标签: processors

My wife and I joke about our “adult” magazines. For her, those are the publications about beading. For me, they’re tool catalogs and Fine Woodworking magazine. The latter I peruse as avidly as a teenager gazing at an illicitly-procured Playboy (or, whatever passes for those sorts of magazines today). And yes, I do read the articles. Then there’s processor porn. Do you want to see thousands of pictures of naked CPUs? Stop by the CPU Shack . In addition to extensive archives of processor pics, the site weekly struts out different bits of silicon technology from the past. As in the aforementioned Playboy, the CPU of the Week gets a complete description. For the high-minded who scorn prurient pictures, the site does have articles which are well worth reading. This is a picture of a hot processor from CPU Shack. It’s Intel’s 3002:   I well remember this device. The description says this nerd-magnet was born in 1973, which is astonishingly early. It’s a bit-slice processor – a two-bit slice of a computer. String together as many as you want to get whatever bus size is needed. Strange, huh? AMD competed with the 2901, a 4-bit slice, with which we designed some high-end graphics gear long ago. I don’t know how popular the 3002 was, but the 2901 gained wide acceptance. Then there’s this sweet thing, a 4004, the first commercially-successful microprocessor:   Yes, at 44 years old the old thing has a few nicks and a fading appearance. For younger readers let me assure you that looks aren’t everything, and in this case the old gal is still a hottie. We all know how Intel was asked to design a calculator chip set, but decided to do a general-purpose computer instead. What is less well known is that the company figured that, while a computer IC would be a nice sideline, it would more importantly increase demand for their memory chips! The 4004 was built of 10,000 nm geometry on 2-inch wafers. You can still buy them. As of this writing a few 4004 chips are up for bids on eBay. One is currently at $1200 - for a four-bit CPU that was obsolete 40 years ago. It came out for about $100, or $600 today. Obviously, these are historical artifacts priced accordingly. It is interesting, though, that their price reflects the opposite of Moore’s Law. Turns out, there’s quite an interest in retro computing, and John Culver, curator of the CPU Shack, has introduced a board for testing your collection of 4004 and 4040 microprocessors. Here’s a picture of the “bare board” that constitutes the MCS-4 Test Board:   Pretty darn sexy, if you ask me. Actually, it comes fully-clothed, ah, assembled, so isn’t exactly a “bare board.” The MCS-4 tester also checks 4040 devices, a souped up four-bitter that came out in 1974. These, too, are available on eBay, and I see one at the moment with a $75 “buy it now” price. For that price one could get a couple of hundred Cortex M0 32-bit processors, replete with gobs of memory and I/O. Each of those could run circles around the 4040. The MCS-4 tester will also let users run their own code on the 4004 or 4040 (CPU chip is not included). A 28C64 EEPROM stores the code. You’d need a programmer for the EEPROM, as the board can’t load code into the memory chip. John tells me that he sold out his first batch in a week, but more are on order. Some of his customers just want to check the functionality of their museum-piece CPUs. Others use them in retro projects, like a nixie tube clock. (Nixie tubes were the vacuum-tube equivalent of seven-segment LEDs back in the olden days when engineers weren’t afraid to tread in voltage domains above 5V.) Calendars that sport monthly pictures of attractive pin-ups famously grace the grubby walls of garages. I wish John would come out with a calendar for techies that featured a hot processor of the month!

