, the answer is they are hiding a lot of things, none of which are positive.
On the connectivity front, neither BT-LE or Wi-Fi are on chip, they are in external packages. This means even if Intel can drop the price of Quark from the $5-20 range they are asking to the <$2 of the faster and far more power efficient competition, you still need expensive radios and PCBs to make a functional device. This again makes Edison a non-starter unless Intel ‘sells’ every one with bags of cash attached.
Power was another in a long line of 32nm Quark Achilles Heels and the 22nm version drops power from flat-out unworkable to simply laughable. How bad is it? Intel wouldn’t give out TDPs, when they take a page from Nvidia’s playbook you know it is a non-starter. Worse yet the 22nm shrink makes the main focus of their messaging, partners adding IP, even more of a financial impossibility than it was on 32nm. In short the minor technical advances are offset by the painful monetary regression.
The platform itself as a developer device is pretty interesting, a lowish power x86 device with all the right connectors has appeal. One could leverage current x86 developers to build embedded devices on Edison instead of ARM SoCs, but anyone familiar with current tool chains would probably have a hard time unlearning the 64-bit world and re-learning decade old workarounds common in 1990 era ISAs. An ARM tool chain would probably be about as much work to get the hang of, and the silicon you target is better in every way than Quark.
In short, Edison is an interesting development platform messaged in a way that was meant to confuse and misdirect. The tech under the hood is marginally interesting until you look at the economics or worse yet the competition at which point it becomes a running joke. Although there may be some things Intel has yet to reveal, given the claims for 32nm versus the actual silicon, don’t get your hopes up. How could Intel sink this low?
The rest of the wearables side was actually much more interesting with a set of earbuds, an earpiece, and a bowl. Heart stopping-ly interesting as it may sound, these three have a lot of promise for all the reasons Quark is doomed. That reason is simple, Intel is not selling the chips that go in to them, the company it is almost assuredly going to sell full devices. What do they do that is interesting?
First up we have the bowl, and like its name suggests it is actually a bowl about 10 inches across. The tech in the bowl is wireless charging via the A4WP standard backed by Qualcomm among others so it should have a wealth of compatible devices. If you throw your device in the bowl, as soon as your back is turned a little door opens up and elves run out to connect power cables to the devices. This has the effect of wireless charging, something that is all over CES in many different forms.
The bowl that personifies Intel wearables
Where Intel steps out of the pack is that the new bowl is, well, stylish. The corners are nicely sculpted and it is obvious that Intel put a lot of effort in to the design. The rear has power and USB ports for old school charging, and even that view is much more elegant than average. Every other wireless charger SemiAccurate saw was either flat or so clunky you would be ashamed to put it anywhere it could be seen. Intel’s was just the opposite, you would want it in a publicly viewable space.
The wireless headset was again pretty nice looking but nowhere near as elegant as the bowl. That said compared to the rest of the Bluetooth earpieces out there this one was in the top echelon of industrial design and it had some interesting tech under the hood. What tech is that? Basically it is always on, not just quick wakeup but listening to the user 100% of the time.
It was designed with always on as the central theme, instead of having to press a button, wait, and then jump through some hoops for your Siri clone to respond, it will always be there. To achieve that goal, you need to have a very power efficient speech recognition engine in the device, a large battery, or both. Given the size of the earpiece above it is clear that you could not fit a car battery behind a little white plastic cover so it is probably a pretty efficient engine. That said we have no idea if it actually worked, much less worked for a full day but the concept is good and Intel should be able to deliver on it soon enough.
Last up was a set of wireless earbuds that measure your heart rate and distance walked in the unit itself. None of these things are unique but all together they are hard to find in one package. Intel’s big push on them is convergence, no need to have three or four devices when you can do it all with one of theirs. Better yet it is one sleek device that looks decent, works well, and has all day battery life too. Once again, Intel did right on this one should the tech pan out as advertised.
So why is SemiAccurate down on Edison, one of the ‘hits’ of CES? How could a development platform be anything but unlocked potential? Because the tech is wrong and even if that could be completely fixed, the finances can’t. Why are we encouraged by the bowl, headset, and buds when they are farther down the economic commoditization ladder than Quark? Because Intel is not selling them as chips against a much more capable competitor.
