ARM: The Non-Evil Monopolist 452
yootje writes "ZDNet is running an article about ARM, a chip-maker who controls more than 80% of the cell phone market and 40% of the digital camera market. ARM shipped 780,000,000 processors last year. ZDNet finds it strange that no one seems to have anything against this company. And maybe it is strange: according to the article many would say ARM is a monopolist, but you never hear anyone say 'ARM sucks!'. But then again, why would they?"
Re:Well let me be the first of many to say... (Score:3, Interesting)
Which is far too often, actually. However, I am led to think that this is always (barring physical damage) a software problem. Some might even add: a problem with Microsoft software. I don't know anybody who has a phone with Microsoft software, but the software on Nokia phones does crash, silently fail (e.g. phone not receiving calls while indicating everything is fine), and have strange limitations (no more messages can be stored, even though a few hundred KB of memory available).
Student's ARM7 clone disappears from Web (Score:5, Interesting)
Because it was part invented by a lady (Score:5, Interesting)
Following on from her success with BBC Basic, Sophie Wilson [dotgeek.org] was asked to help with the instruction set, testing it by hand, on paper !
Re:Why nobody complains (Score:3, Interesting)
http://www.us.design-reuse.com/news/news277.htm
Why is a monopoly bad by default? (Score:3, Interesting)
I don't think people dislike monopolies. They dislike what monopolies have come to represent, and what they can lead to. I don't hate the idea of a monopoly. Do you? I just don't like the apparent and usually inevitable consequences.
Monopolies aren't inherently evil, just like dictators. It's just that in almost every example of their existence, they have shown to be detrimental to individuals, businesses, or society as a whole. A "benevolent", utilitarian dictator with the intent to make life better for his/her people could be beneficial to society. He/she would not be limited by legislation, and could focus on working towards a better future without worrying about bureaucracy or red tape. History demonstrates that any good utilitarian tries to amass as much wealth and influence as possible in order to serve these exact purposes. The more power they have, the better job they can do to serve the people. Humans would do much better with a benevolent dictator that they could ever come close to with any semblance of democracy.
Of course, history also demonstrates that absolute power corrupts absolutely. The day the "benevolent" dictator decides that they've done enough for society and that it's time to serve themselves is the day that everything goes downhill. The unfortunate fact is that those who would make good dictators would never be ruthless enough to attain such power. If they were, they probably wouldn't be in the best interest of the public good.
A monopoly is not bad in theory. If a company or organization had a monopoly on... say, microchips, they could drive the technology much faster and better, because they would control every aspect of it. They wouldn't have to worry so much about their software being compatible with their hardware, because they always know exactly what processor is being used. They wouldn't have to fight with competitors over standards, and could add as much functionality as they wanted, setting their own standards.
Unfortunately, theory is not the real world. In practice, monopolies don't do things because they're in the best interest of the public. They do them because they're in the best interest of the company. (Or at least, the company's officers.) This leads to higher profits (theoretically), but lower customer satisfaction. Some side-effects include buggy software, products that fail or break sooner than they should, etc. Because of this, the getting-screwed-public gets fed up and starts hollering. Thus, everyone hates monopolies. But what if the products and services of a monopoly just worked? I'll bet John Q. Public wouldn't care one way or the other at that point.
The average person doesn't care if something goes well. They become livid when there's a problem. A customer won't usually do very much if a company does their job exceedingly well. They will usually boycott the company and stage a rally if the company does poorly. I took an entrepreneurship class in 1992, and learned that the average person would tell 3 people when they were pleased with a product or service, but 11 when they were displeased. Since the internet became the next big thing (around 1994-1995) those numbers have probably skyrocketed. Humans are a loud, complaining bunch.
So is a monopoly bad? Not inherently, but they usually end up that way. I'd say that no one is going after ARM because their products just work and don't seem to cause problems. Their monopoly has not intentionally shut down any competition, or blatantly violated anti-trust laws. Until they screw us, I say more power to 'em.
