If the Prescott delay is true, what do you think Intel will do?

[AgaBoogaBoo]

This is in regards to: Link

I think if they really can't get it selling to the end consumer before the end of this year in the retail channel, I'm guessing they will introduce a new version of the current processors. They can still ramp up their current cpus maybe 200 more mhz, but after that they will need better cooling than what is currently available. I don't think that will place to large of a problem because they can use copper and a faster spinning fan, or a larger one spinning at the same speed.

I th ink they will either do that, or concentrate heavily on other areas such as mobile computing and ramp the speeds on those up a lot. I'm curious to see how they handle PCI Express...

 

[jim1976]

Hard to say...
But i don't believe that Intel will delay especially now with the arrival of A64.

 

[Duvie]

I know for a fact that the current p4c chips could easily have a 3.4ghz version. If they delay then they should get out another p4c and possibly those phantom p4ee chips....

The prescott is rumored to have 4 steppings already and I don't think the heat thing is fixed....I htink it is time for multiple cores and parallel processing in desktop pc's...We are getting 2 small and too high of speed and thus transistors. Heat and power are starting to take toll on intel chips.

 

[DaveSimmons]

> I htink it is time for multiple cores and parallel processing in desktop pc's

And/or to add more instructions-per-clock like A64 and Bannias
It's pretty impressive of A64 to really be "3200+" at a 2.0 GHz clock.

 

[mechBgon]

I wouldn't be surprised to see another speed bump from the Northwood-C's to tide them over, as long as the mobos can supply the necessary power. In reality, we know that many can (and do), but whether Intel feels it falls within spec might be another question.

 

[AgaBoogaBoo]

Originally posted by: DaveSimmons
> I htink it is time for multiple cores and parallel processing in desktop pc's

And/or to add more instructions-per-clock like A64 and Bannias
It's pretty impressive of A64 to really be "3200+" at a 2.0 GHz clock.

Before I used to read Anandtech, I read a lot of the articles at Cnet News, and multiple cores were something they were touting for some time. I too think they need more instructions per clock to speed up things a bit because I think they have been focusing a little too much on the clock speed side of things. Actually, that is also my guess as to why its being delayed. I think the heat problems affected development/production, but they worked around it only to face the problem of performance. Thats all a theory, but some reason it seems like a logical explanation in my head...

 

[pm]

I know for a fact that the current p4c chips could easily have a 3.4ghz version. If they delay then they should get out another p4c and possibly those phantom p4ee chips....

How could you know for a fact that Intel could easily have a 3.4GHz Pentium 4? Because people who put a lot of money into high-quality motherboards, very high quality RAM, way overspec'd power supplies, big heavy heatsinks with loud fans and well ventilated cases can overclock their parts to about 10% higher (3.7GHz or so)? What will these parts do in some cheap system? I know for a fact that I have never heard of any semiconductor company sitting on a premium part for any time after it's completed validation.

Heat and power are starting to take toll on intel chips.

Intel is first to 90nm. We'll see how everyone else does when they get there. 100W parts are going to be a fact of life in a couple of years. Leakage current increases by 1000% between generations from here on out if you perform simple scaling. Even with improvements to this, leakage is still going to get worse or hold steady from here on out. Power is going to be the big limiter in a few years.

Edit: My apologies to everyone if I come across a bit terse/rude. I'm posting in a bit of a hurry. No rudeness intended.

 

[Ionizer86]

As for Intel specs, if people are pushing to 3.7 safely, Intel probably could do 3.4 relatively safely. In the past, they judged the coppermine to be safe at 1.1, but that was a total disaster. They didn't judge the limit of their chip very well, unlike relatively conservative overclockers (eg running a Barton 2500+ from 1.83 --> 2.2 isn't pushing it too far considering the chip could do another 100 MHz).

 

[Idontcare]

Originally posted by: pm

I know for a fact that the current p4c chips could easily have a 3.4ghz version. If they delay then they should get out another p4c and possibly those phantom p4ee chips....

