One Intel fan never stays quiet

And who are you?

The great protector of Intel's customers?

Or a Brit Rag that hasn't been leaked the real story on Intel's new sockets?

By the by - where was your outrage when you reported that AMD was going to change the AthyWiper64 socket from 754 pins to 940 pins?

Why didn't you editorialize the great pain that would cause AMD's customers - you know - the ones that had motherboards ready LAST YEAR with 754 pin sockets?

Paul R. Engel

Hi Mike

This really shouldn't surprise anyone, considering Intel's past record. Me.

I'm a cheap bastard and always try to get the biggest bang for my hard earned moolah. I guess that's why I go with "Chimpzilla". Good upgrade path, chips cost half as much ( even at the sweet spot") and boards are 1/3 less.

All this for rather comparable performance. Hey, when your pushing around 100 fps, what's a few frames plus or minus? And all this gigahertz don't mean a damn bit of squat in any MS Office apps. Thanx for the heads-up.

Joachim S

Mike,

this is IMO just typical of intel.

I would even go as far as to suggest that this ( see title ) is the main reason why big OEMs go for intel chips. After all - what could be worse for HP, Dell, IBM & Gateway if the customer could just upgrade with a new CPU ? .. now that would hurt future sales wouldn't it !!

What really amazes me is that IT departments ( who are supposed to be knowledgable about IT ) still favour intel over AMD. As a result, I have a healthy disrespect for these so called IT departments - who basically know something about applications programming and system administration - but that's about it I'm afraid.

My PC dealer claims that he's been totally confused about intel's chipsets since the BX chipset - which he claims was the last really reliable intel chipset.

Talking about IT departments, I'm afraid the old rule still applies to those unwilling to take on responsibility or risk -: nobody ever got sacked for buying eg: IBM, Microsoft or intel.

I've also noticed that the mainstream news is "quiet" and AMD's excellent product palette - especially the Opteron. Is Berlusconi not alone I ask ?

Yours truly

Phil G

Thorton's no black hole
AMD pricing itself into a black hole with the Thorton

Hi Mike,

Just thought I'd write in and comment on the story concerning AMD pricing themselves into a black hole... One thing that you didn't consider is this...

What if AMD EOL'd all of the Athlon XP's from the 2500+ down and only continued manufacturing from the 2600+ to the 3200+... This would allow them to use chips that were going to be trashed to sell into the value end of the market. In my opinion, if they do this, then, any chips that get sold into this market become 100% profit as one could consider the cost to be Zero (when compared to the past practice of throwing the bad chips in the barrel).

Then, as the Athlon 64 takes hold, they phase out the Athlon XP and lose the extra 256K cache (thus creating a chip thats less expensive to manufacture) and continue the Athlon FX as the budget processor and NOT the Athlon XP as most now expect... This serves two purposes. It makes the Athlon cheaper to build, and, it makes the Athlon 64 appear much faster and more desireable. (It always seems to work for Intel...) AND, the Athlon FX would still compete reasonably well against the Celeron...

JMHO

Scott Hylton

I seem to recall back in the Coppermine Intel CPU days, the by-then-long-socketed Pentium III and Celeron processors (>= 533 MHz methinks) came from the same die.

Prior to Celeron processors jumping ship to the new die, they were pin-incompatible with the Pentium III. When they switched over, half the P3's 512k cache was disabled, as were SMP pins, but everything else was the same. I'm sure my fellow unfortunate Abit BP6 owners remember this well.

Because the on-die cache space takes up quite a bit of the CPU die, it is possible that a manufacturing flaw in one area of cache could be a recoverable flaw. Simply disable half the cache mechanically or programmatically, and the other non-flawed half will work just fine. Insta-Celeron. This raised Intel's manufacturing yield, thus enabling a lower average selling price per processor to turn a profit.

I suspect that AMD's Thorton is a similarly-recovered chip from the Barton scrap heap, or a slightly die-modified Barton to enable this recovery from certain process errors. Thus, any amount of money they can get from the chip would not necessarily be without profit margin. The way I see it, it's mostly profit.

AMD is trying to recover money from nothing. Thus, Thorton pricing is more apropos to Hawking radiation from a black hole, not the black hole itself.

