>
If I could find a nice board and a reasonably priced PowerPC or other RISC chip that I could run GNU/Linux on and get reasonable performance I would seriously consider it.
It takes some searching around, but you do can find PowerPC (AmigaOne G3 SE, Pegasos), Alpha (API) and SPARC motherboards around. There is also QliTech who preinstalls Debian and other distros in Apple Macintoshes.
>
I'll go along, however, and speculate that they're not recouping their R&D costs at the moment
They are not supposed to recoup such gigantic R&D costs so soon. I don't know exactly what is the expected horizon for recouping investment in a totally new architecture, and moreover here not only Intel but HP too made huge investments.
Rather the point is that Intel can only hope to recoup such investments in any reasonable future -- because it has to keep investing in future generations to keep up with RISC developments -- if it counts on using its monopoly cash hoard to enable it to extend its monopoly on to the 64-bits arena, where it doesn't have a credible history.
That is an unfair advantage no other competitor has, partly because competitors where too stupid, partly because Intel pigbacked on Microsoft's OEM dominance and anticompetitive plots, partly because they have had their own anticompetitive plots like they used to kill Clipper and then break promises made to Intergraph, and finally they resorted to buy their most credible competition, the Alpha and the StrongARM.
I don't really care for Digital or ARM or IBM or Sun, because they have been stupid enough to let all this happen; but in the end it is bad for users because of high costs, high noise, and low performance.
>
If dumping is illegal in domestic trade I'm sure Intel is selling just at or above actual costs of manufacture.
Unfortunately I don't think dumping is illegal internally. Europe could have a say here, since it imports its IPFs IA-32 and IA-64, but still this market is too dynamic for such legal proceedings. The point rather is that it is immoral, and companies should have immoral things they do sticking harder to them.
But mind you, they are probably more than recouping their manufacturing costs, high as they are. What they are certainly doing is burning a large part of their monopoly cash hoard by not recouping R&D, and that to try to extend that self same monopoly.
>
However, Intel isn't a monopoly on the desktop. A little over a week ago AMD anounced it has about 19% market share.
IBM was considered a monopoly when it had 70% of its market, and then they had several major competitors. Intel now has around the same, and even less competitors.
>
Just because it's their architecture that has a monopoly on the PC market doesn't mean that they have a monopoly.
Yes, it means. It is their architecture. They set the trends, they license their own competitors. Look at their current fight with VIA, and their attempts at blocking AMD in the courts. The only reason why they don't have 100% of that market is that they where too liberal with licensing in the 1.980's, but they are not repeating the "mistake" now. If they succeed with IPF IA-64, there will be little place left for the likes of AMD and VIA, and competition would dwindle even more, consequently technical excellence too.
That is why AMD is so keen on going forward with x86-64, even if it means perpetuating an inferior architecture: it is their only chance of survival, short of a total upheaval in the market that would make RISC popular again, presumably under GNU/Linux, and open opportunities for alternative chip designers and second-sources. Unfortunately that is not likely to happen soon, unless open systems suddenly become enforced again. But then perhaps they count on the x86-64 giving them the upper hand over Intel, thus enabling them to eventually push some RISC architecture as a migration path. I wouldn't count on it, though.
>
Had you made the assertion before the Athlon, however, I would have been quite inclined to agree.
As I explained above, the IPF IA-64 is, among many other things, a plan to exclude AMD and other competitors. The Athlon as direct competition sure puts pressure on Intel, and so does the x86-64, but not enough to offset the dismissal of Clipper, MIPS, Alpha, PA-RISC and StrongARM as less direct but ultimately more fundamental competition.
>
See, now you have to go and be a jerk. I hadn't read about VLIW, and thus EPIC's compiler optimizations not carrying forward generations. A simple pointing it out would have sufficed.
Sorry, I apologise.
>
If you could supply a link with detail I would appreciate it.
Besides Digital's Alpha vs IA-64 paper I already supplied you, see a balanced view. It is hard to find good stuff nowadays, because since Intel's PR machine started grinded, Google results got swamped by hypings and "neutral" (bowdlerised) stuff. But if you spend some time looking and reading you will find more.
>
Why Intel, upon aquiring a lot of the talent that went into Alpha, isn't going to do more to further what the Alpha had going is beyond me. Intel would have a hell of an offering if they extended the Alpha.
Basically it is a long time since Intel was an engineer-friendly place, most of the good talent left a long time ago. And even before Intel, Digital had already gone awry and lost many engineers, including the ones who went to AMD and made the Athlon.
Also, a mixture of Not-Invented-Here Syndrome, Featurism Complex and Control Freakism. With the Alpha, instead of only one partner (that's corporate Newspeak better translated into plain language as "People we need now but must stomp out in the future lest they become real competitors"), namely HP, they would have several: Digital/Compaq, HP, API, SamSung, IBM, Sun, SGI, ARM, etc. With IPF IA-64, they bought, neutralised or are cornering every one of them but IBM and Sun. Machiavelian.
>
I don't think that at this stage you can classify the Itanium line as inferior.
I think that if you do your homework you will agree with me that it is inferior indeed. Not only the Itanium line but the whole EPIC concept as applied to general computing. VLIW in DSPs is more than fine enough.
>
I see the Pentium 4 with a core voltage no higher than 1.75V and the G4 fixed at 1.8V.
Voltage by itself means only Intel has perfected their manufacturing process better than Motorola. Nothing about the architecture. And this with they having vastly superior resources, and with all Motorola recent missteps, is really either a compliment to Motorola or a pointing finger at Intel.
Now if you suppose that in a fair world Motorola (or IBM) would have Intel's resources to develop processes for such low voltages as Intel's, you would see PowerPC consuming even less power and consequently generating even less heat, and consequently less noise.
