3 Ghz P4

[LarrySW]

Any ETA on when we should see these? I want to get a new system but would hate to get a 2.53 and then the 3 ghz comes out. It is ok when you have a P4 2.2 and a P4 2.4 comes out but it's different when that first decimal point changes. 🙂

 

[Soulkeeper]

most likely in the fall
depends how hard they feel they gotta compete

 

[Athlon4all]

Christmas season at earliest, prolly Q1'03 according to offical roadmaps

 

[Rand]

End of this year, or early Q1 2003. Probably depends a lot on how pressed they feel by the Hammer', and how quickly AMD's able to oush the ClawHammer into the mainstream.

 

[pm]

depends how hard they feel they gotta compete

Probably depends a lot on how pressed they feel by the Hammer...

No offense, Rand and Soulkeeper, but I disagree. The product will ship when it's ready as soon as that will be. No one in the CMOS semiconductor industry holds back on shipping a faster version of silicon - it costs nothing extra to make a 3GHz than it does to make a 2.2GHz if you are able to make it go that fast. It's not like speed costs extra money... in fact, in the cases, of process shrinks where the CPUs get smaller and faster on newer process technologies, it costs less to produce it... when you can figure out what you need to do to make it.

If the chip can run that fast then you ship it as soon as you have it. No one in the semiconductor industry ever holds back a flagship product simply because, like the movie industry, "the time isn't right." It takes time to modify the design, to find the speed paths, to fix them. Each revision of silicon takes time on the order of months to turn around from design complete, to silicon returned from the fabs, to finding the lastest round of circuits that are holding you back, to fixing them, and then to design complete of this revision. And that's just on the design side, on the manufacturing side, they do a similar release plan for evaluating changes to the "fab recipe". It is not like you have a handle that says "chip speed" on the side of a machine in the fab and you crank the handle to "faster" whenever the competition comes out with a new design. These products have design lead times that are much longer than most people on the forums seem to think.

As far as the original question, I have to decline to coment since Intel employees are restricted from discussing future products - and especially release plans.

Patrick Mahoney
Microprocessor Design Engineer
Intel Corp.

 

[hopeless879]

Originally posted by: LarrySW
It is ok when you have a P4 2.2 and a P4 2.4 comes out but it's different when that first decimal point changes. 🙂

Intel is not going to jump right from 2.533 to 3ghz. There will be various speed inbetween them. You dont have to worry about a massive jump like that.

 

[etalns]

They're nto goign to do a jump from 2.53 to 3. They're going to get as much money from each processor as they can get, if they go from 2.53 --> 3 they are going to lose money in the future as the time between purchase of product and need for upgrade will increase.

Economics 101 anyone?

 

[pm]

They're going to get as much money from each processor as they can get, if they go from 2.53 --> 3 they are going to lose money in the future as the time between purchase of product and need for upgrade will increase.

This isn't how it works. First of all, Intel can't simply put out a 3GHz Pentium 4 right now because it's not available, any more than Cyrix can take their latest CPU and clock it up to 2GHz. In either case, it takes a lot of research and design to get these chips up to higher speeds. It's not easy and it's not quick. But even so, if Intel could release a 3GHz Pentium 4 right now, I'm sure that they would. And it has everything to do with economics - "he who has the best product can charge the most for it" - and right now a 3GHz Pentium 4 would be a very useful chip many in the workstation industry. And as I said, producing a faster CPU doesn't cost any more than producing a slower one once you know how to make the faster one in the first place.

There is a common fallacy that CPU manufacturers gradually release new CPU's incrementally to maximize profits. This is incorrect. Manufacturers gradually release incremental improvements in CPU designs because that's the way that they become available. If a CPU manufacturer could make a huge jump, then why doesn't, for example, Transmeta release a 3GHz version of their Crusoe right now? In order for this theory of incremental improvements to maximize profits to be correct all of the manufacturers would all have to be in collusion with each other. Think about it. This fallacy doesn't make sense when you take into consideration that there is real competition to release the best product as soon as possible. In addition to just Intel and AMD, there are Sun (UltraSPARC), IBM (Power CPU's), HP (PA-RISC), Transmeta (Crusoe), and many others that manufacture lesser known CPUs such as those from MIPS, Hitatchi, NEC, and Motorola. There are plenty of competitors... who would hold back a flagship product on purpose, and why? You only stand to lose out in the long run as everyone catches up to you.

There have been a couple of threads in the Highly Technical forum about this as well. Here is a relatively recent one. And here is a slightly older one that directly relates to this issue.

(on a side note, for those who are Anandtech subscribers, like myself, do you have problems if you click to links to archived messages such as the second link? I get an error message unless I log out. But if I log out, then it works fine.)