Does this mean one man's trash is another man's Athlon FX? I leave that one as an exercise for your readers.

Steve Lewis

We're being shafted
Has Dell's Axim got a big Microsoft problemo?

Yes it bloody well has!

The WM2003 units they're putting out have serious performance issues and all that Dell is doing about it is sticking their fingers in their ears and going "Lalalalalalala I can't heeeeeeear you!"

I actually braved Dell tech support today only to be told that there is no problem, they've heard of no problems. .

Oh well, HP just got themselves another customer - they may be blatantly ripping us off in the UK pricewise (2x US price), but at least the thing actually works right.

"Dude, you're getting a shafting" as the yank ads would say.

Ray

Wires major obstacle to smaller chips
Semiconductor shrink to 90 nanometers a giant leap for chipkind

Hello, Mike

It seems there is a slight misunderstanding about the nature of chipmaking in the public. The transistors shrink, that's true, and the gates shrink too, so the latch time decreases (and speed increases), but the wires are still same old copper (better than aluminium, but anyway).

And since there is no real leaps in wire tech, and with each new process shrink, the chip dimensions decrease, and thus the wires resistance increases (since their diameters shrink) while wire capacitances are almost constant between processes. This poses a big problem.

Since now its possible for the signal to get anywhere across the chi in one cycle, it may not be possible in future processes. And, for that you'd need deep pipelines in new processes. P4 already has this for the future, as it has 2 pipeline stages added just for moving signals through the chip.

Basically, each new process brings us closer to beng "communications-bound" from being "capacity-bound", and while we move to new techs, there comes the need for complex tricks implemented for the cpus to work. Pipelining is going to play a MAJOR role in coming new processes, as we're increasing frequencies. And this also increases complexity of the design process. Thus, moving to 90nm is NOT an easy project to acomplish, and can take a substantial amount of time.

Also, its pretty much possible that its much more feasible to design a completely new processor from ground up to use the new tech process, instead of redesigning an existing design, since the latter can require more effort.

This is a known fact, and there exist several complex techniques to accomplish move to tighter processes, but since their complexity rises while processes shrink, we might not see an AMD Barton scale to 90nm, as its design is not well-suited for that. A more-or-less complete redesign is needed, and for what its worth, it may be better to focus on 64bit processor now, which already has provisions for frequency increases.

Name supplied

Howdy,

You stated:

While there will always be a certain proportion of buyers who want the latest and the greatest, many aren't convinced that an extra bit of megahertz here or there will really make their "computing experience" that much better. Right now, the industry is pinning its hopes on Microsoft's Longhorn and PCI Express to give people - and we mean the mainstream, not the enthusiast - a really good reason to buy.

LOL, the "mainstream" people could careless, heck i would even be surprised if they knew what all that meant. If you asked people what microsoft longhorn was, i bet a good number of people would think microsoft bought a cattle ranch or something. The only people that ARE gonna care are those "enthusiasts", people like me and you. Why would mom and pop care to upgrade? To them it is all just an expensive hassle. Learning a new OS moving everything over...For what? There is NO real compelling reason for the masses to move off of their existing computers. They are more than capable of doing the homework or the internet uses that they do now. I mean, even a 1GHZ machine is MORE than enough for most people.

The "industry" was pinning its hopes on XP, that it would be a reason for new growth. They are just under the same delusion that NEW = BETTER. Change for the sake of change is something most people are not inclined to do, only us fools in the computer industry upgrade our machines for the sake of upgrading. With no REAL LIFE need. The computer industry is going to have to face facts, the industrialized world is saturated with computers now, like what has happened to TVs, VCRs etc...

You do not see Maytag expecting explosive growth in the washing machine world do you? No! The computer industry is just going to have to get over the days of double digit growths and such.

In all honesty there has been no REAL new change in the computer world in the last few years. Yes faster this or larger that. But once they hit the GHz barrier, there has been nothing overly new about them to require a change. Like when color tv first came out, it was a must have to replace that old black and white. That is the kind of thing that is needed in the computer world.

Now, say we finally got a computer system that actually could operate purely via voice input, kinda like you see on star trek. Now that would be the kinda thing that would make everyone want a new system. That would be the spark that would drive the masses back to the store to get a new one.

Not another MS OS or some new bus type

Shawn D Silvas