8 bits is dead

11/7/2010 11:25 PM EST

Jack, great article, very informative.
I almost made in my pants when I read the title though, hehe.

11/8/2010 1:01 AM EST

I agree with you Jack. I have been an embedded developer for more than a decade & have also acted as a consultant, trainee & speaker at conferences related to embedded technology.
I asked people, whenever possible, as to why do they want to know about an ARM processor ? Or why Linux or an RTOS ? A lot of them, to my surprise, said because "that is the trend these days". At that, I had to argue with them as to why do you need a 32-bitter when the good,old friend - 8051 can do your job ?
I have never bought the theory that 8-bitters are dead. In fact, the people who program for such controllers are more smarter than the rest, since they have to settle their world within 256 bytes of RAM & 4 K flash !
Yes, if your application demands the kind of processing, go for a 32-bitter. Else, these days you have very efficient ( in terms of peripheral integration & power ) 8051s et al who can fairly compete with an ARM to deliver your application faster !

11/8/2010 4:09 AM EST

A prediction on the peak 8-bit processors year anyone?

11/8/2010 12:31 PM EST

Jack, the resilience of 8 bit products has been incredible but in technical markets the ever decreasing price per feature eventually wipes out older technology, for example who would buy a 15" LCD display or a 5.25" disk drive today? even if you could!
Both NXP and ST Micro are selling ARM Cortex parts for under a dollar and the dev kits are very cheap as well. At this weeks ARM TechCon in Santa Clara (Nov 9-11) you can see and touch some of these products and decide for yourself.

11/8/2010 4:04 PM EST

I tend to use AVR 8 bitters for really simple stuff and ARM for anything else.

For me the 8-bit "killer features" are:

* Wide voltage tolerance. Many 8-bitters can run from around 2.0 to 5.5 volts - some even wider - which mean you can get away with no voltage regulator and simpler layouts. All ARMs I've seen still need a relatively clean 3V3 or such. That is an extra $1 of hidden costs.

* Extremely low standby power that no ARM can match yet.

Yes, it is true that ARM is constantly intruding into what was 8-bit turf, but the process improvements that help ARM achieve this also work for 8-bitters too. 8-bitters are getting ever cheaper, lower power and finding their way into places which could previously not justify a micro.

11/8/2010 7:23 PM EST

Seems like 16-bit are finding a niche on DSCs (Digital Signal Controllers), what do you think?

11/9/2010 1:34 AM EST

This is like arguing over things like "small screws are dead". I plan to use the best fit chip to the job. If all chips had the same cost and the rest, then I can see using the highest chip to simplify my design time.

11/9/2010 9:10 PM EST

It´s the first time (since 1999 year) that I don´t agree with you Jack :(

IMHO, There is no way to start a new project.. and use a PIC instead of an ARM...

11/10/2010 5:11 AM EST

It seems that Jack writes "the same" article every year or so :) What is the reason that I still use 8-bitters? Simplicity of software development. Period. My compiler generates assembly code so I can see how the program works and I can easily spot the compiler mistakes (if any). It is waaaaay harder to do it with 32 bit MCUs, because you have to read 20 times more spec to understand it, while 8 bitters are almost generic in its form.

11/11/2010 1:17 AM EST

Yes,it is true that the 8 bit controllers will continue to be used as long as the Companies keep producing them.Small size, fewer pins and lower assembly costs favour 8 bits. Simple applications like in toys,control panels etc., and small control applications are all reserved volume markets for 8 bitters.Are we not still using transistors inspite of the availability of multi transistor ICs?, the same logic applies in the case of 8 bitters also.
Laktronics

11/11/2010 2:38 AM EST

Thanks, Jack, for your competent column. Too many of our colleagues tend to follow the phrase: "the bigger=the better", neglecting the important differnces between "microcontrollers" and "microcomputers". Top tier management, often using the same decision makeing rule, will later complain about the tremendous BOM and the total cost of the system.

11/11/2010 7:41 AM EST

Jack, Yes, the Atom powered toothbrush needs a heatsink, put on your marketing hat: "The ZF-1 toothbrush has a heated bristle area for your brushing enjoyment!"

8 bit MCUs are dead - long live 8 bit MCUs

11/11/2010 7:47 AM EST

The press not talking about a technology never implies that the technology is dead, merely that it has become ubiquitous and commoditised. So the premise of the article is flawed; did any one ever *really* believe that?

