AMD and IBM Working Together on Future Chips

oogbla writes "There is a story over at news.com which says that AMD is teaming up with IBM for its sub-100 nanometer process and is de-emphasizing its previous relationships in that area. Also seems that the Silicon-on-Insulator (SOI) technology they were supposedly getting from Motorola isn't going too well and has caused at least one delay to Barton."

Comments

good news/bad news?

[herrd0kt0r]

to me, the most interesting part of the article is this: "The deal also marks a break in an alliance between AMD and Taiwan's United Microelectronics, a foundry that makes chips for other companies Earlier, AMD and UMC agreed to jointly develop 65 nanometer processes and build a fabrication facility together. AMD also said it would use UMC for excess factory capacity if necessary."

at first glance, it might seem like it's bad that AMD is breaking business ties. but the last sentence indicates that the option to tap UMC is still available, which to me means that the relations between the two must not have soured that badly.

looking at the big picture, it seems that AMD has made a pretty decent business move upwards, scaling up as they need, and acquiring a nice, big name to throw around as good PR.

which is not to say that it'll make AMD successful. but you know how dippy people are when it comes to stocks. Joe Trader who was "like, wtf?! UMC? wtf UMC?" might be more like "miammiam, IBM. mmmn, juicy goodness."

Is all this change good for the company?

[SirTwitchALot]

AMD seems to be making a lot of different decisions lately. I wonder if this apparent inability to decide if AMD is in the market or out, if they're going to pander to consumers or the embedded market, whether they're going to use cogswell cogs or spacely sprockets, is going to hinder adoption of Hammer? Especially since Intel (at least tries to give the appearance) that Itanium is a planned technology with a solid roadmap and development plan.

AMD vs Intel

[Elbereth]

It's interesting how all the hardware sites lambasted Intel for the design of the Pentium 4, because it didn't have the raw speed of the AMD Athlon or Pentium 3. What made this all rather amusing to me, however, was that these people weren't around to see the evolution of the 386 vs 486, 486 vs Pentium, and, to a lesser extent, the 286 vs 386. In each of these situations, the previous generation of chip was able to eek a few more cycles over the next generation... in the beginning of the next generation's run! Intel has a very strong history of designing chips that ramp up very well (except for their one CPU engineering failure, the Pentium Pro, which was too ambitiously designed).

I wasn't surprised when the AMD Athlon pulled out ahead of the Pentium 4, then fell very far behind. The Athlon was not engineered to ramp up well over 1 GHz. AMD was very foolish to race to that point, seeing as how long it took them to get working silicon at just 2 GHz.

I'm not saying that I bought a Pentium 4, just that I knew it would eventually overtake the Athlon. I'm quite happy with a cheap Athlon, myself. Semiconductors is a soap opera for nerds. That's why I read The Register, not EE Times.

My guess is that there's going to be a lot more consolodation in the semiconductor and memory world. I bet Micron, AMD, Motorola, and Apple are all going to end up merging, buying out, and/or disappearing in the next few months. Maybe HP will buy them all.

Re:AMD vs Intel

[sean23007]

I hope you misspoke when you said Apple would merge or disappear. Apple isn't going anywhere, as can be seen by their repeated strong showings at their release party conferences and on Wall Street. Apple is a strong company that is in no danger of being purchased. The only thing I can foresee that resembles what you are talking about is if Apple were to buy out Motorola or Micron and so be able to manufacture their own processors and/or memory units. But the next revision of the PowerPC processor went to IBM, not Motorola, and IBM and Apple will not consolidate into one company. Perhaps if Apple bought Motorola (almost impossible), they would start making computers with and integrated G4 that perhaps isn't upgradeable, but runs cooler in less space than a socket processor. A good step for their iMacs and Powerbooks, if they don't do it already. They can't really do the same for memory, as people will always want to upgrade their RAM. They just can't do away with DIMM slots.

In short, your vision of consolidation might not be wrong, but I think you have chosen the wrong companies. AMD and Apple have business alliances with IBM, a very strong company with little to no interest in consolidation with either company. Motorola is into too many things to be bought, but isn't doing well enough to buy anyone. Of all the options, they are the most likely to go under, but I wouldn't count on it. And Micron can continue to sell RAM as long as people continue making software that needs it. They probably aren't in that great a danger, especially in the next few months

Re:AMD vs Intel

[be-fan]

That's what I don't get about those people lambasting Intel for the P4 design. The long pipeline of the P4 was a *design decision* It had certain negatives (huge pipeline flush penalties, insane interrupt and system call latencies, high power consumption, etc) and certain positives (extremely high clock speeds, massive bandwidth and throughput). At the end of the day, what matters is the end result. The P4 is simply faster than any other chip in its class, and is very competitive with a lot of chips (50-70% the FPU performance of a Power4) that are way out of its league. This is how engineering works, plain and simple.