 

[BFG10K]

Nice posts PM although I have to disagree with you a little. Competition and marketing often have an effect on how fast and how soon certain products are released out the door. The entire industry operates that way and I don't see Intel being any different.

As for the 3 GHz part, I expect that Intel might be pushing for a Christmas/new year 2002 release to give the consumers something to be happy about. Failing that I don't expect them any later than Q1 2003.

 

[Athlon4all]

Intel is not going to jump right from 2.533 to 3ghz. There will be various speed inbetween them. You dont have to worry about a massive jump like that.

Well, it wouldn't be beyond them. Remember they jumped from what 800MHz to 1GHz to say "We have 1GHz too!" when the Athlon Classic first hit 1GHz.

I agree with BFG10K. If AMD/VIA didn't exist, then we would still be on Williamette's running at 1.5GHz. And I can easily see Intel jumping from 2.53 to 3.06 so they can in response to the Hammer launch, market "We've got a 3GHz CPU!!!! Hammer, it can only hammer a piece of paper!!!!". Now maybe pm's right, that Intel will release it when its ready, but only because AMD exists. Anyway, Im not interested in a big debate. I just feel that if current roadmaps are correct (stating 3.06 in Q1'03), then I can see if all possible, Intel pushing for an earlier launch than Q1'03. Just MHO.

 

[LarrySW]

I know there will be a 2.6, 2.8, etc before the 3 but I was wondering when the next decimal increase to 3 was. Thanks for the responses, now should I wait it out?

 

[Mavrick007]

The computer industry seems to work different than other consumer products. You don't have to upgrade a frying pan to get it to cook your bacon. It really works with supply and demand. When demand is low, release something a bit better and then wait to see who will bite.

I'm sure that they do have a 3.0ghz in the works right now but because of marketing they will bleed the consumer by pushing the numerous speed increases in between cause they can. Why release the best that they have when you can put something a little bit slower and if the consumer keeps buying it at a constant rate then great, if not, they will release a little bit faster and see how that fares. When that speed weens off, they will release the faster chip, and so on.

They are not going to release the fastest that they have cause then the market would be saturated with the fastest processor and the consumer would want something faster while R&D would be fiddling with new tech working overtime just so that they would have something to market asap. That creates alot of work to come up with a new chip so they're not going to push the best when you will buy the 2nd or 3rd best, otherwise they lose all their resources creating the "next best thing". Plus they probably don't have alot produced at the time so they keep working at decreasing costs on each cut of the silicon which gives them more profit and makes you think you're getting such a great deal when they reduce your sale price.

 

[Rectalfier]

I agree with PM. Many people are basing Intel's ability to produce faster chips, by the good overclocking results as of late. This is a misconception, as most of the people overclocking have to increase the voltage to do so. Intel needs the time to make these chips run at default voltage. Also, it's not neccesarily the highest speed grade that pushes the envelope. Intel need to make sure it can produce enough of it's mid grade model's, as they are the volume sellers and they need to produce enough without pushing demand by selling uber fast processors.

 

[CTho9305]

Originally posted by: pm

It is not like you have a handle that says "chip speed" on the side of a machine in the fab and you crank the handle to "faster" whenever the competition comes out with a new design.

😀

I dont see why people dont trust pm and didnt trust wingz.

think about it like this. if everyone is producing a 2ghz product, for, say $100, and you can produce a 3ghz part, if you only increment by 200mhz, you can charge, say, $150 for the 2.2ghz part. Then the price drops to $100 as you release a 2.4 for $150, and so on. OR you can right away release everything, with the 2.2 at 150, 2.4 at 200, 2.6 at 250, 2.8 at 300, and 3.0 at 350.

How does this make you less money? The vast majority of people and companies do not "upgrade" just because of speed increases. They upgrade when they feel their existing machines are too slow, and they replace them with high-end machines. So if you have a very fast part, then you could get more purchases of the higher-end parts earlier as people buy, say, 2.6 or 2.8 instead of 2.2ghz, paying you more. OR, they will still only want 2.2 and buy it, in which case there is no net effect.

However, the gains from the people buying the fastest stuff from you will put you WAY beyond your competitors.

Hopefully thta makes sense

edit: fixed missing quote, i hope!

 

[MadRat]

pm and Wingnutz-PEZ have nothing to do with the marketing hype released by Intel's spin doctors. They do have everything to do with the teams that put these modern engineering marvels together. Either accept their credentials or don't, but don't think you can bad mouth them considering these guys are some of our more revered members.

