Sign in with
Sign up | Sign in
Your question

PRESCOTT DELAYED

Last response: in CPUs
Share
November 18, 2002 9:26:29 AM

http://www.theinquirer.net/?article=6285


pushed back to Q4 03 at the earliest.

.09 isnt an easy process as some intel faithful would have u belive

More about : prescott delayed

November 18, 2002 11:06:08 AM

Yes, interesting.

AMD announces .09 Hammer pushed back to H104 and Intel pushed .09 Prescott P4 to Q403. One person even intimated that AMD was now a whole year behind Intel in transition to .09. This news seems to suggest that both are a similar timeframe apart. They are about as far apart in release of .09 chips as they were assumed to be 6 months ago.

This news also gives Hammer a full 6 months of ramp time before Prescott comes out, which should once again put AMD firmly on top performance-wise until Prescott and posibly even after Prescott.

Prescott is looking more and more interesting and I think it will seriously *up* theante for performance once it hits, just as I suspect Hammer will do the same.

It's good to see serious performance improvements on the way from both companies!

Mark-

<font color=blue>When all else fails, throw your computer out the window!!!</font color=blue>
November 18, 2002 11:36:27 AM

well on this forum and on others i visit... u have people on both sides who try to COLOR things to put their favorite chip company in the best light. INTEL doesnt MAKE .09 process tools and steppers... they buy them from the same company or companies that AMD can buy them from. So when AMD delayed their .09 chips, I figured the same would happen to INTEL...

Just like for some of the INTEL faithful here to explain how INTEL was sampling .09 parts with such ease that they now are a YEAR away ?
Related resources
a b à CPUs
November 18, 2002 11:43:06 AM

This is actually good news for people buying a P4 now, as it means Intel will likely be forced to continue improvements to the Northwood, giving P4 buyers an extended upgrade path even if the Prescott doesn't work with their current board (no news if it will yet).

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>
November 18, 2002 12:05:02 PM

Rumor has it Prescott will be 166FSB, so it looks like new chipsets/boards all over again.

- JW
November 18, 2002 12:32:48 PM

That means 666Mhz quad-pumped....
WOW!!!

You can get the most powerful computer in the world, but it'll never surpass my genius
<font color=red>Fuahahahahah!</font color=red>
November 18, 2002 1:16:17 PM

I no longer believe it will have a 666 MHz FSB since Intel will chip P4s with 800MHz FSB beginning next year. So, why should they stay with 666MHz?

anyway, you will need a new motherboard for Prescott CPUs since they are based on 0.09 micron technology. A dieshrink by Intel always means a new socket...
November 18, 2002 1:28:58 PM

<i><b>GRRRAAAAAWWWW!!!</b></i>

Please, no caps. It makes you seem too much like AMDMeltdown, and we don't need anyone else like that, no matter which side they're on. :tongue:

Now on to other things...

Interesting that Intel plans to up the FSB again; it's probably a stop-gap bit of marchitecture to help it fight off Hammer. Possible problem with supporting DCDDR 400 though: DDR400 is barely out (not even officially blessed by JEDEC IIRC), has very delicate signaling (thanks to high pin count)--and Intel plans to be able to support it in dual-channel? Intel's having a hard enough time getting DC DDR333 going.

May be time for Intel to grant nVidia a P4 bus license?

<i>I can love my fellow man...but I'm damned if I'll love yours.</i>
November 18, 2002 2:31:17 PM

Well, by the time the 800FSB P4's hit the market, DDR-2 may indeed be available for mass sales in the market. And that would have no problem running at 200MHz DDR.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.
November 18, 2002 4:30:41 PM

I've looked at a lot of roadmaps lately and i'm wondering if Intel is either not releasing these roadmaps and they are just being constructed by whoever feels like playing with photoshop, or Intel is intentionally releasing information that is wrong. Here's my reasoning:

If you look at the roadmaps that have pushed Prescott back, try looking at the P4 northwood line. Most of them have the 3.2 listed sometime in Q2'03 and the 3.4 listed sometime in Q3'03. Is this feasible? Will Intel really only move up their main line consumer processor 200 Mhz in 6 months?? That just seems rather small to me. So my thought is this...since AMD is obviously not coming out with anything soon (their "officially released" processors still aren't completely on the market), maybe Intel is just drawing out the roadmaps to be able to pad their schedule (why not, right?). Then, if AMD looks like it will release something, Intel can jump on it and release before them, or right after.

