Intel and IoT

[Sterny]

I'm a hobbyist that caught the electronics bug a few years ago.  Dipped my toes into all the popular dev boards from Atmel, Intel, TI, ARM, as well as various radios.

When Sparkfun 'partnered' with Intel for the Edison dev boards, I bought a few kits, but never did much with them.  Sparkfun developed a lot of hobbyist level guides and education material.  I recall hearing a lot of complaints with Intel's support.

In my view from the cheap seats, it seems like most of momentum is with established embedded companies like Atmel, Microchip, TI, and various Asian ARM boards.  TI's community and developer support seems to be world class, and Microchip/Atmel and ARM are huge.

With Intel announcing a big IoT strategy,...

http://www.informationweek.com/cloud/intel-ceos-manifesto-details-cloud-iot-strategy/d/d-id/1325284

 ...and after catching a few episodes of their IoT Maker TV show...

http://www.mercurynews.com/business/ci_29755557/makers-tv-show-intel-spotlight

I wondered if any of the real EE's were using some flavor of the Quark in their designs.  What would make your companies switch over to using Intel MCUs from some of the aforementioned companies?

I have no professional experience with Intel x86 CPUs, but I've been a PC enthusiast and I'm aware of their history.  This stuff ain't new to them.  It seems like they've been trying grow something from the 'maker' market - but does that necessarily mean that a company would switch to Intel and move away from all the institutional knowledge built from developing on competing products?

Is Intel just playing the long game and throwing money at it? If so, will it work?

Also settle a bet:

Did they come up with the name 'Intel' as a short form of the term INTegrated ELectronics?

[helius]

Did they come up with the name 'Intel' as a short form of the term INTegrated ELectronics?

That is the story as reported by the Encyclopædia Britannica: http://www.britannica.com/topic/Intel-Corporation

[stmdude]

There's too many players in the IoT arena already...

Sure, Intel is a heavyweight, for sure, but I don't really see how their products fit in _anywhere_.

Basically, they have two products that I think they'll try to market. The Quark, and the smaller Atoms.

Now, in IoT, you have a myriad of different devices, but they basically break down like this:
1. Very small and very battery-powered. Typically sensors and wearables
2. Larger "sensors". Such as coffe-makers, fridges and everything else people are trying to connect.  Power is not a huge issue, nor is size. But price is (as it always is)
3. Border/gateway devices. Basically, it's your WiFi router, but instead of WiFi->Ethernet it's something like ZigBee->WiFi

Now, for (1), Intels quark is batting a solid zero. It's much to large, uses way too much power, and costs way too much. Devices I've designed in this range have MCUs that costs <$1, uses ~3.5uA (with RF), and are ~11 square millimeters.
For (2), Quark or Atom could work, as neither size or power is a big problem. However, price is..  Quarks are around ~9.50$..  And, if an IoT company is already making (1) devices, why not use the same part?
For (3), Atom is a contender, but again, it comes down to price. I'm actually in the process of doing RFQs for a device like this, and the prices I'm looking for is <$4 for the CPU/SoC alone. Atoms start at around $19..

I think that the only way they could offer something interesting would be something like the Quark, but with an integrated LTE Cat0 modem, and a price of about $5.  That way, they would be competitive with the upcoming offerings from MediaTek and Qualcomm, but with the added benefit of a very mature Linux port.

[Kjelt]

Also don't underestimate that companies that already have a couple of generation of products based on an Arm core need to recode their entire platform for x86 if they switch.
So that would only be a valid option if their platform is based on an OS like emb. linux.

[stmdude]

Now, in IoT, you have a myriad of different devices, but they basically break down like this:
1. Very small and very battery-powered. Typically sensors and wearables
2. Larger "sensors". Such as coffe-makers, fridges and everything else people are trying to connect.  Power is not a huge issue, nor is size. But price is (as it always is)
3. Border/gateway devices. Basically, it's your WiFi router, but instead of WiFi->Ethernet it's something like ZigBee->WiFi

For 1, Intel has Curie (based on Quark SE), it seems overpriced, but considering the integrated BLE and 6 axis sensor as well as the coprocessor dedicated for sensors, it is a nice deal.

My <$1 chip also has BLE. No 6-axis though, but none of our products needs a gyro or accelerometer.
While the Curie looks interesting at first glimpse, I doubt even their co-processor can go down to the current-consumption of a decent M0.
Also, what's up with Curie?  There was a big announcement, with promises of datasheets, etc, in Q3 2015.  They're still no-where to be found. I'd love to be proven wrong.

For 2, Intel has Quark D series that competes with 8 bit MCU and 32 bit M0/M0+, but Intel offers plethora choices of libraries and ecosystem backed up by Intel's unlimited R&D money.

True, Intel has great software support. But, in the end, when you multiply your BOM with the number of units that are projected, every cent saved is many thousands of dollars earned. So <$1 chip wins again.

For 3, I think the performance of Quark is not that bad, but I do not think it has a chance. For such task it is better to use a Cortex A5, cheap enough and is compatible with standard SDIO/SPI WiFi chips.