>
I'm not sure I'd call it absurd to have a machine as powerful as the IA-32 line with power requirements at least in the neighborhood of where Apple's are.
The point here is that, even with much less development resources, the PowerPC architecture is so vastly superior that it still gives you a comparable product. Your comparison would have to take into account manufacturing processes to be valid. Just as a means of comparison, see that Apple machines are much silenter than PCs, by virtue of needing less cooling. That with having less advanced manufacturing.
>
I think the part that people miss about the Wright brothers is that they made the fundamental scientific discovery of how to make airplanes turn.
True enough. Too bad their discovery got muddled in the patents war and the consequent "who got there first" dispute.
>
Especially, if getting a patent costs more than an average inventor can affford.
I think the cost must be high anyway, at least for refused patents. One of the problems is too much applications; people should be charged for spurious entries lacking enough prior art research and clarity.
The problem isn't too few patents in the small guy's hands, but too much everywhere. If you do any work with enough visibility, you are liable to get sued for infringement, and the court costs of fighting are so high you end up paying or desisting even because of totally worthless patents.
This keeps the barrier to entry too high, so that nowadays there are big corporations, lots of small box-shifting VARs and retailers, and nothing in between.
In contrast, Digital got started hardly bothered by patents, Sun got by but had to pay IBM US$20M when they were yet quite small in the middle 1.980s, but now, even if the field is technologically very immature, it is hardly possible that there will be another *innovative* Sun, Digital or even Oracle that won't get severily penalised by patents.
>
The problem is that Santos Dumont did not make any fundamental breakthroughs.
That is not true. 14-bis had ailerons before anyone else. He was also the first to use internal combustion engines, while still working in ballons, and to steer dirigibles around. The Demoiselle was also the first ultralight airplane. And BTW, among other inventions he created the wristwatch, commissioning his friend Cartier to manufacture the first wristwatch ever, to be used while flying.
>
according to this page [lycos.co.uk] the Demoiselles, which first flew in 1907, did not have ailerons, or wing warping. Just rudder and elevator.
If this was true, how could he reach 100km/h speed and 20km range as he did? But it was false. The Demoiselle used wing warping, simply because the ailerons used at the 14-bis weren't simple and small enough for such a light airplane.
>
The problem is that without exclusive manufacture and distribution how can one profit from an invention?
In a word, royalties. Obviously there should be an upper limit, and the option of contesting them in courts if they are deemed prohibitive for a given use.
>
Except that "using" cannot be as clearly defined as we'd like. Wrights thought Curtiss was using their invention, and Curtiss thought he didn't.
That would be an argument against patents, because they have the potential of throwing a whole field in disarray, like they are doing now for computing.
>
there seems that there is plenty of evidence that the Wrights were quite ahead of any competition.
I am not disputing that. I am saying that Santos-Dumont had a much better attitude by sharing what he did, and that served better the world than the Wright brothers secret, proprietary attitude.
>
When did 14 bis fly? 1906 or was it 1907?
October 1.906. But the Wright brothers only cared to show their airplanes in Europe in 1.908. By that time Santos-Dumont was already developing the Demoiselle, the first ultralight airplane.
>
Patents can be licensed. The Wrights expected to make money from their invention. Is that so bad?
Yes, because patents can, don't need to be licensed. It they were RAND, Reasonable And Non-Discriminatory, it would be much better. But the way it is, they actually managed to forbid other people from flying. That is umititaged evil, and contradicts the spirit of patent law.
>
You seem to oppose the idea of patents altogether.
No, but I think the current implementation of the concept is counterproductive. Even at the Wright brothers' time, it encouraged them to do things in secrecy, while Santos-Dumont worked in the open. There should be a provision that provided for patents to be granted only to work done publicly. The way it is, the patenting process is a powerful incentive to delay publication for years.
Also, patents should be always RAND. They should reward the inventor, not the manufacturer or distributor.
>
Do you think the patent on public key encryption, let's say, was OK, or not?
Software, processes and methods aren't in the original definition of patents, and have proved to be unmitigated evil. They should be stopped now, and all already granted revoked.
>
equivalent performance at a lower price hardly slaps the Itanium into a "inferior product" category.
Acquisition price is hardly the whole story. First, it is heavily subsidized by the IA-32 business, so it is no indicative of production costs. Second, there is cost of integration, where size and complexity does matter, and cost of operation, where size, power consumption and heat generation does count.
>
what does size have to do with it anyway? A bigger die size will mean lower yield and increased cost, perhaps lower clock rate, but it was already mentioned that the Itanium is selling for less and performing comparably to the Power4.
You just proved my point that Intel is leveraging its monopoly to subsidise its inferior products. In foreign trade the US uses to high taxes and even forbid buying in these conditions -- that is called "dumping".
>
compatability to processor family? Itanium is its own processor family. Itanium 2 will follow the same IA-64 set as far as I know.
That proves you know little... the biggest con of VLIW, of which EPIC is a variant, is that the compiler optimisations used for a generation aren't valid for future generations. That is in contrast to RISC and even CISC, because when optimising for a generation of processors used to benefit future processors, it may actually worsen the performance of future processors in a VLIW family.
Additionally, usually the RISC system builders use better components, do better-balanced systems, and better quality control than the Intel me-toos. Particularly, while the Alpha has traditionally been better than anything else in floating-point, the POWER has excelled at integers. All in all, I wouldn't be surprised to learn that the POWER systems being benchmarked today endured more with less problems than their Itanium equivalents, and they should perform better in business loads. Scientifical loads would have been better hadn't Intel let the Alpha stagnate.
Granted HP's EPIC, of which Intel IPF IA-64 is but an implementation, tries to minimise that. But the price is even more complexity than the original RISCs, like the Alpha. So why bother with going EPIC instead of going RISC? Intel could have pushed the Alpha instead of flogging its own pet dead horse.
