more itanium follies

[jhu]

hp and intel splitting, and it's even being reported by reuters. 10 years and billions down the drain. hp could have continued with their pa-risc and alpha. oh well.

 

[pm]

As one of the engineers living in Fort Collins who is actually personally involved in this, I thought this Reuters article was pretty misleading. Intel consolidated the design teams, and hired several hundred new employees from HP to work on IPF design in Fort Collins, and HP is increasing funding on Itanium development. It's not a negative event, it's a good thing. Everyone that I've spoken to here - many of who are moving coming from HP to Intel - are very pleased. Heck after we heard the news a huge group of us went out to dinner to celebrate.

Jhu, you make it sound like Itanium development is over. In fact, it's the other way around. Intel is creating a large design team to work on future Itanium development, and HP is pleding $3 billion to use on Itanium development.

 

[Insomniak]

When did Intel become the Gilligan of chipmakers?


Recent fark ups:

- Prescott
- Itanium
- Denying demand for 64 bit
- losing server contracts left and right
- 4Ghz canned
- The "Extreme Edition" (HAHAHAHAHAHA)

etc.

They need to get back on the horse.

 

[FishTankX]

Originally posted by: Insomniak
When did Intel become the Gilligan of chipmakers?


Recent fark ups:

- Prescott
- Itanium
- Denying demand for 64 bit
- losing server contracts left and right
- 4Ghz canned
- The "Extreme Edition" (HAHAHAHAHAHA)

etc.

They need to get back on the horse.

The extreme edition games a helluva lot better than the normal prescotts, and can compete reasonably with the FX to a limited degree. Utterly owning it in encoding and such. I don't see it as a 'Fark up'.

 

[VirtualLarry]

Originally posted by: Insomniak
When did Intel become the Gilligan of chipmakers?

You forgot the i432.

The thing is, if you only look at the "loser" chips that Intel has designed/produced, then you aren't seeing the whole picture. Intel, as a company, is big enough to have several design teams pursuing several different potential architectural families, and Intel just waits for the market to pick the winner, before they decide which to put the majority of their corporate effort (advertising/marketing, etc.) behind. Not too many other CPU makers have that option. So where one of their designs flops, another is waiting in the wings, or is being sold into another market segment, waiting to take it's place. It's a fairly brilliant strategy, but it has it's costs. Intel is probably one of the largest corporate spenders in terms of technology R&D, I think, and it shows. Intel just needs to hand control of the company back to the engineers instead of the marketers, and tell them to "just do things right".

I do tend to think that HP's involvement in Itanium was kind of a loss for them, they've spent billions on it, and killed-off their own high-end server chip line, basically conceeding that market to Intel, IMHO. What has HP gotten out of it? A couple Itanium server sales? I think Intel's biggest competitor in this space is Sun, and their Niagara architecture, massively multi-cored/multi-threaded, and with a vertically-integrated OS solution package to go with it to boot. Intel is mostly stuck with MS's OSes, although I'm sure that some of their funding/interest in Linux is intended to lessen their commercial dependency on MS for OSes.

 

[pm]

I do tend to think that HP's involvement in Itanium was kind of a loss for them, they've spent billions on it, and killed-off their own high-end server chip line, basically conceeding that market to Intel, IMHO.

Conceding that market to Intel? Intel sells microprocessors, HP sells systems. The market is not conceded - a relationship is formed that allows HP to become less vertically oriented and allows each company to pursue their competencies.

The real problem with the microprocessor business is who has the money to invest in designing and fabbing one? The design is very expensive in that it requires several hundred engineers and an up-to-date workstation and server network all for several years and then a fairly expensive back-end validation/verification flow filled lots more expensive engineers, and then lots of expensive functional testers and structural testers, logic analyzers, FIB machines, and assorted other chunks of expensive equipment. Then when you are done, you need to put it into a state-of-the-art fab - which currently runs about $3 billion. That's just to produce the CPU and maybe a few chipset parts. Which system company can afford to do this just to sell a couple of hundred thousand chips? Currently only one company: IBM. Sun only has to get the design out the door - they have TI fab theirs.

HP essentially handed off the design of their CPU's to Intel - which makes sense from the perspective that Intel has lots of fabs and needs to keep them filled, and Intel has lots of designers who have experience in CPU design.

As far as the comment about Gilligan's island... I disagree with several of them, and two are duplicates. Still, every company has their mistakes... hopefully 2005 is better than 2004.