How could you know for a fact that Intel could easily have a 3.4GHz Pentium 4? Because people who put a lot of money into high-quality motherboards, very high quality RAM, way overspec'd power supplies, big heavy heatsinks with loud fans and well ventilated cases can overclock their parts to about 10% higher (3.7GHz or so)? What will these parts do in some cheap system? I know for a fact that I have never heard of any semiconductor company sitting on a premium part for any time after it's completed validation.

Heat and power are starting to take toll on intel chips.

Intel is first to 90nm. We'll see how everyone else does when they get there. 100W parts are going to be a fact of life in a couple of years. Leakage current increases by 1000% between generations from here on out if you perform simple scaling. Even with improvements to this, leakage is still going to get worse or hold steady from here on out. Power is going to be the big limiter in a few years.

Edit: My apologies to everyone if I come across a bit terse/rude. I'm posting in a bit of a hurry. No rudeness intended.

I wasn't aware Intel was the first to 90nm. Or that they were shipping 90nm product at all. I must have missed that press release. Oh wait, maybe I found it here Intel delivers 90nm parts...oops, my bad

 

[Redviffer]

Originally posted by: Idontcare

Originally posted by: pm

I know for a fact that the current p4c chips could easily have a 3.4ghz version. If they delay then they should get out another p4c and possibly those phantom p4ee chips....

How could you know for a fact that Intel could easily have a 3.4GHz Pentium 4? Because people who put a lot of money into high-quality motherboards, very high quality RAM, way overspec'd power supplies, big heavy heatsinks with loud fans and well ventilated cases can overclock their parts to about 10% higher (3.7GHz or so)? What will these parts do in some cheap system? I know for a fact that I have never heard of any semiconductor company sitting on a premium part for any time after it's completed validation.

Heat and power are starting to take toll on intel chips.

Intel is first to 90nm. We'll see how everyone else does when they get there. 100W parts are going to be a fact of life in a couple of years. Leakage current increases by 1000% between generations from here on out if you perform simple scaling. Even with improvements to this, leakage is still going to get worse or hold steady from here on out. Power is going to be the big limiter in a few years.

Edit: My apologies to everyone if I come across a bit terse/rude. I'm posting in a bit of a hurry. No rudeness intended.

I wasn't aware Intel was the first to 90nm. Or that they were shipping 90nm product at all. I must have missed that press release. Oh wait, maybe I found it here Intel delivers 90nm parts...oops, my bad

Is the stuff TI makes just as complicated as Intel?

 

[Pariah]

Intel doesn't have to do anything. If AMD could sell every CPU they made as soon as it was made they still wouldn't be able to catch Intel in market share due to Intel's brand name recognition and overwhelming manufacturing capabilities vs AMD. The A64 isn't going to gain AMD any major market share any time soon if they can't get any OEM's to sell them. Anyone seen any advertised systems from any of the major OEM's yet (Gateway, HP, Compaq, Sony? no Dell obviously)? I haven't, and that's not a real good sign for AMD.

 

[mechBgon]

Originally posted by: Pariah
Intel doesn't have to do anything. If AMD could sell every CPU they made as soon as it was made they still wouldn't be able to catch Intel in market share due to Intel's brand name recognition and overwhelming manufacturing capabilities vs AMD. The A64 isn't going to gain AMD any major market share any time soon if they can't get any OEM's to sell them. Anyone seen any advertised systems from any of the major OEM's yet (Gateway, HP, Compaq, Sony? no Dell obviously)? I haven't, and that's not a real good sign for AMD.

Here's an article to that effect: <a target=new class=ftalternatingbarlinklarge href="http://www.xbitlabs.com/news/cpu/display/20030617170716.html">AMD Athlon 64 to be Adopted by HP!
</a>

(Google smiles upon us )

 

[MadRat]

If Intel is having any problems with Prescott its least likely from its automated design putting too much too close together. If its true they nixed 10% of the internal pathway lengths then perhaps there is some hope for a P4-"d" stepping release at 130nm using similar space savings to hold the tide over.

 

[Pariah]

Originally posted by: mechBgon

Originally posted by: Pariah
Intel doesn't have to do anything. If AMD could sell every CPU they made as soon as it was made they still wouldn't be able to catch Intel in market share due to Intel's brand name recognition and overwhelming manufacturing capabilities vs AMD. The A64 isn't going to gain AMD any major market share any time soon if they can't get any OEM's to sell them. Anyone seen any advertised systems from any of the major OEM's yet (Gateway, HP, Compaq, Sony? no Dell obviously)? I haven't, and that's not a real good sign for AMD.

