Author: Jeremiah Penery
Date: 22:35:33 02/21/03
Go up one level in this thread
On February 22, 2003 at 00:37:35, Robert Hyatt wrote: >On February 21, 2003 at 10:19:30, Jeremiah Penery wrote: > >>On February 21, 2003 at 09:45:28, Robert Hyatt wrote: >> >>>On February 21, 2003 at 01:52:34, Jeremiah Penery wrote: >>> >>>>If you can sell a million 3GHz processors today for $600, and another million >>>>3.2GHz processors next month for $600 (plus n 3GHz ones for $400 now), and then >>>>a month later you sell a million 3.4GHz processors for $600, etc., why in the >>>>world would you want to sell 1.2 million 4GHz processors today at $600, and lose >>>>out on ALL those intervening speed grades? Do the math, and it just doesn't add >>>>up. The chip companies make a lot of money by trickling out clock speed >>>>advances, because they can charge more for the highest clocked parts. Releasing >>>>something 2x faster today than what has previously been released is NOT going to >>>>net them more money in the long run. Basically, it amounts to bleeding the >>>>customers for all they're worth - and it works. >>> >>> >>>I don't buy that. Because if I can widen the game between me and my >>>competitor significantly, I am going to get a bigger share of the market, >>>so producing a chip that is stamped slower than what it can actually run at >>>doesn't make sense on the top-end of the market. Yes, I'd take my 3.2ghz >>>line and siphon off some and mark them 3.06 and 2.8 and so forth, if there >>>is a demand down there. But I'm not going to hold back my top-end chips >>>as the more separation there is between me and my competitor, performance >>>wise, the larger my market share. Otherwise there would be no SPEC, no >>>THWP, etc... >> >>Intel already has 85%+ of the desktop market share, and probably even a greater >>portion of the small server share. Killing the competition to gain that extra >>10-15% market share does not net them more money in the short term (the math is >>not complicated), and invites serious problems in the long run, such as possible >>anti-trust investigations if AMD goes out of business. > >If Intel could take that last 15% it would be a _huge_ gain. Because then >they could set the prices as they wanted. > >There is no anti-trust problems with making a chip so much faster than your >opposition that they can not sell their product... There is no anti-trust problems with making an operating system so much more compatible and simpler to use than your opposition that they can not sell their product... Oh, wait, tell that to Microsoft. :) If AMD goes out of business, and Intel uses its new-found 99%+ market share to 'set prices as they want', that would certainly qualify them as a monopoly. >>>>Intel raises clock speed just enough to stay ahead of their perceived >>>>competition. If AMD magically released a 4GHz part tomorrow, do you seriously >>>>doubt that Intel wouldn't be able to follow suit almost immediately? >>> >>>Yes I do. Because they would have already released it to get the lions share >>>of the top-end market where the profit is highest. >> >>Intel _already_ has the "lions share of the top-end market". Even more so than >>the lion's share of the desktop market they have. > >Actually they don't. The "top end" market is dominated by other vendors We were talking about Xeons (I have no delusions about Intel's ability to clock the Itanium monster to anywhere near 4GHz soon), so by "top end market" I meant the 1-4 processor workstation market, dominated currently by Xeons. The market you're referring to below is where Itanium goes. >such as MIPS, IBM and the like. It isn't a big market, but it is hugely >profitable. You don't see intel chips in "mainframe-class" machines, but >you definitely see MIPS and PPC4's and HP-PAs... and even alphas although >their status is no longer clear... Itanium is supposed to go into the mainframe-class machines. Alpha/HP-PA are being killed by HP, in favor of Itanium. MIPS isn't a factor in the high-end anymore - they're used more for embedded devices and such. IBM's POWER series will probably compete well with Itanium for quite some time. SPARC, I guess, will continue to sell as it always has, for whatever reason. The only wild-card is Opteron. Several major vendors are said to be lined up to sell it (IBM, SUN, etc.), but it's not clear whether it will get much market penetration.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.