Mike Barr recently wrote an article wherein he predicted that 32 bit processors will eventually beat or match 8 bitters. I well remember meeting with an analyst around 1990 who told me with great certainty that 8 bits was dead and everything would be, in the near future, 32 bits. Reponses to Mike's article are interesting and argue passionately for both sides of the issue. Chuck Manning thinks that decreasing 32 bit prices will push down their smaller brethren as well. I've made this argument many times in the past. When you can get an 8 bitter for a penny whole new applications will open that we can't imagine today. Chuck also notes that byte-wide processors eat less power, and can tolerate wider power supply voltages than 32 bitters. This is true, and low power is certainly a holy grail of the industry. But I can't see any reason why, sometime in the future, all CPUs won't run off just about any source of energy. Miro Samek says "8bitters make no sense." Part of his argument is that the CPU itself is just a tiny part of a typical microprocessor. Most of the real estate is devoted to memory and peripherals. This is a great argument. Except it's couched in the present tense and is therefore incorrect. Today you can buy an 8 bitter for a third (or less) than the cheapest 32 bit part on the market. That's irrelevant for some applications yet life and death for others. In the future I expect this will change. A Cortex-M0+ in 40 nm geometry requires less than 0.01 square mm of floor space. The CPU itself will eventually truly be an insignificant factor in the transistor budget or die size. But there are three conflicting and confusing trends that toss a little sand into this discussion. First, many low end parts are built with fully-depreciated "antique" fabs at geometries that are almost laughable today. Until and unless parts built with more modern processes have paid for their multibillion dollar fabs, there will be a cost rider associated with the parts. Second, there's another cost that won't go away. Let's face it: the future of 32 bit microcontrollers is ARM, and ARM collects a tax on each part sold. Those numbers are closely guarded, but I have heard rumours that for Cortex-style devices they run tens of cents. Even if all of the other costs were zeroed out, these devices can't compete in the most price sensitive applications. I've long thought that ARM's biggest competitor is the one that doesn't exist yet: a royalty-free open-source CPU supplied with all of the design support ARM provides. Will this happen? Probably. Will it be successful? One trend in the semiconductor industry has been a move away from support of proprietary tools in favour of the freebies, so an open-source CPU would certainly fit the manufacturers' models. But it's hard to see how a free movement can create the huge, mostly compatible, ecosystem ARM provides. Third, silicon costs will continue to drop until they become a non-issue for low-end microprocessors. The package will be where all of the money goes, and there's no reason why high- and low-end microcontrollers won't have the same pinouts and packages. Think a six-pin Cortex part. So, the first and third arguments suggest 32 bits won't cost any more than 8 bitters. The wild card is the second, and it's hard to see how that will play out. I do disagree with Miro's statement "I think that 8bitters still thrive only because of powerful non-technical reasons, such as the immense intellectual inertia of the embedded community." No doubt some of that is true, but costs still drive engineering decisions. There's the parts cost, but also that of tools. I've worked a lot recently with ARM's very nice IDE, but it costs thousands of dollars. Microchip, in contrast, makes PIC tools available for practically nothing. Sure, you can get GCC for ARM and set up your own environment, but that takes time and more expertise than a lot of low-end developers possess. And some teams demand the support they get from a vendor. So will 32 bitters win? Probably, as Mike originally said, in the vast majority of applications. Will that be soon? I doubt it.

Several weeks ago I was cogitating, contemplating, reflecting, and generally ruminating on "this and that" when I happened to think out loud (well, in a column) as to how many microcontrollers I might come into close contact with in a typical day. I'm not talking only about microcontrollers in equipment that I directly interface with, such as the microwave oven and my cell phone and suchlike, but also all of the little rascals that are embedded in the various systems that surround me – lighting, air conditioning, my truck, and so forth. Some time ago – I can't really remember when, but let's say 15 years – I remember reading that the average person comes into contact with – or is affected by – around 20 or 30 microcontrollers with which he or she comes in close proximity each day. But 15 years is an eternity when you think how fast things are moving. As one simple example, it was almost exactly 15 years ago (give or take a couple of days) as I pen these words that the first publically-shared picture via a cell phone took place. On June 11, 1997, Philippe Kahn used his "home-grown" system involving wireless software and a camera integrated into his cell phone to share pictures from the maternity ward where his daughter Sophie was born. He wirelessly transmitted his cell phone pictures to more than 2,000 family, friends, and associates around the world. The world's first commercial offering of a camera and video cell phone came in 1999 in thre form of the VP-210 from a Japanese company called Kyrocera. Since this was originally intended, for face-to-face communication, the lens is located on the front of the phone. Due to the fact that the controls were also on the front of the phone, it was hard for users to capture any images except of themselves (grin). In the early 2000s – about 10 years ago at the time of this writing – cell phones with cameras were pretty darned expensive. Also the picture resolution and quality sucked. I remember telling my colleagues that all I wanted from my cell phone was the ability to make and receive calls. I had no idea about the possibilities inherent in things like iPhones and Android phones with MP3 players and GPS and ... But we digress... my original musings prompted one reader to email me saying the following: The question you ask is one I have been actively asking the non-techie people in my circle. The subject is brought up by me usually when I'm asked what is a microprocessor and what does it do? (My friends know that I use with them a lot). Talk about a leading question... I say show me your phone, and they comply but are puzzled. I then ask to see their iPod (if any) and then their car keys. Then I give them a rough guess as to how many microprocessors they have in hand or have control over. They get the cell phone and iPod readily enough, but are shocked when I go through all the systems in the car that employ microcontrollers. As for the quantity of microcontrollers in my life, I started counting each item around me in my daily activities. After only allowing for one MCU per device (I was not about to tear down everything to count them) I stopped at 45 because I was losing track of what was counted already. It was my turn to be shocked. So, what do you think? How many processors in the form of microprocessors, microcontrollers, and digital signal processors to we come into close contact with each day? Let's include both stand-alone devices and also hard processor cores in other devices; so, in the case of a System-on-Chip that might contain multiple processors in a smart phone, for example, let's count all of these processors individually. In the case of cars, I'm not an expert here, but I would love to know the range of processors one might expect to find in typical modern low-end, mid-range, high-end, and luxury vehicles. So, what say you? Can you add anything to this conversation?  

