Slashdot Mirror
SOCs: Say Goodbye To C's?
Rick Lehrbaum writes: "This [LinuxDevices.com] article describes a new class of Linux-friendly system-on-chip (SOC) ICs that are taking over the 1-chip microcontroller mantle from simpler architectures like the 8051 and 68HC11. And they're going to vastly accelerate the use of embedded Linux in thousands of new designs for intelligent devices, Internet appliances, and embedded systems. Devices covered in the article include include: Intel StrongARM SA-1110, NEC VR4181, STMicro STPC, Mot MPC823e, IBM PPC 405GP, NETsilicon NET+ARM, Aplio/TRIO, Axis ETRAX, LinkUp L7205, Alchemy Au1000, and Cirrus Maverick EP9312." I'd like a walkman-size computer based on that IBM 405GP that runs on AAs for a week ... sort of neat how open source OSes can seep into things like this.
For my senior project, I used an HC11 to a) receive and interpret X10 home automation (extended) codes and b) act as an LCD clock. The idea was to show that you could use many of these cheap devices anywhere in your home and they could all be kept in sync.
Best of all, it worked! With less than 1024 bytes (BYTES) of memory to play with... Imagine what you could do with 16 or 32k.
I mean, using C or any other high-level language, you can barely even compile a "Hello world" inside of 1k.
uCs may not be as elegant as a SPARC, but they've got their uses. Even if it's only for hobbyists and students. You gotta start somewhere.
Also, I don't know what kind of HC11 you're programming, but last I checked, B was an 8-bit register, while X was a 16-bit index register. It's gotta be tough for Motorola's engineers to justify an instruction that only copies B to X.
Besides, you could do that with: CLRA, XGDX. It's not at all counter-intuitive.
--
- Not scalable Your design implies that only one person can add data at a time. Which is fine for a small operation, but if you get a few additions per second then you either corrupt your file (two write handles open) or generate errors for the others. Allowing the others to be delayed is starting to write more backend stuff.
- Not instant Using your method means that data inserted into the system won't show up in the application right away - they have to wait until you do it manually. This means either more work for you, or another weak link in the chain, such as a script run from cron.
- Not universal You can only have one application writing to the file, otherwise you need some API or have multiple copies of the code (never a good idea.) SQL is universal across languages and platforms. SQL isn't quite, but pretty close to Java's Write Once Use Anywhere.
These days, using a database from any kind of application, especially form-handling web applications, is trivial.In my opinion, not scalable is a big problem.
Disclaimer: I am a design engineer on the embedded PowerPC team, interpret this accordingly.
:(
:)
Before I begin: don't be misled. The 68HC11 and the 405gp are two totally different ballparks. They do not compete in the same space.
faeryman sez:
I've followed the development of the for a while now, even having a few email conversations with Jonathon Thompson, Quong Ho Thoc, and Hagr Itstein (three lead developers). I told them about a few of my concerns but it looks like marketing prevailed
I am relatively new with the 4xx PowerPC team, but I've never heard of any of those people; I don't think they are developers (much less lead ones).
I don't see Linux being the right tool for this. I don't want to see this product fail since I know IBM is a good company. By all means everything else they made was a success, but the IBM 405GP looks like it will be a flop.
Umm, our customers sure seem to think it's the right tool. We got so much demand for Linux on 405 that we had to hire extra people to fully support Linux. As for 405gp being a flop, I don't know what planet you are on. 405 is selling so fast that it put a strain on our short term capacity. I don't consider a chip to be a 'flop' when Ericsson, Nokia, Cisco, and Alcatel use them in their products...
(1) Security - This is a big concern for me. Imagine some evil hacker getting control of this baby...now imagine if this was used in your bank or a military instituion. See the problem?
Umm, no, I don't. How exactly do you associate a SOC device with an Ethernet port automatically vulnerable to hackers? Is the 405gp somehow deficient in this regard?
While I commend the design of Open Souce, perhaps allowing the innerworkings of this to be accessable by a hacker is not good, even more so when it's an embedded system.
You are confusing connectivity with security. This article is about SOC's, and as far as their design is concerned they must be properly secured like any other computer system. Save the security tirade for a different forum.