>
Calling Intel a monopoly in this area and then upholding IBM's offering doesn't exactly hold water.
You missed the point. Even if perhaps more due to MS DOS and Windows history and to the stupidity of both IBM and Digital in their competitive offerings, the fact is that Intel has a monopoly in the CPU mass market. Its only opositor there is again the PowerPC, but then it is currently confined to the Macintosh and some Amiga and RISC GNU/Linux diehards. Intel has even bought the StrongARM to be able to extend its position into the PDA space, and has already succeeded there too. Now it is doing the same by buying the competition, namely the Alpha and HP EPIC, the first of which it has killed and the second it adopted. And it is using its resources gained in long dominance of the mass market to pracitise dumping of the high-end marke with the Itanium. That is unfair, if no illegal. In foreign trade, it is definetively illegal.
>
the power consumption by IA-32 processors isn't that absurd.
Yes, it is. Compare it with equivalent PowerPC and StrongARM. No matter what other economies you can get elsewhere, it is still a stupid waste of energy that is maintained purely for marchitectural reasons.
>
Power4 and G5 are totally different CPUs, though they have (nearly) the same instruction set. At the gate level, they share nothing.
The point here is that Apple is tending towards use of IBM processors in its future high-end lineup, already using them in the low end now. That because Morotora has being paying more attention to the embedded market. This will realise economies of scale for other users of the POWER architecture as well.
>
Octave Chanute reported back to the Wrights that, while Santos-Dumont had indeed flown, he had no means of controlling the aircraft except by shifting his weight.
And his report was misinformed. The 14-bis had already ailerons. The difficulty of steering resided mostly in the fact that the pilot was standing, thus incapable of using his feet. The body contortioning was not weight shifting, but activating the control surfaces by strings attached to the pilots' clothes. Awkward, but given it was independent development, and since the Wright brothers were working in secrecy and eager to control airplanes everywhere, European courts were sensible enough to deny their claims.
>
The Wrights had flown a fully controlled glider in 1902! Nobody else had a clue how to do that until the Wrights have shown them.
False. They did they work in secrecy, so necessarily all the European designs, and much of the US ones, were parallel developments.
>
The Wright's patent was on the method and mechanism of controling an airplane in flight, not on the aileron. Aileron is just another implementation of the Wright's system.
So much worse. That means that, had they succeeded in their patents, they would have been able to forestall aviation during 17 years both in US and Europe.
>
I happen to think that patenting their invention was the right thing to do. It was quite an achievement. This was not a "one-click" patent. The Wrights risked they lives to perfect their invention.
The point is that it was not "theirs". It was made of incremental improvements over other people's works preceding and in parallel.
Take the aileron, for one. Patented in the US by Bell in 1.911, invented in 1.909, ignored by the Wrights even if essential for the big airplanes they planned... and already present in the 1.906 Santos-Dumont 14-bis, who favored wing warping in the Demoiselle because this was such a small airplane. With the aileron, if Santos-Dumont had patented it, any Wright patents would have to be exchanged for the aileron patent in order to allow for practical big airplanes and thus generalised, since Santos-Dumont made a point of making his work available for everyone.
But all this was unnecessary, because the European courts saw thru the folly and did not accept the Wright patents.
Had their patent succeeded in Europe and stood in the US, it would have delayed the whole field. So much for patents fostering Science and the Useful Arts...
Re:"Benefits" of killing the Alpha and PA-RISC...
on
Itanium Problems
·
· Score: 2
>
Steven M. Milunovich never heard the axiom: "You need to spend money to make money,"
True enough. They save so much money, but kill their competitiveness against the Power4 and UltraSPARC IV, and at the same time much or all of their differentiation against other IPF IA-64 builders like Unisys, Dell, and even IBM itself.
There is little reason -- beyond inertia -- current Digital, Compaq and HP-UX users will not migrate to IBM, which has it all in house, or Sun, which focus better on RISC on Unix, or Dell, which focus better on Wintel. Thus they save a lot of cash, but loose lots of revenue. True enough most mergers go bad.
I probably forget others, but Unisys for one has built much bigger, meaner machines in the IA-32 space. Not to mention Penguin Computing... pity they could not sell enough of their 8-way systems. In order to sell this they should have centered on applications, perhaps with PostgreSQL and SAP, say, on Debian. But there is only so much one can do.
Re:"Benefits" of killing the Alpha and PA-RISC...
on
Itanium Problems
·
· Score: 2
>
Steven M. Milunovich never heard the axiom: "You need to spend money to make money,"
True enough. They save so much money, but kill their competitiveness against the Power4 and UltraSPARC IV, and at the same time much or all of their differentiation against other IPF IA-64 builders like Unisys, Dell, and even IBM itself.
I probably forget others, but Unisys for one has built much bigger, meaner machines in the IA-32 space. Not to mention Penguin Computing... pity they could not sell enough of their 8-way systems. In order to sell this they should have centered on applications, perhaps with PostgreSQL and SAP, say, on Debian. But there is only so much one can do.
Re:"Benefits" of killing the Alpha and PA-RISC...
on
Itanium Problems
·
· Score: 2
>
My vague impression was that they didn't offer a complete package for the high end
Absolutely wrong. The high end was pretty well covered with wickedly fast, very reliable and well-balanced Digital Unix and OpenVMS systems. The problem is that there was no Sun Solaris for it, and this is basically about loyalty -- no one switches from SPARC Solaris, Power AIX or PA HP-UX just for performance. Ultimately the big failure was the failure of Digital to produce as reassuring a corporate roadmap as Sun, IBM or HP.
>
they didn't offer a good price/compatibility for the low end.
True enough, but not the whole truth. On one side they were stiffled by Microsoft never porting even the basics of its software porforlio. There was no MS Access, Visual Basic came too late, and code was not nearly as well tuned as for IA-32. The result was that, even if the systems themselves were competitively priced with the IA-32 ones, Microsoft Windows itself limited the performance and reliability advantage the hardware granted.