It was 1981, Motorola shipped the 68000 and Intel the 8086. Eight bit processors were the most common; while the 8051 was in production and Z80s were ubiquitous. That same year Jerry Fiddler founded Wind River ( the name was derived from the Wind River mountain range, a Wyoming favorite of Jerry's ). Consultants at first, by 1987 the company came out with the first version of VxWorks, the RTOS that forever defined the company. VxWorks, though much improved and in now available in many different flavors, is still the company's flagship product. In 1993 Wind went public, the first embedded outfit to do so. In the heady days preceding the dot-com collapse the company's stock soared and they bought, well, everybody: ISI. Dr. Design. Embedded Systems Tools. Diab. SDS. What goes up must come down. After the Internet boom imploded Wind River sputtered along for a time and lost money for years. But eventually the ship righted itself and in recent years has been respectably profitable. They have always been the 800-pound gorilla of the industry, one of the few embedded tool vendors with revenues over $100m. Their sales have typically been in the quarter billion to a third of a billion dollar range in recent years. As the economy recovered it was soon apparent that Linux was to be an important force in embedded systems. In 2004 Wind entered that market, and I'm told Linux is now an important revenue source for them (the company has never broken out sales by segment). Some surveys suggest that today nearly half of all 32 bit embedded apps run Linux, alone or in conjunction with a conventional RTOS. 2009 saw a remarkable change: in a move puzzling to analysts Intel bought Wind River. Intel, the inventor of the embedded system, had over the years jettisoned all of the processors targeted at that market, but with the introduction of the Atom is reestablishing its presence in this space ( though, oddly, they skipped the recent Embedded Systems Conference ). In a nice bit of synergy Intel introduced the first embedded processor, the 4004, exactly ten years before Wind River was born. So Wind is 30 this year and the microprocessor is 40. In another bit of synergy some of the first processors VxWorks supported were the embedded offerings from Intel. I'm told various people will be blogging about this anniversary event at http://blogs.windriver.com/ . Hopefully more old-timers will post memories and stories. I congratulate Wind River on their 30 years, and on their continuing contributions for fully three quarters of the era of the microprocessor revolution. I can't think of any other embedded software company that has survived for so long.

I had an interesting discussion with Robert Cravotta regarding four-bit processors. He came up with the Embedded Processing Directory which lists a good number of CPUs, but tiny 4 bitters aren't included. Does that mean they're dead?   Apparently not. In a post about the subject, Robert identifies several areas where these microcontrollers are still steadily cranking away. One example is the Gillette Fusion ProGlide razor , a "manual" device that incorporates a four-bitter. "Manual" is in quotes since it is not electric, other than the circuitry included for running the CPU and powering whatever it is the software controls.   A gift pack that includes the razor and a number of other manly-grooming accessories costs a mere $8.50 in stores. The razor itself has got to run just a buck or two in manufacturing costs. Even an 8 bit CPU that costs a couple of dimes would eat too much of the BOM. I wonder what the microcontroller costs in the millions of units Gillette buys? It must be pennies.   Breathtaking volumes of low-end products scream for the cost advantages of a truly tiny processor.   It seems, though, that 4 bits has fallen off the radar. The processors are never covered in the press. It's hard to even find datasheets. Here's one for the EM6580 from EM Microelectronic. I'm unable to find a programming guide, but the datasheet claims this is a four bit controller with 72 instructions, suggesting it's a long cry from a RISC device.   Small isn't new. Motorola had a single-bit CPU in the 70s that assembled the bit stream into (internal) four bit instructions. Way cool, but hardly new even then. Data General's Nova 1200 minicomputer was a 16 bit machine with a four bit ALU. Logic sequenced operations through the ALU a nibble at a time. The result: Novas were cheap, at least for computers in the pre-microprocessor age.   They were "embedded" in all sorts of applications. We put them into instrumentation. Though we yearned for the much nicer architecture of DEC's PDP-11 units, the price difference made that impossible. Cost has always been an important engineering consideration, which surely is the motivation behind today's continued use of nibble-wide CPUs.   Do you use really low-end processors? Which ones and why?