(2) Expansion architecture - Check the specs on this thing. While a PCI slot is normally a good thing, wouldn't MCA or a propietary bus be better suited for this?
Are you f*cking kidding me? MCA? How many MCA devices can you buy? Not just cards, I means chips (which is what the vast majority of 40x's will be talking to). Almost zippo. Now how many different PCI devices do you think you can find?
Linux runs on the MCA fine, and I think it's low overhead and fault-tolerant properties are better than a run of the mill PCI slot for this. Or a new bus design could be implemented. IBM benefits with better performance, we as a comunity benefit from more GPL code being released. Sound good?
Absolutely not. The whole point of choosing PCI is because it is commodity, fast, reliable, and supported by almost every modern OS. It seems that you are desperate to reinvent the wheel here.
3) Operating system - [flamesuit] I like Linux, but I don't think Linux is the best tool for this. IBM has made the decision to go with Linux, so I'll respect that.
Like I said before, our customers want Linux. Linux is not the only OS we support. Actually you can put damn near any OS on the planet on it; IBM doesn't have support for them all however. You want a lighter weight OS than Linux? Fine, use OS/Open, which is IBM's little creation (works very well and supported too).
Scalibility and performance are key here, and QNX can deliver better than Linux.
Well, if you think so, then there's no reason you can't run it on 405gp.
Again, I don't like being negative but I don't think the IBM 405GP will do that well. I want to be proved wrong though, I want to see Linux progress and gain market share, and I want to see IBM be profitable....but Linux just ain't gonna cut it for this one my friends. Please tell me I'm wrong.
Well, since you asked so nicely...
The opinions I post here have nothing to do with my employer.
I've been building in PICs and AVR microcontrollers into a bunch of devices, and I can tell you, for sure, that they still have a long lifetime ahead of them. I've designed devices that are made in 10's quantities and in 10,000+ quantities and I have a few comments:
When you are designing with microcontrollers, you use the smallest and cheapest that will do the job. It is all about the appropriate use of technology. You don't need Linux to run your microwave oven. 99% of the microprocessors used in embedded systems don't need that much power. They aren't located in PC/104 bussed computers, they aren't in computer racks, they are in devices that are all around us, but not noticed: your microwave, your cell phone battery charger, your car alarm remote... and so on. Price is a very sensitive issue. These system-on-chip devices are very expensive- running $50+ each! If all I need is a $0.73 PIC to do the job, you're a fool (and soon to be unemployed) if you don't use the PIC (or AVR, or COP8, or whatever the latest, cheapest part is)!
The power of a real operating system is undisputed, but use it where appropriate! Rick Lehrbaum's white paper on using 75-200 MHz SOCs to replace 68HC11s and 8051s is ludicrous. If not for the simple fact that 99.9% of your clock cycles would be wasted, think about all the power (electrical) that would be wasted. Sure, Transmeta has some impressive MIPS/Watt numbers, but it doesn't scale well as you go lower. Many applications just don't need that much power. I can run a PIC off a 32 KHz crystal and only draw 50 microwatts off a power source. Not a one of the processors that Rick Lehrbaum mentions will be able to approach that low a power draw, even with a stopped clock.
Microcontrollers are going to be here a long time, just like we still use discrete transistors when we need to. Yes there are some applications that can use these systems on a chip, but for full acceptance they will have to be *cheap*, coming in at a price less than $10.00 each, and preferably less thann $5.00. It will take years for that to happen, and even then, we will still be using microcontrollers. I don't need 20 MIPS to run my microwave or my battery charger, or even my watch.
I design products with 8 bit devices, and I've used a couple larger chips here and there. There are many important features that designers need in microcontrollers:
- LOW COST !! 50 cents less times 10k units/month is a big deal. You could task an engineer for a couple man-months to rewrite code for a 50 cent lower CPU, after an initial release. Usually using an 8 bit chip instead of a 32 bit solution saves at least $10, sometimes much more.
- Low power consumption, under 4 mA is what I generally consider low power. A few years ago I designed a product that uses 9 A, 32 kHz CPU, wakes from full shutdown at 4 Hz. Try that with a linux-capable 32 bit chip! Low power also means a low cost power supply... at 1-2 mA, a resistor and zener diode can sometimes be used!
