July 3, 2001

11:54 PM

Dell takes a survey out on AMD processors.Finally, the only thing that keeps me from buying a Dell is their lack of Athlon and Duron machines. Sounds like the negative reviews of the P4 has Dell running towards the competition.
posted by skallas (21 comments total)

"It's just market research and anticipating future demand," the representative said. It's "a chance to touch customers and get some feedback from them about what they're looking for...but it's not indicative of any certain move.

The willpower to avoid saying "to touch base with the customers" must have been staggering.
posted by skallas at 12:10 AM on July 4, 2001

I'm sold on AMD. The MicronPC I bought two months ago uses a 1.4GHz Athlon, and it f'ing SCREAMS. (Of course, i'm sure a 1.4GHz P3 screams too, but I like promoting competition.)
posted by jpoulos at 10:46 AM on July 4, 2001

There is no 1.4 GHz PIII. They can't make them that fast.

The 1.4 GHz P4, on the other hand, is a dog. We're talking real bow-wow. (Relatively speaking, of course. It's not that it's achingly slow, but rather that it doesn't perform as you would expect based on its nominal clock rate and cost in bucks.)
posted by Steven Den Beste at 10:55 AM on July 4, 2001

Assuming the same processor speed, can anyone explain to me what difference I will notice in everyday use of a PC with an Athlon against a Pentium? Is there any?
posted by normy at 10:55 AM on July 4, 2001

Assuming the same processor speed, you will notice a significantly higher remaining balance in your checking account buying an athlon. AMD has the price/performance metric all tied up.
posted by boaz at 11:20 AM on July 4, 2001

Normy: some benchmarks:

Tech Report
AnandTech
Tom's Hardware Guide
Ace's Hardware
posted by Steven Den Beste at 11:50 AM on July 4, 2001

There's too much hate for Intel in the market. Even though I'm a devoted AMD-fan, I do respect Intel's P4 processor. Good stuff. Has great potential. Although, I feel they made a mistake sticking with Rambus again.
posted by dequinix at 6:16 PM on July 4, 2001

Some extremely negative but relatively convincing technical arguments as to why the P4 is double-plus-ungood here. This was the article that sent me away from Intel...
posted by stavrosthewonderchicken at 10:33 PM on July 4, 2001

A lot of the concessions which Intel had to make on the P4 were due to the fact that they were using a fairly large process (not even 180 nm) for it, and as a result, the dies were immense. Not only does this reduce the number of dies per wafer, but it increases the chance that a given die will be bad (because the flaw rate is a function of surface area). Things like the cut-down L1 cache and the reduced floating point unit were done to decrease the die-size, which was still immense when they were done.

The next rev P4 ("Northwood") will be made with a much smaller process, so they should be able to add back the circuitry they stripped out of the first version. On the other hand, the hugely long pipeline is not changed and that means they still pay a big penalty for a branch mispredict. And by the time Northwood is out, AMD will be humping with the Palomino, which is proving to be a vast improvement, cycle-for-cycle, over the Thunderbird which is already causing Intel so much grief.

The next year is not going to be a good time for Intel.
posted by Steven Den Beste at 11:15 PM on July 4, 2001

> ... by the time Northwood is out, AMD will be humping
> with the Palomino ...

(Steven, baby, you've got to get out more often. I'm not arguing with your numbers. I'm just saying you've got to get out more often. Humping with the Palomino?)

Anyway, AMD vs. Intel? Much of the pro-AMD talk is just anti-Intel sentiment from the same people who will never buy a Microsoft product because Microsoft is a very big corporation they have decided to hate forever. If AMD chips were slower and more expensive, many people would still look for reasons to buy them because the alternative is, again, a giant corporation that they have decided to hate forever. If AMD (or Cyrix or whoever) becomes the big guy, they will hate that company.
posted by pracowity at 3:31 AM on July 5, 2001

A lot of the concessions which Intel had to make on the P4 were due to the fact that they were using a fairly large process...

While i'm not exactly sure what all that means, Steven, what has AMD done differently?

If AMD chips were slower and more expensive, many people would still look for reasons to buy them because the alternative is, again, a giant corporation that they have decided to hate forever.

pracowity, that's speculation based on...what? While I admit there is plenty of blind hatred of Microsoft out there (how did we start talking about microsoft again?), I don't see any of that here. The fact that the Athlon is cheaper alone is valid enough reason to prefer it to the P4. This thread has produced several concrete reasons why the Athlon has an edge over the Pentium. How can you dismiss as baseless bitching?
posted by jpoulos at 7:08 AM on July 5, 2001

> pracowity, that's speculation based on...what?

Based on human nature. The little guy usually likes the underdog and reacts against the big guy. Look at "intel sucks" and "amd sucks" in google searches. Look at message boards. Look at your own comment above about how you were "sold on AMD" not because you were in a position to compare Intel and AMD chips, but because you are happy with your AMD chip and you "like promoting competition" with the big guy.