Re:Not just a monopoly. (Score:5, Interesting)
Using your monopoly position in illegal anticompetitive ways however, is.
Sort of quasi-off-topic, but here goes:
In Finland we have a rather interesting and deliberate monopoly situation in regards to gambling. Slot machines, tables and casinos are all controlled by RAY [www.ray.fi] (decided by the state, I believe), but RAY on the other hand is a non-profit organization. RAY actially funds all sorts of cultural and social service activities. The same applies to Veikkaus [veikkaus.fi], which controls the lottery, betting on sports and similar stuff.
The result is that gambling in Finland is indirectly giving money to charity, weird, but nice in it's own way. I guess I'm just trying to say that even a regulated monopoly can be a good thing, sometimes anyhow.
they are in the list (Score:4, Interesting)
Arm designs chips and then lets other license them and make them. So intel and the others in the list got a simple choice. Do their own work or pay ARM for their work.
So plenty of competition and hardly small ones. It just seems that some of the big boys prefer think giving ARM money makes a better deal for them. After all it is not like IBM or Intel can't design their own chips.
Re:Shipped? (Score:5, Interesting)
indeed, i don't think ARM shipped any processors at all. ARM designs and licenses cores. from low powered arm7's in your run of the mill mp3 player, to a 400+mhz arm9/strongarm/xscale in high end pdas. arm-based chips are produced by dozens of manufacturers in many countries. arm cores run linux (and have a big developer community), wince, and multiple embedded operating systems.
i think the real failing of the linked story, however, is that ARM IS NOT A MONOPOLY. sure, they may ship more chips than anyone else. they make a good product. but in the embedded world, there is choice. mips, 68000, super-h, powerpc, dozens of proprietary architectures, even low end x86. if arm decided to pull some of the stuff that intel and microsoft try, they'd have the bottom pulled out of them as everyone migrates to their favorite arch of the day.
Re:Because it was part invented by a lady (Score:5, Interesting)
Phillip.
Re:I kind of like ARM (Score:3, Interesting)
The mysterious disappearance of Piccolo (Score:1, Interesting)
About 10 years ago they presented the Piccolo DSP add-on to their ARM processor. Using the AMBA processor bus you can add features to the processor core and Piccolo did so too.
Like the rest of the ARM designs it looked clean, tidy, elegant and actually rather inventive. As it hapened I was on a networking design for a new high speed compact system and we needed just something like Piccolo, so I immediately jumped on the opportunity, contacted them for samples and more docs
Having advocated intenally for ARM and put some of my rep on the line, this did cost me a bit and I am still disappointed in their rotten customer service.
While out of the business I still tracked development there and it appears that only small numbers of people used Piccolo. ARM itself now seems to disown it, you get no docs on ARM's own website on it.
I have been unable to determine what happened to the Piccolo and would appreciate any explanations from this community.
Re:Shipped? (Score:3, Interesting)
I don't think ARM had anything to do with it, in fact I'm pretty sure it's Intels attempt to take market share off ARM.
Re:Not just a monopoly. (Score:4, Interesting)
In an average criminal case, the DA has to prove that the accused had an opportunity to commit the crime (if the accused claims he didn't, at least). That doesn't mean that every person who had an opportunity to commit the crime did it, but that none of the persons who didn't have any opportunity did. There are probably 100,000 people who can't account for where they were at the time in the OJ Simpson case, and lived close enough to the crime scene that they could theoretically have had an opportunity, but that doesn't mean we should put them all on trial.
A company that isn't a monopoly has no way to commit certain antitrust violations, but a company that is a monopoly can. That's all it means, can and not did.
Instruction set is yummy (Score:1, Interesting)
The ARM instruction set is really nice to use. It's so simple that programming it is easy. After being exposed to this I couldn't go near x86. Of course it helped that Acorn computers had built-in assemblers, and you could knock up quick programs in which the listing could be half BASIC and half assember.
Phillip.
