By Ted Speers
Eight hours into my flight to Shanghai (45B) I decided to watch my “Public Enemies” DVD. It’s no “Zombieland,” of course, but it did keep me occupied for 1 hour and 59 minutes—or until it decided to stop playing moments from the climax where Dillinger presumably gets filled with lead while leaving a movie theatre with the infamous “lady in red.” No amount of cleaning the disc of dust and fingerprints seemed to relieve my frustration. So guess what I decided to do? “Write a blog posting?” you ask. Ooh! That’s a bingo!

This blogging adventure started with a request to write a blog about low power. I’m not one to be easily constrained so I’ve pretty much done as I’ve pleased. Here’s an attempt to please my masters.

A while back, I wanted to find out if power really matters (so much for pleasing my masters). I decided to read the first fifty user reviews of 22 phones offered for sale by my beloved service provider, Verizon Wireless. What I found was pretty conclusive.

No matter what type of phone we’re talking about (basic, feature or smart), more than 40% of the reviewers (blue plus green regions) took the time to comment about battery life (power matters), whether good or bad. To the first order, 15% of basic phone users are dissatisfied with the battery life of their phones, 25% of feature phone users and 35% of smartphone users.

Drilling into smartphones, it appears that no OEM has a monopoly on customer dissatisfaction, though admittedly the Blackberry 8830 World Edition and the Samsung Omnia are hanging out in ”basic territory.”

I’m going to make a bold prediction and say that this situation (>35% of smartphone reviewers are unhappy with their battery life) is not likely to improve.

Avicenne (never heard of them before but The Economist has) seems to be predicting that battery energy density is going to double every 15 years or so. For the fun of it, I sketched in a Moore’s Law curve (doubling every 18 months) and a conjectured hypothetical curve for mobile customer desires residing somewhere in the gulf between the Moore’s Law curve and the Avicenne prediction. (BTW, I thought for sure I was destined for immortality when I postulated Speers’ Law in the ’90s: FPGA companies double the degree by which they exaggerate gate counts every 18 months.) This gulf we’ll call the Power Matters Zone or PMZ for short.

Obviously we have a crisis on our hands, which thankfully means the government should be stepping in to fix things any moment now. The first thing they should do of course is mandate that every cell phone contain at least one low power FPGA. (How’s that for a marketing strategy?) It certainly wouldn’t be the least effective government mandate. We’ve seen the addition of an FPGA to a feature phone increase the battery life by a factor of four. (That’s 30 years of battery progress!) How we accomplished that magic feat and other war stories from the handset front lines will be covered in a future posting.

Before I compose myself for landing, I’d like to offer a free Actel SmartFusion Development Kit to the first commenter who can point out the Quentin Tarantino reference.

–Ted Speers is an Actel fellow.

By Ted Speers
What is the future of the FPGA industry? Well, if you listen to analysts, who appear to be listening to the hype machines of the industry Goliaths, FPGAs are getting so cheap and so awesome that they are going to eat into the massive ASSP market. That sounds credible and, you can argue, even follows the playbook laid out in “The Innovators’ Dilemma” (quoted by everyone but understood by few) whereby a disruptive technology gains hold in the lower end of a market with some disruptive attribute (reprogrammability) but then grows like a weed to overwhelm the incumbent technology (ASICs and ASSPs). Mainframe →Mini →PC followed that course.

So how has that inevitable dominance coming along. The following data is from iSuppli circa 2008 (before the crash).

Some decent progress but a little too early to declare victory, in my opinion. Certainly it makes the claim that FPGAs are going to disruptively displace ASSPs suspect when a victory against ASICs looks far from inevitable. How is the battle against ASSPs going?

It looks like the battle against ASSPs has been a losing battle for PLDs and ASICs. How are the PLD giants going to reverse this trend? My prediction is that they can’t because of several inconvenient truths. The first is elasticity. If you lower the cost/price, you make up for it in volume. In the scatter plot below, semi value per equipment drops as equipment unit volume increases (orange line). However, total semi revenue increases as equipment volume increase (blue line). Going full custom pays, in other words. The second inconvenient truth (for which I have no clever graphic to offer as proof so you’ll just have to take my word for it) is that software development costs dominate ASSP development costs. Millions of lines of codes are typically shipped along with the silicon. The cost of converting your hardware to standard cells is not that expensive a proposition when considering the total cost of a program. In fact, thanks to FPGA prototyping, converting to standard cells is very low risk and high reward.

While the FPGA industry has maintained a singular focus for more than a decade now of eating into the tantalizingly vast and seemingly vulnerable ASIC/ASSP market, another programmable product category came out of nowhere to quickly equal the FPGA market in size and continues to outgrow it: the 32-bit microcontroller.

You can make the argument that the 32-bBit MCU has eaten away at the FPGA market from underneath, like disruptive technology wants to do. As the processors become more capable, there is less need for logic or hardware. Instead of looking up at the enticing ASIC and FPGA market, the real challenge is from the encroaching 32-bit MCU market. If you can’t beat them, join them.