- Multiple vendors, or at least some assurance that the components will be available in the required quantities.
- Programmable program memory on-board... with in-circuit firmware download is a plus.
- Small physical size
Often times these considerations at much more important than cpu horseposer. For example, Microchip took off about 6 years ago, offering one of the most limited feature-poor instruction sets on the market, but they did all the important things very well. Their chips were cheap, low power, small size, and they offered EPROM based chips at low prices.... but the most important thing they did was they offered flexible purchasing, made possible by selling blank EPROM based devices, at a time when Motorola had inflexible purchasing requirements for masked-rom based parts.Most projects in the embedded market just don't need a lot of CPU power. At high volumes, it's easy to pay even the most expensive engineers and programmers to re-write code to run on a cheaper chip.
I should probably disclose that I have a small website with 8051 related resources (open source), so take my words with a grain of salt, but until 32 bit microcontrollers are less expensive and use less power than their 8 bit competitors, I'd expect the bulk of the market will probably stay with the 8 bit chips.
PJRC: Electronic Projects, 8051 Microcontroller Tools
Okay. I know you will very well call me a stick in the mud for this one, but I must be a bit more pessimistic than the article or the general air for the IBM 405GP is.
:(
I've followed the development of the for a while now, even having a few email conversations with Jonathon Thompson, Quong Ho Thoc, and Hagr Itstein (three lead developers). I told them about a few of my concerns but it looks like marketing prevailed
While yes, I am a fan of Linux and OSS (hell, I've used been running Slackware since version 2 and my firewalls run OpenBSD), I don't see Linux being the right tool for this. I don't want to see this product fail since I know IBM is a good company. By all means everything else they made was a success, but the IBM 405GP looks like it will be a flop.
Why?
(1) Security - This is a big concern for me. Imagine some evil hacker getting control of this baby...now imagine if this was used in your bank or a military instituion. See the problem? While I commend the design of Open Souce, perhaps allowing the innerworkings of this to be accessable by a hacker is not good, even more so when it's an embedded system.
Check out these sites, they explain why the needs for your desktop's security (which Linux can provide) are on the other end of the spectrum for bank/B2B/military security (which Linux cannot provide):
The CIA's spin
Military disablement
cpsr.org
(2) Expansion architecture - Check the specs on this thing. While a PCI slot is normally a good thing, wouldn't MCA or a propietary bus be better suited for this? Linux runs on the MCA fine, and I think it's low overhead and fault-tolerant properties are better than a run of the mill PCI slot for this. Or a new bus design could be implemented. IBM benefits with better performance, we as a comunity benefit from more GPL code being released. Sound good?
(3) Operating system - [flamesuit] I like Linux, but I don't think Linux is the best tool for this. IBM has made the decision to go with Linux, so I'll respect that. But I must say that WindowsCE or QNX would be better. We know who WindowsCE is backed by, but I must admit Mico$oft'$ embedded OS department knows thier stuff. Look at the recent Sharp handhelds - fine work and I think the same design could be applied to the IBM 405GP. If you don't want to recognize MS products though, I can understand. QNX would be just as valid (and in some ways such as power usage and latency) even better than WindowsCE and Linux. Scalibility and performance are key here, and QNX can deliver better than Linux. [/flamesuit]
Again, I don't like being negative but I don't think the IBM 405GP will do that well. I want to be proved wrong though, I want to see Linux progress and gain market share, and I want to see IBM be profitable....but Linux just ain't gonna cut it for this one my friends. Please tell me I'm wrong.
,
faeryman
I'll agree with you there to an extent. My point is that if I have some boring little website that does not get that much traffic, and the data I have is not that critical, then why do I need to set up a full-fledged database system? For your concerns with two write handles, you could use flock(), a second lockfile, or a combination of both to try and minimize that risk.
The applications I dealt with did not require any instantantaneous access to the data by anyone, so the extra step of copying a week's worth of data (mainly surveys) and importing it into a local database was acceptable to the client. They did not see the point in receiving every last response immediately; it would likely bog them down if they dealt with this data on a daily basis. So in this case, I feel justified in using a system like this; there was no SQL access set up on their systems, so I did not have to create it solely for this task. MySQL or Postgres would have been another thing for their administrators to watch which was not needed.