They did supported GNU/Linux, but at a time when it was strictly geek-only. And even so, they never had OEMs enough to build a strong platform. Also, lack of the seemingly infinite Intel-like scale of resouces left some important holes in their lineup, such as notebook and low-end processors.
Compaq could have pulled it with its resources, but its box-shifting corporate mentality eventually killed the Digital-inherited engineering heritage, and the whole thing went into suicide mode by pushing harder the lower-margin IA-32 side of business. Just as depressing as Digital shooting itself in the foot by calling Unix snake oil and failing to capitalise in Novell and Apple interest in using the Alpha.
>
There's not much like it, performance-wise, and on a cycle/dollar scale, it's in a class by itself. Smokes US-IIIs, walks away from the Alpha, and keeps pace handily with the Power4, at a more academicly-tolerable price.
So the issue here is price... but Itanium is still hotter and bigger than Power4. The comparision to Alpha I dismiss out of hand since the Alpha processors are neglected by Intel itself. But if the Power4 has equivalent performance being smaller, cooler, and more compatible in its own processor family, then we have again an issue of Intel leveraging its monopoly to push an inferior product, as Microsoft does with its Windows.
Now the Power4 is poised to win more economies of scale with the G5 or G6 generation going into Apple Macintoshes, AmigaOnes and POP GNU/Linux systems, besides the AIX ones. This will be interesting to watch.
It strikes me as absurd that all this global warming talk says nothing of the absurd power consumption by IPF & IA-32 processors. A concerted effort for a RISC migration, and for properly configuring aggressive power saving in all corporate and domestic systems, would go a long way.
>
if you use a modified OSL app on your website, you need to make your changes public. As I read the GPL, you don't need to make your changes public
Currently, that is GNU GPL v2, you are right. But the FSF has already endorsed the Apero FSL as a draft for this change in GNU GPL v3. This is to keep Internet ASPs honest.
>
Many words in aviation come from french (eg. fuselage etc). So what.
It means it was in Europe that aviation popularised.
>
The Wright's plane used wing-warping mechanism (so they weren't ailerons).
Yes, just as Santos-Dumont nos. 19 to 22, the Demoiselles.
>
The point is that the Wrights understood how airplanes turn and they could build a machine that could do it.
Just as Santos-Dumont did. But when he did, he did not kept it to himself.
>
Santos Dumont's first heavier than air flight was in October 1906. By then Wright's patent was already granted and they were trying to sell their working airplane to the US Goverment.
And kept the whole aviation field in the USA in a check, having the power to forbid people from using their patents. Which should never been granted in the first place, because all they did was build upon other peoples work in parallel with people from other countries. The whole situation parallels the incredible injustice of the patent granted on the telephone by a few moments difference in the filing time.
That while Santos-Dumont produced the Demoiselle and gave away both the concept and the projects for everyone to use, thus precluding Wright brothers attempts at hindering the progress of aviation in Europe too.
>
The functional and performance difference between the two approaches isn't really that huge anymore, since this CISC->RISC translation doesn't slow things down a whole lot.
Wrong, it does slow things a lot, ceteris paribus. As ceteris are notparibus, Intel and AMD just throw a lot of resources to get CISC chips as fast as RISC ones. In order to do this they make big, expensive-to-project-and-manufacture, hot, energy-hogs of some chips.
Just when mainframes are not water-cooled anymore, and there are RISC processors that achieve even better-than-mainframe performance, if you wanted to get mainframe performance from IPF (IA-32 & IA-68, or x86 & EPIC) processors you would need water cooling, or nearly so.
>
The Itanium is the first to abandon that approach, and say "it's up to the compiler to make sure stuff doesn't mess up when we pipeline."
Yes, but what are the trade-offs? The IA-32-to -EPIC translation layer still makes for a too-big processor, and even without it lots of efficiency are lost in attempts to ease off the burden on the compiler and the lack of consistency among chip generations. Alpha in particular, and RISC in general, are still much more efficient and elegant.
Could not agree more. Moderators, please, Insightful +1 for the parent until he gets at least four.
I would add that the basic issue is free market without neither ethics to strive for the better, nor education to be able to choose and attain it. In absence of ethics and education, justice and schooling would do as a second best, but even this is being lost.
>
that's how Intel and AMD can be competitive (and often beat) their current RISC competitors.
But they are not competitive. They only beat their RISC processor by the, ironically, Chinese Army strategy: throwing an absurd level of resources, creating an absurdly big, hot chip and producing it at nicely big scales.
If Intel would suddenly change heart and put in place a migration path to, say, the Alpha with MS WXP, Debian GNU/Linux running Gnome2, DEC OSF/1 Unix and the likes of them -- or if IBM & Apple got a clue to start selling nice cheap PowerPC microsystems that could run AIX, Mac OS X, preconfigured Debian Gnome2 GNU/Linux, Amiga and the likes of them --, thus getting the same economies of scale as the x86 processors enjoy, then we would see in the market what is already true in the fabs and technical specifications: RISC processors are cheaper to manufacture, smaller, more efficient.
>
RISC instructions are so simple that small programs compile to large binaries. This means that the code cache needs to be several times as large to hold the same amount of high level code. This is where the RISC with run-time CISC to RISC hardware translation actually out-performs a pure RISC implementation.
This is stupid. The bigger RISC cache is more than compensated by totally avoiding needing the CISC-to-RISC translation, which takes much more chip real state then the cache, adds much more complexity to the project, and in addition slows down the whole kabooza, by adding both processing stages and development delays.
>
Take a look at the newest Dell machines. They don't have a fan on the heatsink (which is necessarily small and fast, and thus noisy). Instead, they have a very large, slow fan near a gian heatsink. The whole setup is *really* quiet.