Clearly, this strategy is fraught with just as much peril as the aforementioned ASIC/ASSP strategy. First of all, the incumbent players like Renesas or STMicroelectronics aren’t exactly less scary than Qualcomm. You can also argue that it has been tried before without much success: Atmel and Triscend in the 8-bit domain and Xilinx and Altera in the 32-bit domain. You could also argue that a ‘soft’ processor is a better technology given its infinite flexibility. I’d even admit that Altera abandoning Excalibur for Nios was the right move.

But it’s different this time. Only four companies that I am aware of have figured out how to make a business out of programmable logic, whereas you need two hands to count the relevant competitors in the micro space. Programmable logic is just difficult. The technology can be bought, the know-how on how to sell and support programmable logic is a core competence than only a few have successfully mastered.

Going after ASSPs (when ASICs haven’t really even been taken) seems like tilting at windmills. Time will tell, of course. Sit back and enjoy.

Oh, 13D to Frankfurt if you’re wondering.

By Ted Speers
Last time we spoke, I was in seat 22G on a United 777 returning from Seoul and whining about my seat. I now speak to you from 7B on a United 747 to Hong Kong and, thankfully, I have much less to complain about. Though I have plenty of room to get my work done, I also have lots of things to distract me, including an on-demand video system with a large display. In reality, the film selection is not that compelling, which explains why I’ve sunk to watching ‘Zombieland’ for my second helping.

Who knew that ‘Zombieland’ would launch itself into my Pantheon of great films (somewhere between ‘Godfather Part II’ and ‘The Seven Samurai’). What’s not to like, after all, about a movie where the most compelling character (an effective zombie killer portrayed with exquisite understatement by Woody Harrelson) rails against Hostess Sno Balls: “I hate coconut! … not the flavor, the texture!” I submit that that sensory assault is magnified exponentially when the coconut is tinted in any of the common DayGlo shades.

For this edition of ‘Musings from Seat 7B,’ I have decided to steal some of the rules for surviving in a zombie-infested world and apply them to the rules for effective innovation.

Rule #1: Cardio
Zombies are easy to beat in a sprint (try running with outstretched arms) but they never stop. Preparing for a sprint when innovating will get you eaten up. You should expect a long slog.

Rule #2: Double tap
Shooting a zombie twice has no downside (note to screenwriter: what if you’re short on ammo?) and it may save your life. The odds of succeeding with your first attempt at creating a new market or technology are not great. Keep trying.

Rule #3: Beware of bathrooms
In other words, know what you don’t know. There could be a zombie locked in that stall, or hiding in the backseat (see rule #31). I can’t stress enough how important it is to raise yourself from unconscious incompetence to conscious incompetence when embarking on an innovative journey.

Rule #4: Wear seatbelts
This enables you to crash your car into a wall and send the zombie through the windshield. Likewise, preparing for a rough ride is necessary when you are innovating. Internal and external naysayers will be constantly trying to cut you down to size or throw obstacles in your path, to say nothing of the real technical obstacles you’ll be facing.

Rule #6: Cast-iron skillet
When used effectively, a most satisfying implement for inflicting zombie death.When you’re innovating, you want to bring some serious technical weight to bear on the thorny issues.

Rule #12: Bounty paper towels
Prepare for some nasty messes. Zombies are notorious for their tendency to spew strange body fluids (especially when their skull is impaled with a cast-iron skillet).I can’t quibble with the choice of Bounty over Brawny but I know you can get 12 rolls of Bounty at Costco whereas their house brand is single ply and totally inadequate. There are lots of messes along the way when you’re innovating. If you’re lucky, your early adopters will be spared the unpleasantness, but if not, send a team in with the Bounty and everything will be spotless in a matter of minutes.

Rule #18: Limber up
Obviously suffering a soft tissue injury can be untimely when warding off a zombie attack. In innovation, you want to pace yourself properly. It helps to bite off what you can chew, get to market, and then take another bite.

Rule #22: When in doubt, know your way out
While I don’t know this from personal experience, I’m pretty sure that getting cornered by zombies is indeed a bad thing. I have been in the uncomfortable position of having to define a product that I know nothing about. Hiring someone who can define similar products in their sleep helps a lot.

Rule #29: The buddy system
The tried-and-true system for guaranteeing our safety that we first learned in kindergarten even works amongst zombies. Partnering either for essential technology or an ecosystem network is a key element in everyone’s playbook today.

Rule #31: Check the backseat (See rule #3)

Rule #32: Enjoy the little things
Being on high alert all the time and having to fend off creatures wired to consume your flesh can get you down. So can reporting news for the fifth time to executives (see ‘creatures wired to consume your flesh’) that your project has slipped again or that some early adopter has forcibly removed you from his building. I’ve found that poker parties fueled with The Macallan 18 are an effective remedy.

Now I must take leave to enjoy these peanuts while they are still warm.