Re:Shipped? (Score:2, Interesting)
The reason why? (Score:4, Interesting)
Secondly, they are based in a country (the UK/EU) that actually UPHOLD it's competition laws; and thus they couldn't get away with what Micrsoft has in the US.
a couple of reasons I like ARMs (Score:1, Interesting)
- LDMFD R13!,{R0-R12,R14} - load multiple with full descending stack, stack pointer in R13, into registers R0-R12 and R14.
- ADDCS R0,R1,R2 LSL #4 - if carry flag clear, left logical shift R2 by 4 places, then add to R1 and put result in R0.
And these particular condition codes etc are not specific new CISCy instructions, but can be incorporated in *any* appropriate instruction - the barrel shifter doing the LSL for example being just another stage in the magical pipeline.This kind of elegance you just don't see in other instruction sets.
Re:I kind of like ARM (Score:5, Interesting)
ARM advertises -just not in the magazines you read! Further, ARM isn't a monopolist, they just happen to be the most successful and oldest of the companies that supply this type of item. There is also Tensilica, MIPS, and ARC to name three of their competitors.
They also have done a good job of propagating their technology by giving some of it away! What say you? Yep. They have published the specs for the AMBA bus which has become the defacto standard for connecting things together inside a chip.
Now -they didn't give away their own implementations of this stuff, but the spec is more than sufficient to build the structure in a couple of days.
Perhaps ARMs biggest success has indeed been their market path. They have done deals with every major chip manufacturer so that I can get access to their designs by merely paying royalties. I can go give them 750K up front if I want their IP to use myself, or I can pay maybe 50 cents a chip instead. This gives me a lower entry price with only the foundary guys paying the 750K. In one fashion they get paid twice!
In any case, they aren't the only ones on the market, merely the most successful.
Re:ARM--- (Score:2, Interesting)
The only difference is that they are still in the consumer electronics market, you do not have any kind of customization, or any speed requirements for most of their applications.
Do you remember the last time you hated some company for making that chipset in your VCR?
U R what U R accountable 2 (Score:3, Interesting)
Sure, ARM is easier going than alot of other outfits, and we don't notice them as much because they deal mostly with companies instead of individuals so their effects on peoples liberty aren't directly noticed as much. But's lets make no mistake about it, there is no nice way to enforce patents any more than there is a nice way to rape people, sooner or later something is going to half to give.
Further down thread you will find complaints about how ARM shut down an open source core project at www.opencores.com. As this kind of movement gets more and more popular - don't be supprised when their nasty side starts to reveal itself.
Why ARM is not resented. (Score:3, Interesting)
Think about the IT industry. Are the best products the ones that usually win?
We want to believe that merit is rewarded; and to some degree it is. But the only way we can believe this absolutely is either to close our eyes or to buy into a tautological definition of merit: merit is that that succeeds. There's no particular mystery as to why ARM is not resented: it is dominant and it has technically superior products. It confirms our cherished belief that if you build a technically superior product, you will win (ignoring the history of desktop ARM of course), and so we feel well disposed towards them. What really motivates contempt among the technologically sophisticate is not success, or copyrights, or patents, but when mediocrity wins and undermines our belief in the fairness of "the system".
A note about the tautological definition of merit. I was an MIS director in the late 80s early 90s. At the time we were on an exponential growth curve for personal computer adoption in business. Apple had a product that was superior to DOS (and later Windows) in so many ways it was laughable to compare the two. However, it cost more than twice as much to equip people with a Mac as with a PeeCee; in an era when a typical computer order was by the truckload, this was huge. The rest, as they say, is history.
Now, was Microsoft more meritorious than Apple? Well from the point of view of their shareholders there is no question this is true. The situation from their customers' standpoint is murkier.
Yes, the availability of a cheaper, lower cost personal computer probably sped the adoption of computers. We probably reached the 90% point of equipping office workers a couple of years earlier. Yet even back in the 80s there were studies even then that the cost of training and supporting users dwarfed the costs of buying the box, but most people could not quite bring themselves to believe this could possbly be true. Like in many things, the human capacity for inconsistency was amazing: it was not common for CEOs and top brass to have Macs and the troops to be equipped with PCs. Clearly they understood the value of quality to their own productivity. But, a loading dock full of computer crates was a tangible sign that you were making progress in "computerizing" your business.