Here's an article to that effect: <a target=new class=ftalternatingbarlinklarge href="http://www.xbitlabs.com/news/cpu/display/20030617170716.html">AMD Athlon 64 to be Adopted by HP!
</a>

(Google smiles upon us )

I have no doubt these companies will eventually get around to releasing A64 products, but there was nothing at launch, which is pretty embarrassing for AMD for such a major product, and there still is nothing even officially announced. That link is from early June, and there is no mention of an A64 system on HP's site. Who wants to wager whether of not Dell and other major OEM's will have Prescott systems ready to ship at launch?

 

[mechBgon]

What about Cray, Sun and IBM's support for Opteron, or is that not close enough to A64 to suit you?

 

[Pariah]

Last I checked, the A64 and Opteron were not the same product. If those contracts are the only sales AMD gets out of the Opteron/A64/A64 FX the company is as good as gone.

 

[mechBgon]

You're so pessimistic at this time of night, Pariah

"HP will offer the Athlon 64 processor as an option on select configure-to-order HP Pavilion and Compaq Presario consumer desktop PCs later this year," company spokeswoman Kristin Reeves wrote in an e-mail message. "The company is also expected to offer the Athlon 64 processor on select HP Pavilion and Compaq Presario notebooks in the 2Q/3Q 2004 timeframe." In Europe, certain desktop models will begin featuring the processor in the next few weeks, she added.

from here

I'm still trying to pry an Athlon 64 out of HP.com. I must admit, "Athlon 64" is one hard-to-search-for name. Lots of results, none of which are what I want.

 

[mechBgon]

Heh, HPshopping.com has gone nighty-bye for the time being I'll see what I can find in the morning.

 

[Pariah]

You're not listening to what I am saying. I said there were no systems available at launch. How is that a debatable point? There weren't. There are still none now. Again where is the point for debate? I also stated that I know there will be some at some point in the future, but as of now there have been no official announcements excepts hints dropped by HP that they definitely will be. You can stop looking at HP's site, this is in the link you provided:

""HP will offer the Athlon 64 processor as an option on select configure-to-order HP Pavilion and Compaq Presario consumer desktop PCs later this year,"

So at some point they will be releasing systems at an unspecified date, but as of a week ago they still didn't know when. Intel has very little to worry about with the sad rollout the A64 has had so far, and despite what diehard AMD backers would have you believe, Intel is not in a position where they have to release the Prescott on time to keep their dominant position in the market.

 

[mechBgon]

I suspect that HP would rather release the A64's with 64-bit WindowsXP, personally. What do you think?

 

[Pariah]

It's possible. Has MS hinted at a possible release date yet?

 

[mechBgon]

Originally posted by: Pariah
It's possible. Has MS hinted at a possible release date yet?

"First half of 2004?" That's what I'm seeing here, however accurate it might be.

My guess is AMD would like to bring down the cost of making A64 dies by getting their 90nm act together. I wonder if Dell will ever sell their souls and try an AMD-based model...

 

[Pariah]

Well, if MS is planning on a first half 2004 release and HP is planning on a 4th quarter 2003 release, I doubt that XP 64bit is the cause of the delay, unless HP plans on shipping systems with a beta OS.

 

[imgod2u]

And/or to add more instructions-per-clock like A64 and Bannias
It's pretty impressive of A64 to really be "3200+" at a 2.0 GHz clock.

You don't really "add" more IPC. IPC is a statistic and depends highly on the application being run. You can add things to take more effective use of idle clockcycles and idle transistors and/or add more transistors/clockcycles, however, such as an integrated memory controller and a packing/unpacking stage to improve integer execution efficiency and possibly improving IPC. The word there is "possibly".

 

[thorin]

I said this a few days ago but I'll say it again:

No big deal, even if they did have manuf problems or whatever they've managed to keep up "the fight" without releasing it so it's all that much better for them to spend time perfecting it, and working out supporting chipsets etc..... Why rush when there's no reason to.

Thorin