–Ted Speers is an Actel fellow

By Ted Speers

In eager anticipation of my impending jetlag, I opted to craft my inaugural posting to Low Power Engineering. After all, my attempt to complete today’s NY Times Crossword was less than stellar (what the hell is a 3 letter word for ‘rowdydow?’) as was my effort to dust off my FreeCell skills. So here I sit, in seat 22G on a United 777 bound for SFO from Seoul pondering how to dazzle you with insightful commentary about power.

I’ve been at Actel for almost 22 years, heading the company’s product planning for about a third of that time. Sitting here at 35,000 feet off Juneau is an appropriate place to start relating my thoughts about how we are going to make serious money off our low-power technology (secondary of course to our primary mission of saving the world from the global warming and peak oil hangover resulting from our binging on cheap energy). Actel has been making money for decades now selling our Firm Error immune FPGAs to the commercial aviation business. I’d venture a guess that there are more than 100 of our parts on this very plane responsible for delivering me and my 200+ new friends safely to our destination. The same story is repeated on every Boeing and Airbus jet that you care to name. I’m proud of that.

We expect to continue to make products for this important market. There are even strong indicators that other industries are waking up to the pernicious effects of atmospheric neutrons and I’m sure I’ll figure out how feature that observation into a future posting. I’m returning from Korea because our new goal is to have at least one low power Flash based FPGA in the hands of every passenger on this plane. It hasn’t been easy but we’re going to do it.

The reasons it hasn’t been easy are quite complex actually. Over the course of my next few posts, I’ll do a deep dive into some of these complexities. But that’s not what I’m going to discuss now. I thought it would be interesting and informative to look at power (energy really) from about a 30 million foot view. From this vantage point, we can decide where it might be interesting to do a deep dive in future posts.

So what do things look like from 30 million feet? It looks just like this:

This is the energy balance of our little blue dot. Thanks to such phenomena as the strong nuclear force, gravity and mass-energy equivalence, we currently have about 174 PW of power invading our world. BTW, what’s a PW? I’m glad you asked. That’s a petawatt or 10¹⁵ Watts. If you can follow the figure, about 52 PW gets reflected straight back out into space while the remaining 122 PW gets absorbed by the earth (89 PW) and the atmosphere (33PW). For completeness, I included geothermal energy which is estimated to be a paltry 30TW (enough to make large mountains if given enough time but not enough to hold a candle to the sun) and tidal power.

What’s with the red arrows? Well, what comes down, must go up. In this case, the 122 Watts of incident power heats up the system until an equal amount of power ultimately gets radiated back into space (thank you quantum mechanics). Again, for accuracy, I am speaking of a system in equilibrium. There are estimates that as much as 430 TW of that incident power isn’t escaping at the moment which some speculate may have some future impact on your flood insurance rates.

Now coming down to about sea level, where humans tend to congregate, what is the power story there?

Currently, we’re consuming energy at a pace of about 15.4 TW. How much is that? If we crammed all 6.8 GB (gigabutts) of us into an elevator, we could all ride to the 80 thousandth floor and come back again to do it the next day (assuming the kids don’t press all the buttons of course). This elevator today is largely powered by hydrocarbons. If all of our energy came from hydrocarbons, we would exhaust proven recoverable reserves in about 160 years (or about 60000 elevator rides). Proven conventional nuclear reserves could only power the world for 4 years which would lead one to look elsewhere for the 70000^th elevator ride. Nuclear advocates will correctly point out that adoption of fast breeder technology could make our nuclear resources effectively limitless. It is surprising to observe that the source of power for earthquakes and volcanoes could only take us on two rides a day.

If we could harness all the solar energy that doesn’t get reflected, we could all take 4 trips to the moon each day. As it is, today we only ride to the 850^th floor of our 80000 floor journey on direct conversion of solar power. Others are much more voracious in their solar appetite. Biomass uses 95 TW (or 6 elevator rides) worth of solar energy each day to convert CO2 to oxygen (thank you chlorophyll) and another 71 TW is converted to wind energy.

Finally, what do we do with all of this energy we consume (we don’t actually use it for elevator rides).

Well, a little less than third of it gets effectively flushed down the toilet due to inefficiencies in energy conversion (eg. Coal to electricity) and transport (losses in the grid). A little more than a third gets used to make things (includes growing things (agriculture) and finding things (mining)). About a fifth gets used to either move things or people around. About an eighth keeps us comfortable in our homes and another sixteenth keeps our businesses running. I did find it interesting to note we could pay for most of our commercial energy needs (with the added benefit of losing a lot of weight) by giving up eating.

Well, I’ve been informed that my computer needs to temporarily suspend consuming energy so that’s it for today. I can’t promise that I won’t be taking another high-level view of energy but I definitely will be exploring how electronics can make an impact on how energy is consumed. Hopefully at this point, we may have some idea where to look for opportunities to do just that.

–Ted Speers is an Actel fellow