Category Archives: Blog

Over the past couple of months, we’ve been running a series summarizing the excerpts taken from Embedded Computing for High Performance by João Cardoso, José Gabriel Coutinho, and Pedro Diniz that embedded.com has published. The context for the book (and this series) is the growing importance and ubiquity of embedded computing in the age of the Internet of Things – a topic that seems to grow more important by the day. Earlier posts in this series focused on target architectures and multiprocessor and multicore architectures; core-based architectural enhancement and hardware accelerators; and performance. In this installment, we cover power and energy consumption.

The section on power and energy consumption offers information on techniques for reducing power and energy consumption. It’s pretty much the densest excerpt that we’ve seen to date, so you’re definitely advised to take a look at the entire section. But for a quick summary:

The authors set the stage by providing a number of formulae for power consumption (static power + dynamic power consumption), static power consumption, and dynamic power consumption. They then offer these tips:

The static power depends mainly on the area of the IC and can be decreased by disconnecting some parts of the circuit from the supply voltage and/or by reducing the supply voltage.

And, as dynamic power consumption “is proportional to the switching activity on the transitors in the IC”:

One way to reduce dynamic power is to make regions of the IC nonactive and/or to reduce supply voltage and/or clock frequency.

They then discuss ACPI (Advanced Configuration and Power Interface), which the operating system or applications use to manage power consumption.

Energy consumption is not directly associated with heat but affects battery usage. Thus by saving energy one is also extending the battery life or the length of the interval between battery recharges.

The authors devote next address the topic of dynamic voltage and frequency scaling.

Dynamic voltage and frequency scaling (DVFS) is a technique that aims at reducing the dynamic power consumption by dynamically adjusting  voltage and frequency of a CPU. This technique exploits the fact that CPUs have discrete frequency and voltage settings as previously described. These frequency/voltage settings depend on the CPU and it is common to have ten or less clock frequencies available as operating points. Changing the CPU to a frequency-voltage pair (also known as a CPU frequency/voltage state) is accomplished by sequentially stepping up or down through each adjacent pair. It is not common to allow a processor to make transitions between any two nonadjacent frequency/voltage pairs.

This section contains information on measuring power and energy, and on dark silicon.

We’ll be running one more installment in this series in a few weeks. After that, we’ll take a break, but will likely resume the series as embedded.com brings more excerpts of the book online.

 

Embedded.com is running excerpts from Embedded Computing for High Performance, the highly informative (and, yes, educational) book by João Cardoso, José Gabriel Coutinho, and Pedro Diniz. In my last post, I summarized the initial article, which focused on target architectures and multiprocessor and multicore architectures. This post covers core-based architectural enhancement and hardware accelerators.

Core-based enhancements

In this excerpt, the authors address how CPUs have been further enhanced, beyond multicore, by incorporating more than one multicore CPU, integrated as a chip multiprocessor (CMP). They offer two ways in which CMPs can be organized.

The first is Single Instruction, Multiple Data (SIMD) units. These are:

…hardware components that perform the same operation on multiple data operands concurrently. Typically, a SIMD unit receives as input two vectors (each one with a set of operands), performs the same operation on both sets of operands (one operand from each vector), and outputs a vector with the results.

Intel microprocessors (MMX, SSE, AVX ISA extensions) have included SIMD units since the late 1990’s, and ARM got in the act more recently with SIMD extensions to ARM-Cortex (NEON technology.)

Typical tasks that SIMD units take care of are basic arithmetic and other operations, like square root. SIMD unit enhance performance by their ability to simultaneously load from and store to memory multiple data items, taking advantage of the entire width of the memory’s data bus.

One bit of advice the authors offer:

To exploit SIMD units, it is very important to be able to combine multiple load or store accesses in a single SIMD instruction.

The second option for organizing CMPs presented is Fused Multiply-Add (FMA) units.

Fused Multiply-Add (FMA) units perform fused operations such as multiply-add and multiply-subtract. The main idea is to provide a CPU instruction that can perform operations with three input operands and an output result.

Both Intel and ARM have microprocessors that include FMA.

Hardware Accelerators

Here the authors discuss two types of hardware accelerators, GPU accelerators and reconfigurable hardware accelerators (absolutely something we’re quite familiar with here at Critical Link).