Intel is not talking about what they are going to sell in this space but given the level of polish that these designs show have you have to assume they will be put on the market with minimal changes. Intel has zero experience in this realm, they are not a device maker, they are a component maker. We will go out on a limb and postulate that you will see this trio available from multiple vendors with a non-Intel label on them before the year is out.
If Intel sells devices as a whole, effectively as a white box wearable thingy, or even the PCB plus software and services, they will make many times what they could for the silicon alone. If the category is a hot one there could even be decent margins on the line, large enough for Intel investors not to punish the stock for once. As long as the tech is real, the designs are solid, and the vendors are carefully chosen, Intel could do really well in wearables. We think the tech is real now or will be soon enough, the designs so far are really good, and if the team can get this far, there is reason to believe they won’t botch the vendor choices.
Stepping out of wearables we come to the other big push for Intel at CES, gesture recognition. The tech involved was an interesting 3D camera module that was first shown of at IDF in September. It has two cameras and a color module so it is unlikely to be a time of flight mechanism like the XBox One Kinect device we wrote up earlier (here and here). While there was nothing more said about the tech, we have requested more details on what the silicon actually does and how.
Luckily for all of us, gesture recognition is going to have the same warm reception that most recent forced technological pushes had, that is to die in flames. Sure it is early in the life cycle, sure things will get better and cheaper, and sure there will be niches where it is extremely useful. For the moment though, gesture recognition is expensive, rarely works much less works well, and annoys the hell out of the victi…user. For some reason people don’t buy laptops which annoy them more for a higher cost, look at the sales of Windows 8 for more on that consumer trend.
Intel demonstrated a lot of gesture functions at CES that mostly worked, and software to back it up. One was an alphabet learning game for kids where they just held their hand out and moved a basket on a screen to catch things starting with the appropriate letter. It worked quite well. Kudos to Intel for turning learning into something that will make a child’s arms hurt within 10 minutes, what could go wrong with equating learning with physical pain in young children?
Similarly all the gesture recognition features shown had a neat demos but, well, none applied to the general case. As the author said to Mooly Eden after the keynote, “Wow, this is going to be great. I can just see how this will transform the next Intel analyst day. ‘Now we will bring out the Intel Dance Squad to Interpretive Dance our 2013 financial results spreadsheet. Go for it guys!’ What could go wrong?”
In short, all the gesture demos at CES that SemiAccurate saw at best barely worked. They are a long long way off, and there wasn’t a single one that did anything better than a touch screen, voice, keyboard, or mouse. Touch had a killer app or two that could be very useful in some niche markets, gesture lacks that. It may exist but for the general case it is doomed. The quicker this turkey goes away, the better off we will all be.
In the end CES was a mixed bag for Intel. Edison was a nifty development platform that is aiming to sell a chip that isn’t economically feasible. Wearables are a bright spot because they don’t have the same fatal economic flaw as Quark, plus Intel appears to be willing to sweat the details. Keep an eye on this line, it will have legs. Luckily for humanity, legs are not arms, and gesture recognition will quickly go the way of 3D and Windows 8. All in all, for the future trend setting score book, one out of three isn’t actually all that bad.S|A
Have you signed up for our newsletter yet?
Did you know that you can access all our past subscription-only articles with a simple Student Membership for 100 USD per year? If you want in-depth analysis and exclusive exclusives, we don’t make the news, we just report it so there is no guarantee when exclusives are added to the Professional level but that’s where you’ll find the deep dive analysis.
Latest posts by Charlie Demerjian (see all)
- SemiAccurate has Skylake-SP die shots, sizes, and more - Jul 21, 2017
- Skylake-SP has a diverged core - Jul 19, 2017
- Intel’s Purley platform architecture is a step forward - Jul 13, 2017
- Intel launches Purley aka metal Xeons - Jul 11, 2017
- AMD’s Epyc has lots of connections - Jun 26, 2017