The press only talk about the latest technologies and product launches because there is little more of interest to be said about commodity parts. No one needs to be informed of their availability because we all know already. After all, you don't see many press launches for a new line in resistors, but you know you can get them!

No one is likely to launch a 32 or even 16 bit chip with 8 pins for example, but many applications require no more and the manufacturing and development cost of laying down a high density SMT QTFP or BGA with 64 or more pins only to track-out 5 of them is prohibitive and frankly bonkers.

11/11/2010 8:50 AM EST

5 years ago I visited a consumer electronics customer who was getting a stripped down 6502 (if you can believe a 6502 could be stripped down) in bare die form for $0.09 per piece.

11/11/2010 9:32 AM EST

I really agree with you Jack, the 8-bit processor will never, ever disappear. I've been a software engineer for over 30 years now, and I don't recall there being even a dip in the number of 8-bit processors sold world-wide compared to 16 and 32-bit.

11/11/2010 10:19 AM EST

Why don't we talk about how 4 bit micros will never disappear? I assume there are still lots of toys, timers, switches, etc. that still use 4 bit processors. Is anyone still using 4 bitters?

11/11/2010 11:42 AM EST

Since C does a "good enough" job compared to assembly, the micro chosen is not necessarily as important as in the past. Having said that there are a ton of applications (and always will be) where 8 bits is more than sufficient. What 32 bits does for you is that in systems where you might need several 8 bit micros, you can possibly use one 32 bit which tuns out to be less expensive that several of the 8 bitters.

11/11/2010 1:53 PM EST

I use the 8-bit vhdl "Picoblaze" microcontroller in most of my Xilinx FPGA designs. It's essentially "free", and makes control and data flow applications easy.

11/11/2010 2:50 PM EST

I used to use PICS, but now am firmly an Atmel AVR (8-bit) freak. Wide range, great instruction set, free tools (AVRStudio and gcc), killer devices that are easy to write drivers for, good power consumption, built-in osc to 32MHz, plenty of pins (or just a few on the ATTiny), built-in USB support if you need it, etc.

The documentation is good - the appnotes are real good, the code examples are good if you strip out the weird comments (yeas, they are there for the automatic documentation package - they just get in my way).

One of the things that drove me away from PICS is the really bad PIC24 CCS compiler - is is very buggy. Microchip does not do a good job qualifying third-party products. I lost 60 hours of my life in the middle of a deathmarch from the CCS bugs. I even had to write my own UINT32 divide routine, the CCS version would fail sometimes.

Not that I mind ARMs - I have used an ARM7 in the past and like the ISR shadow registers. Also, when I use an ARM I have two (I am the one-armed bandit...). Still - that is way too much horsepower for what I normally use. (I also like Coldfires for the higher-end stuff. Checkout www.netburner.com - great stuff if you need that level.)

11/12/2010 10:09 AM EST

One more thought from me (a bit outside this 8-bit discussion). In many cases (like mine) it is far more difficult to find a good supplier (manufacturer) rather then choose the perfect architecture (8 vs. 32). There are so many different manufacturers making all these thousands of types of MCUs yet when you design something for the high volume production it has to be available. Before I pick the MCU type I always try to survey the suppliers and ask them if the part is popular and is there any chance it will become obsolete soon. On the other side more popular parts are more likely to be cheap as well.

It is not just a question of what type of MCU I like but more how practical it is to use it and because 8-bitters are more generic in its form I can use almost any of them.

Here in Europe especially Atmel had a bad reputation for not meeting demand and delaying supplies. Unfortunately our "own" manufacturer - ST (that I am stuck with) joined this trend last year (they stopped the production for few months expecting the recession to be longer and last year they were giving 50 weeks of lead time for some of their parts).

Maybe it is some idea for a survey or an article, Jack? What MCUs do you use? Why? What problems do you encounter with supplies?

11/12/2010 11:51 AM EST

I would say that the ARM7 plus Cortex M0/M3 is definitely the 8051 of the 21st century. The part is available from lots of vendors and is available in a wide range of package sizes and I/O capabilities.