If pm said no 3GHz part is ready for release then its the honest truth. He has nothing to gain from lying. Nor would he willingly encourage Intel to release overclocked parts just to have the fastest computer part on the market. His investment in time and effort is worth far more than a temporary speed crown.

 

[pm]

Competition and marketing often have an effect on how fast and how soon certain products are released out the door. The entire industry operates that way and I don't see Intel being any different.

Which part of the semiconductor industry operates like this? If you have a 3 year design time, how can you have any specific idea what the competition will be?

Can anyone cite even one specific example of a great product and that was ready to ship, but wasn't for marketing or competitive reasons? I can think of plenty of products that shipped earlier than they should have (rushed to market), but I can't think of a single one that was held back from release for anything other than technical reasons.

In my 10+ years in the semiconductor industry, I have never even heard of anyone holding back a flagship part. There is no reason to do so. You can charge a higher premium if you get it out the door as soon as it is ready. By sitting on it, all you do is watch your lead disappear and your ASP's(average selling price) evaporate. Companies will hold back lower end products for marketing reasons, and sometimes a design will not be released due to lack of infrastructure (ie. lack of a chipset, lack of compatible memory, etc.) but I have never heard of anyone with a high-end product just letting it sit there ready to go waiting for a good time to release it.

A CPU isn't like the latest "Harry Potter" movie where you attempt to maximize revenue by releasing at a certain time, a CPU is like Nokia having a tiny cell phone that has phenomenal battery life. Why would you hold this product back from the market? If you sit around and wait, it's more likely that your competition will release their version and you will then have to match them in price. You release it as soon as it's ready, and you sell it for a higher price because it has a feature set that is compelling enough to command a higher premium.

 

[CTho9305]

pm, YGPM

edit: new avatar 🙂

 

[Acanthus]

PM tell me there is a reason that Intel Pentium 4 2.2a processors werent released with the 1.6 and 1.8a? Almost all 1.6a procs will go to 2.2 at default voltage, all they had to do was increase the multiplier, is there "multiplier technology" intel doesnt have? come on now 🙂

///EDIT Oh and i also remember at IDF LAST YEAR Intel was demoing a Pentium 4 at 3.5Ghz, this CPU was supercooled with an unspecified type, but it was also utilizing the "hyperthreading technology" that allowed the pentium 4 to execute multiple threads at once. Considering what they had then, and it was in use and working for the demo at IDF at 3.5Ghz, you have to specualate the yields of that CPU if it were not handpicked, or supercooled. I would make a safe assumption that the CPU demod there was also capable of 2.2+ Ghz with air cooling and 1.5v.

 

[goog40]

From Anand's GF4 article:

They[Nvidia] have consistently been producing new graphics parts in record times; NVIDIA was so ahead of schedule in fact that they could have released their newest GPU, the NV25, last December. The Santa Clara based company was on track to release the NV25 in December but held off in order to prevent cannibalizing sales of their GeForce3 line that was doing so very well.

I don't doubt that Intel isn't able to sell 3GHz P4s now, but hypothetically if they were able to right now, I don't think they would. There's no reason to at the moment since there's a decent gap between their fastest processor and AMD's fastest.

 

[BFG10K]

Let me put it to in another way PM: I'm not claiming that Intel purposely holds back but rather that they accelerate because of what the competition is doing. If it weren't for AMD I doubt we'd be seeing the 2.53 GHz P4 right now.

 

[imgod2u]

Has anyone mentioned that Intel has already demonstrated a 3 GHz air-cooled P4 at IDF? The point is not that they don't have a 3 GHz P4, it's how many they have. How many people have been able to overclock their P4 to 3 GHz on standard air cooling (never mind the retail heatsink and fan, which Intel MUST guarantee that the processor works fully stable with). Watercooling a P4 to 3 GHz is far different than being able to make enough of the chip that works with the retail HSF in mass volumes and still maintain a sizable profit. Now while I don't doubt that Intel could potentially speed up the process by pumping more money into R&D and refining the design to do 3 GHz that is stable when run on the standard HSF, this would put quite a dent into the budget. Remember the 1 GHz fiasco? Sure they had 1 GHz P3's but yields weren't good enough so there was a huge supply shortage. You don't see that with any of the P4's today. Why? Intel is ahead in the MHz game and there's no reason to release a processor before yields are good enough to provide mass volumes of them. I'm sure if Intel really tried, they could release a fully stable 2.6 GHz chip or even 2.8. But they may estimate the time taken to refine the P4 design to do 3 GHz with stock cooling to take longer and hence releasing the 2.6 GHz and 2.8 GHz now would mean they'd have to wait a long time before releasing another chip. In that case, they may indeed be holding back to gain a profit.