On Sat, Nov 24, 2001 at 12:08:29PM -0600, Carl Wilhelm Soderstrom wrote: > here's something that's been nagging at me; and maybe those who know more > about SGI can answer this. <SNIP> > I think the world will be poorer, for the lack of competition in processor > designs... Intel has a pretty big stake in the status quo; the shakeup > involved in going to the IA64 architecture has lost them ground to AMD. > without Alpha, MIPS, and PA-RISC processor teams working away; will we see > less innovation in the processor world? there'll certainly continue to be > research at universities, and independent groups; but with Intel holding an > even larger monopoly on processor markets, there may be less chance of new > processor designs making it to market. Carl, you're perceptive to pick up on this thread, but it's worse than you think. If you go search google in like comp.os.vms, comp.sys.dec, and some of the other ones, you'll find much enlightening discussion buried undere lots of FUD, dating back to the June 25th announcement to kill Alpha. Not only is it greatly reducing the competitive market, but it's throwing the baby out with the bath water. First off, they are 10 years behind the curve. MIPS has been 64-bit since '92. HP-RISC goes back about that far. SPARC is 5-10 years into it. What else? IBM's had 64-bit for 5-10 years. Alpha has been going 64-bit since really about '89 or so. Secondly, there exists on Compaq's website (don't have the URL, but it has been widely popularized of late and should be fairly easy to turn up there or on someone's site) an engineering report detailing how and why the Itanic cannot match, much less exceed, the performance of the EV8 Alpha which was killed in dev. One rumor was that Intel bought Alpha and the Alpha design teams because they can't make Itanic competitive without help. What better way to compete than to hire the competition and buy their competing product? I can't say that there are any teeth to this rumor, but it has never been dispelled by anyone. If you search usenet, look for posts from (Brandon?) Batson, an Alpha engineer that let some interesting details slip. > I think Linux, and other Open Source software is the tool with which to > break processor architecture monopolies. As much as I would like to agree, I don't think I can. What open source will do is allow us oddball geeks -- and I should know: I have *lots* of non-Intel hardware in the basement! -- to keep using things that are no longer considered commercially viable by the manufacturer. But new sales are driven by applications, and I can't think of a real-world application that can't be run on Intel. So the marketing pukes tout "industry standardization" and the developers and users have no motivation to be different. So it goes. > with Linux, if someone comes out with a new processor design (take a > look at http://www.f-cpu.de/), and convinces hardware makers to build > compatible hardware; it's possible to just port the OS and compiler, > recompile your Linux apps, and it's as easy to run on the new processor as > the old one. OK, but the thing is it takes millions of dollars to "come out with a new processor design." Even if the design work could be done like a "free" system, someone has to spend a lot of money on fab. The thing about Alpha and now Itanic is that the fab engineering (building smaller faster silicon) is at least, maybe slightly more, than half of the battle. Getting people to build boxes around a chip is nowhere near as tough -- I think that could be done. But big money has big inertia. HP killed off PA-RISC to deal with the Compaq merger, and Alpha's gone too. > so I guess I'm saying that even if Intel takes over even more of the > processor market; all hope is not lost, it's just going to take us a while > to break their stranglehold. "While there's life, there's hope." The fact that you can walk over to MicroCenter and play with a NEC sub-laptop running on a Transmeta Crusoe is one ray of hope. (I didn't notice if it was running Linux, though...) -- I used to like HP before computers, and once I even liked Compaq, but I liked DEC better than HP and Compaq put together.