Regarding the horsepower issue. An ARM has the horsepower to use an RTOS in even the most stripped down version. Using an RTOS is a game changer in an embedded product. I always used a simple round robin workloop with interrrupts for my 8 bit apps. After developing a couple of projects with ARM processors using an RTOS I can say, I am never going back. The amount of state logic that gets replaced by a couple of well partitioned tasks communicating with queues is incredible. The software design becomes much simpler and easier to document and explain to others. That alone is sufficient reason to switch.

11/12/2010 1:56 PM EST

With cheapest ARM MCUs costing well below 1$, the raw part cost is not the deciding issue. Still, 32-bitters tend to have higher system cost because they need tighter power supplies, BOR circuits, fine tolerance PCBs, etc. They also tend to have less advanced peripherals: you rarely see ADC/DAC analog front ends on the ARMs. Finally, the low power modes and the fast wake-up times on 8-bitters often result in much lower power profile.

None of these seem to be insurmountable: I hope manufacturers like TI and especially ST and NXP will keep pushing low-end so that it'd be as easy to stand up an ARM-based one-off as it is currently to breadboard a PIC or AVR system.

Having said that, I suspect that PIC/AVR/6502/8051 will go on as long as there are people who know and like them.

11/12/2010 5:39 PM EST

8-Bit controllers: they are still the best learning tool in embedded electronics. You gotta start someplace and the simplest place is the best place.

11/14/2010 11:42 PM EST

Everybody seems to be speaking from their own application point of view. What jack says here is from embedded industry perspective and he is right in saying that 8bit MCUs are not going to be dead. The scope of embedded apps is so great that we cannot straight jacket all uses to 32 bit scenarios. With lower geometries they are becoming more and more attractive to host of other apps which came hardwired.

11/15/2010 12:01 PM EST

Another factor in the 8 vs 32 vs 64 vs... discussion is the total system complexity. You can start with an idea and a few parts and have a PIC or AVR project up and running in just a few hours - design, prototype and coding. The barrier to entry is very low.

It's not just seasoned embedded engineers working with these things. Engineers from other disciplines are more and more frequently being tasked with adding a bit of electronic control to their old mechanical devices. In cases like that, that low barrier to entry can be the difference between success and a failed product.

With an ARM, you can certainly do a lot more than with an 8-bit processor, and the chip prices can be pretty comparable, but you need to spin a PCB or get a development board before you can do much of anything. You frequently need to worry about things like line-level conversion and higher speed PCB design. Not that those are uncommon issues, but they do add time and cost to the project over and above just the chip price.

11/16/2010 3:29 AM EST

Jack,
You're damned right! A good engineer designs something for $2, a better one does it for $1. This thought I learned from a Canadian Engineer Reinhard Pildner (worked for DSC in Toronto) and these guys produce alarm panels with only 8-bit u-controllers in them. The functionality of those 16kBit u-controllers will amaze you! And I bet that they earn the most money out of them, because they are sooo cheap. Another thing: If you need a small micro in an FPGA, do you start with 32 bits? Of course not! You first look if you can do the job with a 8 bit Pico-Blaze. The 32 bit one will cost you too much cell's. And if you can't do it with one, you programm 2 in your Xilinx FPGA...

11/16/2010 11:29 AM EST

You can do pretty useful stuff with an AVR (look at Arduino and all its clones). Cost-wise there are Arm Cortex M0 etc for around the same price, but harder to prototype because of power supply and package options - you need a 100 pin surface mount break-out board before you start (and a steady hand with the soldering iron). But there is an ARM version of the Arduino (see LeafLabs Maple) and absolutely tiny versions of the Arduino (see Teensy++). Sure makes starting a project easy when you have 128K of onboard FLASH, a USB interface, 50+ user definable pins, UARTS etc and no need for external DRAM etc etc

If you only need to handle external signal protocols of 200KHz or so, the AVR chips are fine. If you need to do faster things than that, or signal processing then probably need an ARM core.

11/16/2010 10:30 PM EST

Cost of the micro is a function of the number of units sold vs the NRE (broadly speaking).


If you are going to make millions of units, it makes sense to spend the money to save a nickel on each unit - the NRE is made up quickly. But .. if the units sell for 5x or 10x of the BOM, then engineering time is more valuable than the nickel.


I am told doorknobs have 4-bit bangers in them (never had a chance to take one apart :^) - I suspect it's a 4004 or similar. You don't need more that that, and they need to sip power.