I didn't dismiss anything as "baseless bitching," though. Prices are prices and I'm not arguing with them. And I don't have a favorite chip or chipmaker. If you love AMD, that's fine with me.

I would, however, like to see people discuss prices that matter more to the average buyer, not just, say, to the impecunious teenage hobbiest who assembles his computers in his bedroom from the cheapest parts he can find. Hot-rodders have always been insane about "squeezing performance" out of the things they build at night, whether they were building cars or computers, and numbers always mattered most to them. They want their machines to race past other machines; they aren't necessarily concerned with normal speeds needed for normal life.

How much does a full computer system running an Intel chip cost if you walk into the store and buy it? How much does it cost to buy the same system at the same store but with an AMD chip? How fast do the buyer's e-mail and word processing and browser applications run with an Intel chip compared to an AMD chip, and do most buyers (not just the gamers and hobbyists) notice the difference and feel the cost?
posted by pracowity at 9:23 AM on July 5, 2001

Pracowity: "How much does a full computer system running an Intel chip cost if you walk into the store and buy it?" How long is a piece of rope?

What computer system? How powerful? Configured how? The answer to your question is a price somewhere between $1000 and $6000.

"How much does it cost to buy the same system at the same store but with an AMD chip?" You can't; without the P4 it isn't the same system.

I think I know the question you think you are trying to ask, and I'll try to answer it. For two systems based on P4 and Athlon which have about the same speed-of-execution of software, the Athlon system will save you about $300, more or less (depending on the configuration). Six months ago the price differential was more like $600, but Intel is selling its chips at a loss now to remain competitive. The biggest source of the price differential is the RAM, because RDRAM is preposterously expensive compared to SDRAM or DDR-SDRAM, so that savings will depend critically on the amount of memory in the system.
posted by Steven Den Beste at 10:55 AM on July 5, 2001

Look at "intel sucks" and "amd sucks" in google searches.

Actually, if you add the binding quotes, this "evidence" ( questionable measure of anything, really) is pretty watered down:

"amd sucks" and "intel sucks"
posted by fooljay at 4:41 PM on July 5, 2001

Oh, I forgot to add my question. Is there any type of software which just won't run on the alternate chips which will run on an Intel chip?
posted by fooljay at 4:42 PM on July 5, 2001

It's reported that some of the latest Cyrix processors (from Via) won't run some commercial software. However, the Athlon and Duron have no problems I've ever heard of.
posted by Steven Den Beste at 6:51 PM on July 5, 2001

> For two systems based on P4 and Athlon which have
> about the same speed-of-execution of software...

So we're really talking about maybe a $300 difference on a purchase of "somewhere between $1000 and $6000." Probably closer to $6000 or the speed wouldn't much matter. And $300 is, what, just 5 percent of $6000? The real price difference is a lot less than you might think, listening to people go on about it.

And market pressures (including price pressure from AMD, of course) are forcing Intel to lower the cost of a system even further by allowing the use of cheaper RDRAM, the cost of which is the "biggest source of the price differential" between the two choices, right?
posted by pracowity at 3:00 AM on July 6, 2001

Probably closer to $6000 or the speed wouldn't much matter

It's closer to $3000. That's what the top of the line Dell and Gateway systems go for these days. To spend $6,000, you've got to try pretty hard.
posted by jpoulos at 6:24 AM on July 6, 2001

There are a number of issues here.

For a while, Intel had no processor which was actually at execution parity with the fastest AMD systems. For instance. right at the moment the 1.4 GHz Thunderbird and the 1.8 GHz P4 are closely matched. But AMD is about to pull into the lead again with the release of the Palomino for desktop.

Second, AMD's fundamental production costs are lower, and Intel is having to sell its processors at a ruinously low price just to remain competitive. AMD just announced a warning but will still make money. Intel's been struggling badly and it is bleeding cash at an unbelievable rate. It won't be able to sustain these prices for much longer; it's going to have to hike them somehow. Probably it will do that by eliminating its RDRAM price subsidy. That has been its problem for a long time: there isn't any "cheaper RDRAM".

And $6000 is very, very unusual for a desktop computer. ($6000 is a decent multiprocessor server.) Right now the typical price is in the ballpark of $1500, and a $300 difference is quite significant at that price. (A really expensive desktop computer runs $3000, but only hardcore gamers and similar fanatics buy those.)
posted by Steven Den Beste at 6:35 AM on July 6, 2001

> That has been its problem for a long time: there isn't
> any "cheaper RDRAM".

What's this, then?
posted by pracowity at 6:55 AM on July 6, 2001

"This" is Rambus's attempt to drop the price premium of RDRAM from 300% to more like 100% compared to DDR-SDRAM. It's still too expensive. (It's also not on the market yet.)

I suppose technically that it qualifies as "cheaper" by a strict interpretation of the word. It does not, however, qualify as "competitively priced".

And by the way, that article is an interesting historical note, because it dates from 2/27/2001 and lists Rambus's stock price as $42.75. It's now under $11.
posted by Steven Den Beste at 8:05 AM on July 6, 2001