Very interesting, thanks for the info.
But the fact remains that an equivalent RISC processor would be smaller, cheaper to manufacture ceteris paribus -- that is, given equivalent economies of scale --, use up less energy and generate less heat, thus requiring either a smaller heatsink and therefore enclosure, or a even slower, smaller fan that turns on less often, or even none fan at all as in the Apple Cube.
>
The government owns the big companies; that is the definition of communism (ownership of the means of production by the "proletariat").
Not by any means. In Communist theory, they have tried the Proletariat Dictatorship, which is called this side of the Bamboo Curtain by the less hopeful name of Totalitarian State Capitalism. Communism would be a further stage after education has eradicated egotism, and then the production means would be owned by the communities -- hence Communism.
>
None of that has changed. Call me when it does.
What did change was that the Party has decided that the Proletariat Dictatorship has failed in almost all goals except keeping a small clique in power, and the failure was risking even the inner ring. So they are keeping both the Communist Newspeak and the totalitarian government, but changing the production mode into a form of Mercantilism, where the government cooperates with a few chosen business to advance what is officially national, or popular, interest.
Come to think of that, even the US is becoming Mercantilist with all the ever-crescent restrictions to immigration and trade.
>
If the hardware design habits of the Chinese are anything like their software programming efforts, then the Dragon will be reverse-engineered and rebranded Pentium.
This being not a for-profit fly-by-night sweatshop, but a research institute, rumour has it that they cloned Alpha.
I hope they did, because there is no microprocessor architecture that holds more promise then the Alpha, and it is a shame on the US supposedly pro-competitive, efficient culture that it has been cancelled due to Digital being inefficient in marketing it and then Intel not wanting the competition.
>
You used the plural of "glory", in the following sentence: I wonder why only First-World Westerners are allowed any glories.
Oh, yes. That phrase. Tip: it was intended to be ironic, because not only the poster I responded to defended the Wright brothers, he also dismissed Alberto Santos-Dumont works with derisive words. So I was questioning his exclusiveness to the point of debasing his opponents.
>
You thus begin a pointless contest with the Americans, which I now realize is probably your real goal.
I do not think it was pointless. Have you noticed that in defending the Wright brothers all these people from USA not only overlook the moral and scientific implications of secrecy and patents, they also tried to deny all of other peoples contributions? Just to name one, present in 14-bis 1.906 flights, the aileron.
>
I don't think it's accurate to say that either of them are "the real inventor" or "invented the airplane". Anyone who examines history with even the slightest degree of objectivity will realize that these are simplistic and trivial claims.
In another quote I said that, just like with the automobile, there was not one inventor to the airplane, but several. Now, trying to make either Mr Curtiss or the Wright brothers the sole inventor(s) is ridiculous and shows how much the USA as a whole is ignorant of the rest of the World and History.
>
the other pretenders to the throne built one or two failures or uncontrolled short hop planes and then retired from the field.
Nothing could be further from the truth. 14-bis gradually improved, and the Demoiselle was used practically for individual aerial transportation with full steering over several kilometers, being the first widely-used (for the time) light aircraft.
Slashdot Mirror
It takes some searching around, but you do can find PowerPC (AmigaOne G3 SE, Pegasos), Alpha (API) and SPARC motherboards around. There is also QliTech who preinstalls Debian and other distros in Apple Macintoshes.
They are not supposed to recoup such gigantic R&D costs so soon. I don't know exactly what is the expected horizon for recouping investment in a totally new architecture, and moreover here not only Intel but HP too made huge investments.
Rather the point is that Intel can only hope to recoup such investments in any reasonable future -- because it has to keep investing in future generations to keep up with RISC developments -- if it counts on using its monopoly cash hoard to enable it to extend its monopoly on to the 64-bits arena, where it doesn't have a credible history.
That is an unfair advantage no other competitor has, partly because competitors where too stupid, partly because Intel pigbacked on Microsoft's OEM dominance and anticompetitive plots, partly because they have had their own anticompetitive plots like they used to kill Clipper and then break promises made to Intergraph, and finally they resorted to buy their most credible competition, the Alpha and the StrongARM.
I don't really care for Digital or ARM or IBM or Sun, because they have been stupid enough to let all this happen; but in the end it is bad for users because of high costs, high noise, and low performance.
Unfortunately I don't think dumping is illegal internally. Europe could have a say here, since it imports its IPFs IA-32 and IA-64, but still this market is too dynamic for such legal proceedings. The point rather is that it is immoral, and companies should have immoral things they do sticking harder to them.
But mind you, they are probably more than recouping their manufacturing costs, high as they are. What they are certainly doing is burning a large part of their monopoly cash hoard by not recouping R&D, and that to try to extend that self same monopoly.
IBM was considered a monopoly when it had 70% of its market, and then they had several major competitors. Intel now has around the same, and even less competitors.
Yes, it means. It is their architecture. They set the trends, they license their own competitors. Look at their current fight with VIA, and their attempts at blocking AMD in the courts. The only reason why they don't have 100% of that market is that they where too liberal with licensing in the 1.980's, but they are not repeating the "mistake" now. If they succeed with IPF IA-64, there will be little place left for the likes of AMD and VIA, and competition would dwindle even more, consequently technical excellence too.
That is why AMD is so keen on going forward with x86-64, even if it means perpetuating an inferior architecture: it is their only chance of survival, short of a total upheaval in the market that would make RISC popular again, presumably under GNU/Linux, and open opportunities for alternative chip designers and second-sources. Unfortunately that is not likely to happen soon, unless open systems suddenly become enforced again. But then perhaps they count on the x86-64 giving them the upper hand over Intel, thus enabling them to eventually push some RISC architecture as a migration path. I wouldn't count on it, though.