8-bit micros will be around for a long time, for the reasons most folks have posted - ease of use, cost of entry, don't need the power. I have a back-burner project - an LCD controller - and there is absolutely no need for a 32-bit 400MHz ARM.


I have used ARMs and Coldfires, and really like them, but only when I had a real need for that sort of horsepower. It's like putting a muscle car engine in a VW bug - way overkill that can wipe you out.


On the cost of entry - with most 8-bit chips you can get DIP packages and use a cheap/free PCB layout package and one of the proto board houses (I like www.apcircuits.com) so you can lay out the board at a reasonable scale, solder sockets onto it, and get something working cheap. You need a higher class of tools and more expensive board houses (a factor of 5x to 10x) if you need to goto 5 mill or even 3 mill traces.

11/17/2010 3:21 AM EST

In terms of technical strategy, 8-bit controllers are good to have in your tool kit, because they can be efficient solutions.

In terms of career strategy, they are commodity items that require only commodity skills. In fact, there are probably enough off-the-shelf 8-bit solutions that an engineer might not be required. If you design an 8-bit solution, I can't imagine you will be getting a lot of recognition for it. Though good engineering shouldn't be influenced by such politics, it is.

In terms of market strategy, there is such a low barrier to entry regarding PIC use, that your product will be easily copyable. Complexity is a barrier to entry. It buys you time in the market place (that may be why we see so much bloatware around). PICs won't buy you any time.

In an ideal world, an engineering solution should be as elegantly simple and inexpensive as possible. It should always efficiently match the problem. That's what makes it an art as much as a science.

But, in the real world....

11/17/2010 10:58 AM EST

I agree with Jack and believe the 8-bit controllers will have a long life yet. More often than not, you are looking for the next big controller for your next project which is pretty much the same as your current project but you need an extra uart or an additional I2C interface. Not every system will (ever) need an RTOS or even a stack running on the uC. Cypress Semi, for example, is betting on the trend that people want to integrate analog functions and peripheral functions and an 8-bit uC in one device A(PSoC). It has recently announced shipping the 750 millionth PSoC and it started making the PSoC only in this decade.

11/18/2010 4:50 AM EST

8-bits will never die. Large industries (automotive, consumer appliances) are cost driven. No automotive vendor will use 32-bit platform for simple CAN node endpoint or window lifting unit or anything else that I was working on before in my career. The bottom line is that MCU suppliers will set their prices based on the amount of silicone they need to manufacture the micro, period. This will never go lower than 8-bit unit.

Also someone mentioned that it's difficult to put good amount of functionality in limited ROM/RAM when using 8-bit. Now I work for FSL and the amount of complexity related to 32-bit platforms, their derivatives and issues related to different compilers is sometimes overwhelming... Not to mention HW/SW safety standards that are arising in some industries now.

The bottom line? Get real - no one will pay extra for not needed processing power or additional develompent costs to ensure something that is easily achieved on dinosaurs.

11/19/2010 6:46 AM EST

All said and done it is up to us engineers ( especially in the senior level) to keeps these 8-bitters alive. Don't bow to the management pressure from the top and don't have the fear of being ridiculed by the younger recruits who may have done their college projects with the latest 32 bitters just to get good grade.

11/20/2010 4:59 PM EST

Jack you have previously issued a standing mandate for us engineers to use the best tools. I suspect think the best tools are NOT those available for 8-bit machines, are they? (assuming non-ASM development I mean).
Enlighten me, someone; for most of us it is tough to get to use several different tool chains, let alone become productive enough with them to be able compare them.
I guess defining 'best tools' would be a mega-thread of itself...

11/20/2010 11:09 PM EST

I agree with Bala. Integrating more functionality (analog and others) on 8-bit micro in more desirable. For majority of my designs I mainly use 8-bit micros. I only sometimes need 16 bit and only thrice I have employed Popwer PC as part of Xilinx Virtex II Pro.

Understanig need of indusrtry and application is more crucial and atleast for next 10 years, I will emplying 8-bit micro controller to solve many new requirements.

11/21/2010 10:35 PM EST

Yes when I read the title, I doubted. Its very difficult to kill 8 bit. Unless you are dealing with video and audio processing or any other high end processing, you can do a lot with 8 bit. Clear architechture, many interrupts function, sleep mode, power down mode, its fascinating to work with. I think every engineer starts his/her career with 8 bit.