My homework for today?

A little knowledge of the M4 and sensor fusion...

Diya Soubra is a CPU Product Marketing Manager for Cortex-M ARM Processors at ARM. He has 20 years of experience in the semiconductor industry, during which he held various positions in engineering, product marketing and business management. Just prior to joining ARM, he worked with hi tech start-ups to develop their business. Before that he was in charge of product marketing for devices for VoIP and broadband gateways. He has also developed various software and hardware products for communication protocols and infrastructure systems while working for Rockwell Semiconductor, Conexant Systems and then Mindspeed Technologies. He received a B.S. in Electrical Engineering from the University of Nebraska at Lincoln, a Master of Science in Engineering from the University of Texas at Austin and his MBA from the Edinburgh Business School. He holds 1 patent.

Samsung''s gear 2?

Yup it was an ARM M4, must just burn the battery, but they must want some math unit to run some serious algos on it?

http://www.arm.com/products/processors/cortex-m/cortex-m4-processor.php

Energy efficient digital signal control

The Cortex-M4 processor has been designed with a large variety of highly efficient signal processing features applicable to digital signal control markets.  The Cortex-M4 processor features extended single-cycle multiply accumulate (MAC) instructions, optimized SIMD arithmetic, saturating arithmetic instructions and an optional single precision Floating Point Unit (FPU). 

commentary;  QUIK wants expertise in both Fixed point AND Floating point.  So the FFE is the fixed point unit (?) and the floating one will come with the M4 part of the SoC is my conclusion.

These features build upon the innovative technology that characterizes the ARM Cortex-M processor family.

Responsiveness and low power

In common with the other members of the Cortex-M family of processors, the Cortex-M4 has integrated sleep modes and optional state retention capabilities which enable high performance at a low level of power consumption. The processor executes the Thumb®-2 instruction set for optimal performance and code size, including hardware division, single cycle multiply, and bit-field manipulation. The Cortex-M4 Nested Vectored Interrupt Controller is highly configurable at design time to deliver up to 240 system interrupts with individual priorities, dynamic reprioritization and integrated system clock.   

QUIK's bits and pieces will allow it to be off most of the time, but still be always on and fusing data, other approaches will be the penumbral M4, a mic is left on, or one other sesnor is left on, they will call this always on ( mic) also, but it won't know where you are or what you are doing; there will be "always on" devices that don't know context and will not be aware in a useful sense.....................    

Easy-to-use technology

The Cortex-M4 makes signal processing algorithm development easy through an excellent ecosystem of software tools and the Cortex Microcontroller Software Interface Standard (CMSIS) .

Its 32 bit,

Maximizes software reuse?

So if they do a beta on an MCU it sill seem like what they know, very familiar?

Here is ARM IoT page

http://www.arm.com/markets/internet-of-things-iot.php

From Sensor to Server

The Internet of Things (IoT) is the collection of billions of end devices, from the tiniest of ultra-efficient connected end nodes or a high-performance gateway or cloud platform, intelligently connected and interoperating with servers and services. ARM’s technology’s breadth and diversity from silicon IP to software IP, combined with its partnership approach and ecosystem meet the needs of rapidly evolving secured interconnectivity of IoT, and provides the quickest path to market with connected chips and platforms. ARM drives and simplifies the current and future IoT applications and services to become truly ubiquitous and intelligent.

So, this is NOT so far away.  It must be the S3?  Does anyone think the S3 will NOT be the one with the ARM core?  Thanks in advance for any thoughts.

iya Soubra > Diya Soubra's Blog

Diya Soubra's Blog

 1 post

Using a variable clock to reduce power consumption for IoT endpoints

Posted by diyasoubra Oct 3, 2013

Connected objects for IoT pose a strong technical challenge with regards to power consumption.

The standard deployment scenario is a battery operated object that needs to last a few weeks if not months in the field without any service call or recharge.

Hence, unlike smart phones that need a recharge every day, or standard mobile phones that need a recharge every week, IoT objects have to operate for long periods without any recharge.

One idea that comes to mind, is a variation on the Big.Little scheme.

In that scheme, demanding tasks run on the big processor, everything else runs on the little processor. Total power is reduced dramatically since most operations require only the little processor.

Assuming an IoT connected object with an RF block, a Cortex-M4 and a few sensors, we can achieve the same dramatic power reduction by changing the clock frequency of the processor.

For most of the time, the processor would run at low clock frequency to service the sensors. Once every few minutes or based on specific events, the processor would change the clock setting to switch into high performance mode in order to transmit secure packages over the RF link. Overall power consumption is dramatically reduced since the processor would run at low clock frequency most of the time.

This sounds simple in theory, but the design has to take into account the whole system design including all the buses and peripherals attached to the CPU. Asynchronous bridges may be required to create distinct clocks to protect peripherals. Also, the change over must be clean to retain the integrity of the clock signal.

I have not done a full system design on this idea yet but I am sure there are other considerations such as settling time for the PLL after a change in frequency.

has any one seen a product on the market yet with a Cortex-M device that uses this technique?

So they are trying any way they can to make it work.  QUIK has the best bits and pieces for needs of

the IoT SoC?

That is the inference to gain subjective probabilities on?