Of course now people know the costs of training and support in spades, which is why you hear people bandying terms like "TCO" about when they thought this was mere flummery fifteen years ago. It's just the cost of other things like network security that they're blind about. I can confidently predict the world will settle on a cheap, half-assed solution to this problem and deal with the negative consequences for years to come.
The point is long winded story is not to say that Microsoft is evil for having succeeded with a product that was not very good. It's to point out that the tautological theory of merit ignores the way that real peole focus on the very short term. Often the company that wins is the one that keeps their customers focused on the short term. ARM in fact, got a lucky break; they are in a sense a failure in the desktop processor market, but succeeded by finding a niche in the embedded processor market. If for some reason their design drew several times the power, they could well be another entry on the roll of talented failures.
No monopoly in embedded (Score:4, Interesting)
I used to work on a high-volume embedded product. The first generation used a popular Motorola chip, the 2nd used one based on ARM. Most of our C-code remained unchanged when we switched cores. Just some hardware-abstraction layer stuff, and that was less than 5% of the code.
2 reasons (Score:1, Interesting)
2) Double standard because they don't suck.
Re:ARM--- (Score:3, Interesting)
The RIAA and the MPAA aren't trying to force anyone to listen to their music or watch their films, either. They're trying to enforce a physical-property model on an effortlessly duplicable product, but that's hardly evidence of anything resembling a monopoly.
Re:I kind of like ARM (Score:3, Interesting)
Intel doesn't seem to have such great CPU designers (they did start to get better with the 486 onwards. Still...), but they seem to have very good process and fab engineers.
ARM make quality products. (Score:1, Interesting)
I remember when the first ARM cheips came out. Back then the companies acronym meant Acorn Risc Microprocessors. Yes, the same Acorn that made the Atom, Electron, BBC and Master range of computers. The ARM first made it's appearance in the Archimedes range of desktop computers and spanked the equivalent Intel CPUs (Hmmm, doesn't that sound familiar).
When Apple were hunting round for a CPU for the Newton they came across the ARM chipset and "were so impressed they bought a piece of the company". So the company became Advanced Risc Microprosessors instead.
Their CPUs started urning up in all sorts of places, including car gearboxes (Porsche's Tiptronic gearbox) as well as numerous other devices.
The difference between them and other "monopolistic" companies is that they work with their customers instead of ignoring their customers and acting like a bunch of assholes. That is the way you win customers and grow in strength.
In fact, I am willing to be that the only line of CPUs that has come out of the bowels on Intel without a continuous stream of faults if the StrongARM based chips.
MUL arguments, LDRH displacement, nonorthogonality (Score:2, Interesting)
for some reason that I would expect has to do with how the instruction is implemented, one of the source operands (I forget whether it's the first or second) can't be the destination. Sure, you can swap the operands, unless you want to square a number (and don't need to keep the original around), but it's a pain.
Granted on x *= x, but in other cases, can't swapping mul operands be handled within the assembler rather than in the compiler, limiting the scope of the non-orthogonality?
Some sizes of load/store have 12-bit displacements; the ones added later, I think, (16-bit and one of the signednesses of 8-bit) only allow 5-bit displacements. A 5-bit displacement field is pretty darned small.
Granted, but what about those architectures that don't allow any nonzero displacement? Be glad you even have a displacement in order to allow quick access to structs. So in your code generator's documentation, give the following advice to programmers: when optimizing private structs for ARM, put "small" elements first to speed access.
Agreed, Thumb is designed to allow smaller code size, but the cost is major non-orthogonality.
Still, even in ARM mode, the code density still beats x86, if only for larger registers and thus fewer mov instructions to get variables on and off the stack. The 8- and 16-bit microcontrollers with which Thumb is designed to compete have even worse non-orthogonality.