As I explained above, the IPF IA-64 is, among many other things, a plan to exclude AMD and other competitors. The Athlon as direct competition sure puts pressure on Intel, and so does the x86-64, but not enough to offset the dismissal of Clipper, MIPS, Alpha, PA-RISC and StrongARM as less direct but ultimately more fundamental competition.
Sorry, I apologise.
Besides Digital's Alpha vs IA-64 paper I already supplied you, see a balanced view. It is hard to find good stuff nowadays, because since Intel's PR machine started grinded, Google results got swamped by hypings and "neutral" (bowdlerised) stuff. But if you spend some time looking and reading you will find more.
Basically it is a long time since Intel was an engineer-friendly place, most of the good talent left a long time ago. And even before Intel, Digital had already gone awry and lost many engineers, including the ones who went to AMD and made the Athlon.
Also, a mixture of Not-Invented-Here Syndrome, Featurism Complex and Control Freakism. With the Alpha, instead of only one partner (that's corporate Newspeak better translated into plain language as "People we need now but must stomp out in the future lest they become real competitors"), namely HP, they would have several: Digital/Compaq, HP, API, SamSung, IBM, Sun, SGI, ARM, etc. With IPF IA-64, they bought, neutralised or are cornering every one of them but IBM and Sun. Machiavelian.
I think that if you do your homework you will agree with me that it is inferior indeed. Not only the Itanium line but the whole EPIC concept as applied to general computing. VLIW in DSPs is more than fine enough.
Voltage by itself means only Intel has perfected their manufacturing process better than Motorola. Nothing about the architecture. And this with they having vastly superior resources, and with all Motorola recent missteps, is really either a compliment to Motorola or a pointing finger at Intel.
Now if you suppose that in a fair world Motorola (or IBM) would have Intel's resources to develop processes for such low voltages as Intel's, you would see PowerPC consuming even less power and consequently generating even less heat, and consequently less noise.
The point here is that, even with much less development resources, the PowerPC architecture is so vastly superior that it still gives you a comparable product. Your comparison would have to take into account manufacturing processes to be valid. Just as a means of comparison, see that Apple machines are much silenter than PCs, by virtue of needing less cooling. That with having less advanced manufacturing.
True enough. Too bad their discovery got muddled in the patents war and the consequent "who got there first" dispute.
I think the cost must be high anyway, at least for refused patents. One of the problems is too much applications; people should be charged for spurious entries lacking enough prior art research and clarity.
The problem isn't too few patents in the small guy's hands, but too much everywhere. If you do any work with enough visibility, you are liable to get sued for infringement, and the court costs of fighting are so high you end up paying or desisting even because of totally worthless patents.
This keeps the barrier to entry too high, so that nowadays there are big corporations, lots of small box-shifting VARs and retailers, and nothing in between.
In contrast, Digital got started hardly bothered by patents, Sun got by but had to pay IBM US$20M when they were yet quite small in the middle 1.980s, but now, even if the field is technologically very immature, it is hardly possible that there will be another *innovative* Sun, Digital or even Oracle that won't get severily penalised by patents.
That is not true. 14-bis had ailerons before anyone else. He was also the first to use internal combustion engines, while still working in ballons, and to steer dirigibles around. The Demoiselle was also the first ultralight airplane. And BTW, among other inventions he created the wristwatch, commissioning his friend Cartier to manufacture the first wristwatch ever, to be used while flying.
If this was true, how could he reach 100km/h speed and 20km range as he did? But it was false. The Demoiselle used wing warping, simply because the ailerons used at the 14-bis weren't simple and small enough for such a light airplane.
In a word, royalties. Obviously there should be an upper limit, and the option of contesting them in courts if they are deemed prohibitive for a given use.
That would be an argument against patents, because they have the potential of throwing a whole field in disarray, like they are doing now for computing.
I am not disputing that. I am saying that Santos-Dumont had a much better attitude by sharing what he did, and that served better the world than the Wright brothers secret, proprietary attitude.
October 1.906. But the Wright brothers only cared to show their airplanes in Europe in 1.908. By that time Santos-Dumont was already developing the Demoiselle, the first ultralight airplane.
Yes, because patents can, don't need to be licensed. It they were RAND, Reasonable And Non-Discriminatory, it would be much better. But the way it is, they actually managed to forbid other people from flying. That is umititaged evil, and contradicts the spirit of patent law.
No, but I think the current implementation of the concept is counterproductive. Even at the Wright brothers' time, it encouraged them to do things in secrecy, while Santos-Dumont worked in the open. There should be a provision that provided for patents to be granted only to work done publicly. The way it is, the patenting process is a powerful incentive to delay publication for years.
Also, patents should be always RAND. They should reward the inventor, not the manufacturer or distributor.
Software, processes and methods aren't in the original definition of patents, and have proved to be unmitigated evil. They should be stopped now, and all already granted revoked.
Acquisition price is hardly the whole story. First, it is heavily subsidized by the IA-32 business, so it is no indicative of production costs. Second, there is cost of integration, where size and complexity does matter, and cost of operation, where size, power consumption and heat generation does count.
You just proved my point that Intel is leveraging its monopoly to subsidise its inferior products. In foreign trade the US uses to high taxes and even forbid buying in these conditions -- that is called "dumping".
That proves you know little... the biggest con of VLIW, of which EPIC is a variant, is that the compiler optimisations used for a generation aren't valid for future generations. That is in contrast to RISC and even CISC, because when optimising for a generation of processors used to benefit future processors, it may actually worsen the performance of future processors in a VLIW family.
Additionally, usually the RISC system builders use better components, do better-balanced systems, and better quality control than the Intel me-toos. Particularly, while the Alpha has traditionally been better than anything else in floating-point, the POWER has excelled at integers. All in all, I wouldn't be surprised to learn that the POWER systems being benchmarked today endured more with less problems than their Itanium equivalents, and they should perform better in business loads. Scientifical loads would have been better hadn't Intel let the Alpha stagnate.