Yep, didn't think about that, but yea, you should be able to do this with a Quark as well.  But, yea, price..

[stmdude]

I'm curious, what $1 chip has BLE?

nRF51822 (128KB+16KB variant), WLCSP packaging, 100K+ volume

[stmdude]

100K+ volume

That deterred me .

They're still kind of cheap when buying them in single quantity in QFN packaging. <$4

[VK3DRB]

Heads will roll as Intel morphs. Sadly, many of those heads are those who helped make the company what it is today. Now they are no longer needed, they are discarded. I cannot understand why the 12,000 employees cannot be redeployed. Surely Intel can afford it. (I saw this one sided loyalty with IBM, except with IBM was probably a lot worse in Australia in that they ruthlessly revoked allocated retirement savings from employees.)

Here is a copy of the letter from Intel's CEO to Intel employees...
https://newsroom.intel.com/newsroom/wp-content/uploads/sites/11/2016/04/krzanich-restructuring-memo.pdf

[Kjelt]

Latest news from the CEO of Intel about Intel and IoT
https://newsroom.intel.com/editorials/brian-krzanich-our-strategy-and-the-future-of-intel/

As I read it he still believes the law of Moore will continue and that a lot of Intel cards will be put on the cloud servers (as I interpret it).
I don't know what to think  of it, a lot reads like management BS. It reads as if all future IoT devices are useless without their connection to the cloud.
Hmmmm I wonder what strain that will put on the reliability of internet connections in the home, without it nothing will no longer work

[stmdude]

It reads as if all future IoT devices are useless without their connection to the cloud.
Hmmmm I wonder what strain that will put on the reliability of internet connections in the home, without it nothing will no longer work

Right now, pretty much all IoT products become useless if the internet connection goes away.

Once people start figuring out that "the cloud" doesn't really help anything (except data-gathering from your customers), this might change.
The product I'm working on right now doesn't need an internet connection to do its main functions.  Actually, it would detract from the product, as people expect some things to work regardless if the internet is available, or even if the power is on. Also, latency.

[RGB255_0_0]

I only really see IoT useful in devices like medication and medical equipment. Devices like the Amazon Echo symbolise laziness; likewise with fridge-freezers that order milk - though I think Smart FF are useful (like warning you that you should consume x before it goes off (saves wasted food).

[coppice]

I'm curious, what $1 chip has BLE?

Prices vary quite a bit, depending on the chip's features, but there are 70 cent BLE SoCs.

[linux-works]

 ...and after catching a few episodes of their IoT Maker TV show...

http://www.mercurynews.com/business/ci_29755557/makers-tv-show-intel-spotlight

slightly OT, but I'm curious - what do people think of that show, overall?  has anyone seen it yet?  its already on its 4th episode.

[linux-works]

Also don't underestimate that companies that already have a couple of generation of products based on an Arm core need to recode their entire platform for x86 if they switch.
So that would only be a valid option if their platform is based on an OS like emb. linux.

see this:

https://www.zephyrproject.org/news/linux-foundation-announces-project-build-real-time-operating-system-internet-things-devices

its not linux.  its not arduino.  and its what intel thinks will be the 'new arduino' API, so to speak.

only time will tell if people jump on this or not.

[Sterny]

see this:

https://www.zephyrproject.org/news/linux-foundation-announces-project-build-real-time-operating-system-internet-things-devices

its not linux.  its not arduino.  and its what intel thinks will be the 'new arduino' API, so to speak.

only time will tell if people jump on this or not.

That's interesting, thanks for the link.

One thing that caught my eye was them mentioning that Intel bought Altera and Wind River.  I knew about Altera, but hadn't heard about Wind River.

As I understand it, Wind River develops VxWorks, which I know it be the RTOS for a lot of industrial and financial grade systems. What are the chances Intel puts VxWorks IP into the Zephyr project?

I haven't gotten around to working with TI-RTOS, nor FreeRTOS.  But I wonder what's going to happen to FreeRTOS with Intel and their growing consortium putting resources behind Zephyr.   Seems like it would attract people being an open source project.  Port it to any architecture and you'd have something unifying, no?

[Sterny]

slightly OT, but I'm curious - what do people think of that show, overall?  has anyone seen it yet?  its already on its 4th episode.

I watched the first episode, and have the rest waiting to be watched on the Tivo

I wasn't really entertained by it.  I never watched Mark Cuban's show or Trumps, but I knew of the format, and this seems to mimic it.  People come up with something based on Curie, and a panel of people comment and critique it - Intel's president is up there judging, but the rest weren't anyone remarkable to me.  You see Intel badging everywhere, and the studio looks like a futuristic-trying gameshow. Winner gets a million.

The projects are things you'd might see on Hackaday, maybe a bit more polished. RFID and sensor ear tags for cattle seemed to be a good idea - they had a dog running around getting all the measurements from the cows with a gateway device attached to his collar.

It's on TBS for me.