Re:AMD vs Intel

[kma]

Intel has a very strong history of designing chips that ramp up very well (except for their one CPU engineering failure, the Pentium Pro, which was too ambitiously designed).

While your overall thesis is correct, your parenthetical diss of the PPro is ass-backwards. The PPro core is the same core that has been tweaked into the PIII. It had an approximately seven-year productive span, during which it was very competitive, although it was considered a monstrosity in its original incarnation as the Pentium Pro. I.e., the PPro is exhibit A in your argument, not some sort of exception to be excused.

Re:This should happen (more commentary)...

[$$$$$exyGal]

IBM, of course, will also benefit. The company rivals Intel in semiconductor research, but its chip sales are only one-eighth as large. Licensing its technology, and manufacturing chips for other companies, opens revenue streams for Big Blue.

Licensing technology out to others is exactly what IBM should be doing in this case. It helps the industry and that will, in turn, help them. This is great for IBM in the short term and the long term.

--naked

Re:IBM + AMD = APPLE G5?

[Anonymous Coward]

What is with you people and spelling? Try "boost," and "behemoth." Unless Apple really does need a "bost" to beat the "bohemith." I mean, really.

Re:Motorola is Slow???

[Daniel_Staal]

Motorola is never *achem*G5*achem* late!

You're only late if the product is shipped...

This is not a major alliance

[Anonymous Coward]

Everyone who's talking about Apple 64 bit hammer-based PPC whatever is totally off the mark. They're cooperating on future technologies that we won't see for years and which have nothing to do with processor design, only processor technology. There will be no visible effects from this for at least the next year, and even if/when it does lead to change, it will be change most people won't even notice.

Re:IBM slow... AMD... hot

[binaryDigit]

IBM is simply offering production services to AMD, this is not another AIM alliance or anything like that. So IBM mgmt slowness should not be an issue at all. AMD simply is another IBM Semi customer.

As for IBM mgmt, well yes, in many ways IBM is the poster child for the slow and ponderous company. However, when they decide to do something (and do it right, well as much as can be expected) they can be the unstoppable force. RS/6000 and ThinkPad are two excellent examples.

If You're Thinking "PowerHammer", "IBM Athlon"...

[0x69]

If you're thinking "PowerHammer", "IBM Athlon", etc., then you REALLY don't get it. This is like GM and Ford cooperating on expensive research needed to meet new Federal safety & pollution standards:
- An engineer who used a saw to cut GM & Ford airbags or catalytic converters open a few years later could see similarities in the technology.
- Someone who was expecting to see a Pontiac Mustang or Lincoln DeVille would be SOL.

Celeron...

[MamasGun]

"Mighty fine nerve tonic...and boy do I have nerve!"

Cartoon reference...anyway, I digress...

The first Celerons were junk. No cache. Wretched things. Ugh. However, the Mendocino Celerons, the Coppermine Celerons and best of all the Tualatin Celerons were almost as good as their respective PIII "big brothers." Great price point made them the choice for anything that didn't absolutely, positively need all the caching the PIII provided.

However, the "Celeron...eew!" equation has become a reality again. Basically the chips Intel are selling as Celerons are very neutered P4s. Avoid them like the plague.

Not so sure it was a bad business move...

[Kjella]

I wasn't surprised when the AMD Athlon pulled out ahead of the Pentium 4, then fell very far behind. The Athlon was not engineered to ramp up well over 1 GHz. AMD was very foolish to race to that point, seeing as how long it took them to get working silicon at just 2 GHz.

It gave AMD some very good credibility, having the fastest processors, not just being some copycat always behind. Without that credibility, they would never have gotten Athlon MP in on the market. Most likely they wouldn't get money for the investments they'd need (and still do) to keep up with Intel otherwise.

It showed that Intel could be beaten, at least for a short while. Kinda like the gfx cards. Geforce, Geforce DDR, Geforce 2, Geforce 2 GTS, Geforce 3, Geforce 3 Ultra, Geforce 4... Radeon 9700! Ok Nvidia might strike back just as hard with Geforce FX, but it's the same thing.

Besides, it's not like AMD is really far behind. I've seen AMD2800+ in the stores, Intel has 3.06GHz (assuming those PR ratings are still close to valid). Of course AMD is now playing pretty much every design trick in the book (FSB, additional layer, minor core improvements+++) to keep up, so they need SOI and/or Hammer fairly soon, but they're still in loop.

Kjella