Linux is a huge beast. Look at it: You have support for IPv4, IPv6, IPX, parallel ports, serial ports, interprocess communication, filesystems, ethernet, etc. You can modularly add or remove these features, but only to an extent. You can run Linux on a 386, but you still need at least 8 MB of RAM to do basic functions. Many microcontrollers at most address 1 MB or 2 MB of memory! That includes your RAM, ROM, etc. Normally, not even half this address space is used. Operating systems designed from the start to fit within the limitations of a these systems (some with as little as 32 KB of RAM or below still - the 68HC11's I used only had 2 KB of EEPROM space onboard) are likely to do a better job than those that are modified and stripped down to do so.
While a high-end consumer device that needs ethernet access might be a good canidate for one of these new Linux-running chips, there will always be room for the smaller and older microcontollers and microprocessors. Remember the Z80? This microprocessor ran Timex's computers back when 16 KB would cost you US $100. It is still available today; Texas Instruments uses it in their calculators that cost about US $80 for the entire thing. Likewise, old microcontrollers are used because they are ready available in bulk -- cheap. A microcontroller for Linux may be a great idea, but likely costs a fortune.
For example, a modern Z180 (with two serial ports built in, a board with their C routines, RTOS, etc.) in quantities of 1,000+ would likely cost me US $50 -- each. That's half the cost of your modern microwave. Compare it to the lowly PIC's we use -- in the same quantities, these chips only cost about US $8 each in the one-time programmable variety. It isn't as fancy, but would be fine for controlling your average clock radio or answering machine. Which would you want in the next item you buy?
"Which means nine months from now (products take roughly the same time to gestate as human babies), the results of this frenzy of post-PC development will begin to emerge in a big way.
Obviously my problem is not that all my products are defective, they're just premature!"I'll take the red pill. No! Blue! AAAaaaahhhhhhhhh"
- Monty Python meets the Matrix
While doing web scripts, I often find myself writing simple databases. These forms do trivial things like take a users form and add it (comma or tab seperated) to the end of a text file. While I could have used a complete SQL backend, I chose the simple append to file approach. This is because my forms were purely meant to be imported into a database on another system - there was no need for them to be entered in a manner where they would be quickly searchable locally.
So instead of connecting to an SQL server, logging in, sending the command "INSERT INTO mytable VALUES data_1, data2,..., data_n;", waiting to hear if it worked, and closing the connection, I simply appended a line to a file. When I wanted to read the file, I downloaded it, viewed it locally, and zeroed the online copy so it could be filled again. What is wrong with that?
Compare this to my work with microcontrollers. I do work on Z180's, the PIC series, Basic STAMPs, and the 68HC11's (you can get a good student deal on these from Motorola - ask them). I have done work in both C and pure assembler (or in the case of the stamp, their BASIC). Guess whose programs comes out largest? Those in C. While the assembler routine itself for the task at hand is similar, a bunch of additional preloading code added by the C compiler is added. Imagine how much bloat a crude real-time operating system (RTOS) such as Linux would add if I did not need it.
If I'm purely watching inputs and outputs, and need to scan a few interrupts, I do not see the need to have Linux in my design. Granted, I'm a huge Linux user myself, but putting a stripped-down version in a microcontroller seems to be like shoving an elephant into a tin can. Real-time OS's for microcontrollers have been around for a while; some are designed to take up less than 2 KB. Why do we need to adapt Linux to a task that has already been solved?
We've been headed for systems on a chip for quite some time know. I remember, when working for Creative Labs, that they had a chip version of the Sound Blaster Pro waiting in the wings to be placed on a motherboard (as opposed to a card). But there was no demand for it! Looking at my IBM desktop, I'd say its no logner the case. (Sound card is integrated.)
I'd have to say though, as much as I like technology, the thought of all my appliances having a fulling running OS of some sort and hooked up to a network really really scares me.
...annoying microcoulombs. If there's one metric unit I can't stand, it's the C. In fact, the prefix should be removed entirely, because you need to use the HTML character entity µ, which timothy doesn't seem to know about.
Gates' Law: Every 18 months, the speed of software halves.