GPU accelerators were initially used for graphical computation, but these days they also support additional application areas, including scientific and engineering apps. Then there’s our friend, the reconfigurable hardware accelerator:

Given the ever-present trade-off between customization (and hence performance and energy efficiency) and generality (and thus programmability), reconfigurable hardware has been gaining considerable attention as a viable platform for hardware acceleration. Reconfigurable devices can be tailored (even dynamically—at runtime) to fit the needs of specific computations, morphing into a hardware organization that can be as efficient as a custom architecture. Due to their growing internal capacity (in terms of available hardware resources), reconfigurable devices (most notably FPGAs) have been extensively used as hardware accelerators in embedded systems.

The authors give a shout out to components that combine FPGA and DSP. As I said, right up our alley. Here’s a block diagram of a reconfigurable fabric that includes both FPGA and DSP. It doesn’t look anything like the sort of block diagram we’d present to a client, but I had my reasons for including it here.

As with my earlier post, I encourage readers to head over to embedded.com to read the full excerpts (which are all accompanied by ample diagrams and other illustrations).

Today is Valentine’s Day, so I guess it’s appropriate to write a heart-related post.

What prompted it was a piece I saw on EE Times a week or so ago. In a post entitled “Implants Raise Security Awareness,” Joanne Emmett addressed the topic of medical device security. She started by going back a decade to the time when the doctors who were giving then-Vice President Dick Cheney a heart defibrillator replacement “ordered the device’s manufacturer to disable its wireless capacity.” There were concerned that the device could be hacked, and Cheney assassinated via technology. In a weird case of art imitating life, in an episode of the thriller Homeland, the serial number of the fictional VP’s pacemaker is stolen and a terrorist uses the information to rev up the pacemaker, triggering a fatal heart attack.

More than a decade on, the security of networked medical devices is a serious concern for many reasons. Concerns are growing as devices are engineered to gather and report increasing amounts of data, most of it collected and transmitted using widely available, off-the-shelf software. (Source: EE Times)

As Emmett points out, the real worries aren’t about attacks on individuals like Vice President Cheney or the VP on Homeland. They’re around data mining, and using the information for financial gain.

Hacked data can reveal commercially valuable information, such as performance data on competitors’ products. Such data can be used to exploit weaknesses in a rival’s marketing or used to modify the company’s own offerings.

Networked devices provide feedback on a wide range of patient data such as blood pressure, respiration, blood enzyme levels, and other health conditions. Insurers that issue individual life insurance policies often purchase this information — repackaged by third parties to be of apparently legitimate origin — to assess clients for insurability and to set premiums.

She also notes that fitness devices, like Fitbits, also produce large quantities of biometric information.

The FDA has released some guidance on cybersecurity for medical devices, and legislation has been introduced that would “codify protections for medical device data.”

Emmett’s post was particularly timely given the recent news that fitness device data, from devices worn by U.S. soldiers, could potentially be used to identify their locations. Heat maps developed by Strava, a GPS tracking company, aren’t real-time depictions, but could help enemies figure out sensitive information. When the heat maps show affluent regions in the United States and Europe, where so many millions of people wear fitness trackers, the picture is pretty dense. In poor regions in the Mideast and Africa, where fitness consumers are few and far between, what shows up on the maps are the areas where Fitbit-wearing soldiers congregate.

An Australian student was the first to notice that a heat map of Syria showed where soldiers congregated. A journalist found information that likely gives away the location of a CIA base in Somalia. A Twitter use claims that “he had located a Patriot missile system site in Yemen.” (The heat map shown here is of Kandahar Air Force Base.)

The U.S.-led coalition against the Islamic State said on Monday it is revising its guidelines on the use of all wireless and technological devices on military facilities as a result of the revelations.

“The rapid development of new and innovative information technologies enhances the quality of our lives but also poses potential challenges to operational security and force protection,” said the statement, which was issued in response to questions from The Washington Post.

“The Coalition is in the process of implementing refined guidance on privacy settings for wireless technologies and applications, and such technologies are forbidden at certain Coalition sites and during certain activities,” it added. (Source: Washington Post)

Technology that both “enhances the quality of our lives” while at the same time posing “potential challenges” pretty much sums up the Internet of Things when it comes to the consumer end of the product spectrum. There may not be someone out there who’s going to hack someone’s pacemaker and throw them into cardiac arrest, but a lot of the security challenges that have been addressed in the industrial IoT have yet to be seriously taken when it comes to consumer devices, including some life and death ones.