Granted HP's EPIC, of which Intel IPF IA-64 is but an implementation, tries to minimise that. But the price is even more complexity than the original RISCs, like the Alpha. So why bother with going EPIC instead of going RISC? Intel could have pushed the Alpha instead of flogging its own pet dead horse.
You missed the point. Even if perhaps more due to MS DOS and Windows history and to the stupidity of both IBM and Digital in their competitive offerings, the fact is that Intel has a monopoly in the CPU mass market. Its only opositor there is again the PowerPC, but then it is currently confined to the Macintosh and some Amiga and RISC GNU/Linux diehards. Intel has even bought the StrongARM to be able to extend its position into the PDA space, and has already succeeded there too. Now it is doing the same by buying the competition, namely the Alpha and HP EPIC, the first of which it has killed and the second it adopted. And it is using its resources gained in long dominance of the mass market to pracitise dumping of the high-end marke with the Itanium. That is unfair, if no illegal. In foreign trade, it is definetively illegal.
Yes, it is. Compare it with equivalent PowerPC and StrongARM. No matter what other economies you can get elsewhere, it is still a stupid waste of energy that is maintained purely for marchitectural reasons.
The point here is that Apple is tending towards use of IBM processors in its future high-end lineup, already using them in the low end now. That because Morotora has being paying more attention to the embedded market. This will realise economies of scale for other users of the POWER architecture as well.
Not so long ago an entire processor took less than 10W...
And his report was misinformed. The 14-bis had already ailerons. The difficulty of steering resided mostly in the fact that the pilot was standing, thus incapable of using his feet. The body contortioning was not weight shifting, but activating the control surfaces by strings attached to the pilots' clothes. Awkward, but given it was independent development, and since the Wright brothers were working in secrecy and eager to control airplanes everywhere, European courts were sensible enough to deny their claims.
False. They did they work in secrecy, so necessarily all the European designs, and much of the US ones, were parallel developments.
So much worse. That means that, had they succeeded in their patents, they would have been able to forestall aviation during 17 years both in US and Europe.
The point is that it was not "theirs". It was made of incremental improvements over other people's works preceding and in parallel.
Take the aileron, for one. Patented in the US by Bell in 1.911, invented in 1.909, ignored by the Wrights even if essential for the big airplanes they planned... and already present in the 1.906 Santos-Dumont 14-bis, who favored wing warping in the Demoiselle because this was such a small airplane. With the aileron, if Santos-Dumont had patented it, any Wright patents would have to be exchanged for the aileron patent in order to allow for practical big airplanes and thus generalised, since Santos-Dumont made a point of making his work available for everyone.
But all this was unnecessary, because the European courts saw thru the folly and did not accept the Wright patents.
Had their patent succeeded in Europe and stood in the US, it would have delayed the whole field. So much for patents fostering Science and the Useful Arts...
True enough. They save so much money, but kill their competitiveness against the Power4 and UltraSPARC IV, and at the same time much or all of their differentiation against other IPF IA-64 builders like Unisys, Dell, and even IBM itself.
There is little reason -- beyond inertia -- current Digital, Compaq and HP-UX users will not migrate to IBM, which has it all in house, or Sun, which focus better on RISC on Unix, or Dell, which focus better on Wintel. Thus they save a lot of cash, but loose lots of revenue. True enough most mergers go bad.
I probably forget others, but Unisys for one has built much bigger, meaner machines in the IA-32 space. Not to mention Penguin Computing... pity they could not sell enough of their 8-way systems. In order to sell this they should have centered on applications, perhaps with PostgreSQL and SAP, say, on Debian. But there is only so much one can do.
True enough. They save so much money, but kill their competitiveness against the Power4 and UltraSPARC IV, and at the same time much or all of their differentiation against other IPF IA-64 builders like Unisys, Dell, and even IBM itself.
I probably forget others, but Unisys for one has built much bigger, meaner machines in the IA-32 space. Not to mention Penguin Computing... pity they could not sell enough of their 8-way systems. In order to sell this they should have centered on applications, perhaps with PostgreSQL and SAP, say, on Debian. But there is only so much one can do.
Absolutely wrong. The high end was pretty well covered with wickedly fast, very reliable and well-balanced Digital Unix and OpenVMS systems. The problem is that there was no Sun Solaris for it, and this is basically about loyalty -- no one switches from SPARC Solaris, Power AIX or PA HP-UX just for performance. Ultimately the big failure was the failure of Digital to produce as reassuring a corporate roadmap as Sun, IBM or HP.
True enough, but not the whole truth. On one side they were stiffled by Microsoft never porting even the basics of its software porforlio. There was no MS Access, Visual Basic came too late, and code was not nearly as well tuned as for IA-32. The result was that, even if the systems themselves were competitively priced with the IA-32 ones, Microsoft Windows itself limited the performance and reliability advantage the hardware granted.
They did supported GNU/Linux, but at a time when it was strictly geek-only. And even so, they never had OEMs enough to build a strong platform. Also, lack of the seemingly infinite Intel-like scale of resouces left some important holes in their lineup, such as notebook and low-end processors.
Compaq could have pulled it with its resources, but its box-shifting corporate mentality eventually killed the Digital-inherited engineering heritage, and the whole thing went into suicide mode by pushing harder the lower-margin IA-32 side of business. Just as depressing as Digital shooting itself in the foot by calling Unix snake oil and failing to capitalise in Novell and Apple interest in using the Alpha.
