I'm in a nitpicker's mood, so here goes.
Joerg and Xed wrote:
_____________________________________
" . . . If anything AMD is far from being done with its single core processors and i dont doubt intel will get to 4.5 or so before throwing in the towel with prescott. . . . "
"Erm, Intel already cancelled the 4 gig prescott =) "
_____________________________________
COMMENT: Before the 3.46EE vs. Athlon 64 FX-55 shootout, industry publications noted that both AMD and Intel were prototyping dual-core processors, to address simultaneously the issues of heat and speed. It is likely that the 3.8E (or is it EE?) will be the last in the Prescott line.
_____________________________________
berkut7 and wchou wrote:
"They have finally figured out that it's better to be profitable than to have large market share. "
"Oh it'll get much higher if Intel falls over on its feet. not to mention duron and xp as well, chuckle. "
_____________________________________
COMMENT:
First, there are several "markets". There is an "office PC" market, a "student's" market, a "home-PC" market, an industrial PC market, and a fruitcake-geek-over-clocking-obsessive-compulsive market. To address multiple markets, companies provide "product differentiation." Not only do they produce different processors at different manufacturing costs and different prices to capture shares of different markets, they try to piggy-back differentiated products onto the same manufacturing process -- thus the Gallatin core resurfacing as a Northwood production run with a disabled L3 cache.
But in the long run, it is he who captures the largest share of loyal customer base who survives. Thus, survivability trumps "profitability". If product demand is elastic, then the profits are to be had through increasing Q, quantity produced and quantity sold at the most competitive price to the largest number of customers. If product demand is inelastic, profits are to be had through restrained production and prices chosen to maximize the returns from one-time buyers.
These companies are not monopolies -- they are just dominant firms. So the "play-offs" continue. If the Athlon 64 FX-55 could find a big demand among the office-pc or home-pc market, prices would fall to some point just above average cost that would reflect recoupment of the average fixed cost of R & D and the variable cost of manufacture. But the FX-55 is not soon going to find a home among markets with this many computers or computer users. Given the number of "buyers" with a "need" for the FX-55, the company won't recoup average fixed cost and its manufacturing costs by charging Celeron prices.
Same with the Intel EE CPUs.
One more point. L2 and L3 cache memory (it IS memory, after all) -- has to use the fastest memory available, otherwise it is no good as cache memory. So when you go from 512K to 1 MB to 2 MB of cache, you increase the cost of the CPU dramatically, and further, given the number of transisters required to add that cache, manufacturing costs increase -- which is to say about the same thing.
All of these factors determine price for Intel EE and AMD-64-bit processors. With only two MAJOR competitors in the marketplace, prices are not going to be as low as with five or ten or fifteen competitors. But even so, with only two competitors in the marketplace, prices will still reflect the cost of the technology and the size of the market that demands it.
And there will be stunning developments around the corner. Too bad CPUs are not like vintage wine. They don't become more valuable with age, and a fluke of some reseller going bankrupt is not likely to result in an auction where $15 bottles ($1,000 processors) find their way to customers' hands for 99 cents ($250).