So the issue here is price... but Itanium is still hotter and bigger than Power4. The comparision to Alpha I dismiss out of hand since the Alpha processors are neglected by Intel itself. But if the Power4 has equivalent performance being smaller, cooler, and more compatible in its own processor family, then we have again an issue of Intel leveraging its monopoly to push an inferior product, as Microsoft does with its Windows.
Now the Power4 is poised to win more economies of scale with the G5 or G6 generation going into Apple Macintoshes, AmigaOnes and POP GNU/Linux systems, besides the AIX ones. This will be interesting to watch.
It strikes me as absurd that all this global warming talk says nothing of the absurd power consumption by IPF & IA-32 processors. A concerted effort for a RISC migration, and for properly configuring aggressive power saving in all corporate and domestic systems, would go a long way.
Currently, that is GNU GPL v2, you are right. But the FSF has already endorsed the Apero FSL as a draft for this change in GNU GPL v3. This is to keep Internet ASPs honest.
It means it was in Europe that aviation popularised.
Yes, just as Santos-Dumont nos. 19 to 22, the Demoiselles.
Just as Santos-Dumont did. But when he did, he did not kept it to himself.
And kept the whole aviation field in the USA in a check, having the power to forbid people from using their patents. Which should never been granted in the first place, because all they did was build upon other peoples work in parallel with people from other countries. The whole situation parallels the incredible injustice of the patent granted on the telephone by a few moments difference in the filing time.
That while Santos-Dumont produced the Demoiselle and gave away both the concept and the projects for everyone to use, thus precluding Wright brothers attempts at hindering the progress of aviation in Europe too.
Wrong, it does slow things a lot, ceteris paribus. As ceteris are not paribus, Intel and AMD just throw a lot of resources to get CISC chips as fast as RISC ones. In order to do this they make big, expensive-to-project-and-manufacture, hot, energy-hogs of some chips.
Just when mainframes are not water-cooled anymore, and there are RISC processors that achieve even better-than-mainframe performance, if you wanted to get mainframe performance from IPF (IA-32 & IA-68, or x86 & EPIC) processors you would need water cooling, or nearly so.
Yes, but what are the trade-offs? The IA-32-to -EPIC translation layer still makes for a too-big processor, and even without it lots of efficiency are lost in attempts to ease off the burden on the compiler and the lack of consistency among chip generations. Alpha in particular, and RISC in general, are still much more efficient and elegant.
Could not agree more. Moderators, please, Insightful +1 for the parent until he gets at least four.
I would add that the basic issue is free market without neither ethics to strive for the better, nor education to be able to choose and attain it. In absence of ethics and education, justice and schooling would do as a second best, but even this is being lost.
But they are not competitive. They only beat their RISC processor by the, ironically, Chinese Army strategy: throwing an absurd level of resources, creating an absurdly big, hot chip and producing it at nicely big scales.
If Intel would suddenly change heart and put in place a migration path to, say, the Alpha with MS WXP, Debian GNU/Linux running Gnome2, DEC OSF/1 Unix and the likes of them -- or if IBM & Apple got a clue to start selling nice cheap PowerPC microsystems that could run AIX, Mac OS X, preconfigured Debian Gnome2 GNU/Linux, Amiga and the likes of them --, thus getting the same economies of scale as the x86 processors enjoy, then we would see in the market what is already true in the fabs and technical specifications: RISC processors are cheaper to manufacture, smaller, more efficient.
This is stupid. The bigger RISC cache is more than compensated by totally avoiding needing the CISC-to-RISC translation, which takes much more chip real state then the cache, adds much more complexity to the project, and in addition slows down the whole kabooza, by adding both processing stages and development delays.
Very interesting, thanks for the info.
But the fact remains that an equivalent RISC processor would be smaller, cheaper to manufacture ceteris paribus -- that is, given equivalent economies of scale --, use up less energy and generate less heat, thus requiring either a smaller heatsink and therefore enclosure, or a even slower, smaller fan that turns on less often, or even none fan at all as in the Apple Cube.
Not by any means. In Communist theory, they have tried the Proletariat Dictatorship, which is called this side of the Bamboo Curtain by the less hopeful name of Totalitarian State Capitalism. Communism would be a further stage after education has eradicated egotism, and then the production means would be owned by the communities -- hence Communism.
What did change was that the Party has decided that the Proletariat Dictatorship has failed in almost all goals except keeping a small clique in power, and the failure was risking even the inner ring. So they are keeping both the Communist Newspeak and the totalitarian government, but changing the production mode into a form of Mercantilism, where the government cooperates with a few chosen business to advance what is officially national, or popular, interest.
Come to think of that, even the US is becoming Mercantilist with all the ever-crescent restrictions to immigration and trade.
This being not a for-profit fly-by-night sweatshop, but a research institute, rumour has it that they cloned Alpha.
I hope they did, because there is no microprocessor architecture that holds more promise then the Alpha, and it is a shame on the US supposedly pro-competitive, efficient culture that it has been cancelled due to Digital being inefficient in marketing it and then Intel not wanting the competition.
Oh, yes. That phrase. Tip: it was intended to be ironic, because not only the poster I responded to defended the Wright brothers, he also dismissed Alberto Santos-Dumont works with derisive words. So I was questioning his exclusiveness to the point of debasing his opponents.
I do not think it was pointless. Have you noticed that in defending the Wright brothers all these people from USA not only overlook the moral and scientific implications of secrecy and patents, they also tried to deny all of other peoples contributions? Just to name one, present in 14-bis 1.906 flights, the aileron.
In another quote I said that, just like with the automobile, there was not one inventor to the airplane, but several. Now, trying to make either Mr Curtiss or the Wright brothers the sole inventor(s) is ridiculous and shows how much the USA as a whole is ignorant of the rest of the World and History.
Nothing could be further from the truth. 14-bis gradually improved, and the Demoiselle was used practically for individual aerial transportation with full steering over several kilometers, being the first widely-used (for the time) light aircraft.