The other obvious thought is that Intel is simply slowing down due to laziness. Although i doubt that is the case in this economy (and also considering their recent presumptions that Intel-servers will start bringing in more revenue than RISC-servers).

And i'm still wondering how well AMD is going to succeed in the next few months. Yes, they just dropped a ton of their workforce, but the savings from that is barely enough to cover the next quarter losses, if their revenue rate continues as it has. No one wants a completely Intel-controlled market, come on AMD, do something! :) 

Athlons and Pentiums are just melted rock. Who’s rock is better? Who cares, let’s play some games
a b à CPUs
November 18, 2002 5:23:47 PM

Not if it's pin-compatable, or don't you remember PIII EB's on the BX?

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>
a b à CPUs
November 18, 2002 5:25:15 PM

Die Shink does not equal new socket, the Tualatins went to a die shrink and used the same socket, just a couple pin definitions changed-which meant you could use them with an adapter, or modify your board.

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>
November 18, 2002 5:50:52 PM

wolverinero79:

I agree.

I think the economy is slowing intel down. Who wants a $600 processor? Everyone is buying the $200~300 ones because they're plenty fast enough for 95% of the population. If they can get the speeds out of the 0.13 line and still keep up with AMD why bother shelling out all the $$$ to rush 0.09 into production when you can take your time and produce a better chip for less money eight months down the road.

Part of it is probably the excellent yields on the C1 stepping - people are running them well over 3ghz. I know this is probably an extreme example but there's a guy at overclockers.com running a 1.8A C1 @3.5ghz @1.525V (95% o/c - default voltage - standard aircooled). If intel can sell 3.4ghz+ SMT Northwoods for $600 and keep the 3.06 in the $200~300 range I dont blame them for holding back on Prescott.

That'll keep AMD in the running as well - as long as they can make the chips they keep releasing on paper. AMD wants at most 18% waste from wafers (82% yield). Dresden can make 50m 0.13 chips a year (12m a quarter) for AMD, but AMD only expects to sell less then half that - and I don't think it's from lack of demand. I wouldn't be surprised if they're getting more then 30% (or even closer to 40%) waste. If they can't make 2ghz+ Tbred-B chips (2700+, 2800+) chips in volume, it's going to be even harder to make 2ghz+ Barton chips in volume because of all the extra transistors the 512k L2 adds. They need a lucky stepping like intel has with the C1, and they need it soon.

BTW: Nice sig!


Crashman:

Good point, but if it's really a 4x200 FSB I don't think many existing boards will do 200FSB.

- JW
November 18, 2002 7:05:59 PM

Vacs, as many said, on a 200MHZ FSB we are talking of three possibilities:

1)Intel goes for the first time for P4, in asynchronous clocks for their HIGH END. I doubt that 95%, Intel just won't go make people get 200MHZ FSB bus P4s with 166MHZ memory.
2)We rush in DDR 400 which currently is no longer accepted by major corporations, has been nearly shunned by JEDEC. DDR-2 is close, but even then, some people are doubting its success now that other technologies are rising, as well as DDR-3 technology found, if I am not mistaken, by ATi and some others.
3)Rambus gets RDRAM PC1600 out. At the moment, PC1066 is barely found around, I still have no seen any Canadian shop selling it around me, nor on the Canadian web. Can you imagine the time for PC1200 to come out?
And THEN PC1600? There are even yeild doubts about over 1200MHZ speeds! There's little chance to none of getting something from Rambus soon, to cover 200MHZ FSB speeds, let alone the 166MHZ! That would mean even if Intel uses 666MHZ FSB, Rambus may have trouble supplying the correct modules!

Therefore, only DDR so far, with Dual Channel, can help, and in case number 2, it can't if 800MHZ FSB is chosen.
So I doubt we'll hear of it any soon, possibly after H1 04.

EDIT: I just checked the Inquirer link itself. It sounds too good to be true, it's just not "Intel". Does anybody feel that way?

I was glad however to see lower end P4s get HT, unfortunatly the idiocy here is that it will launch in Q2 03, at a time few will buy those anymore. Worse is that I can forget about putting this in my friend's new Xmas system.
--
*You can do anything you set your mind to man. -Eminem<P ID="edit"><FONT SIZE=-1><EM>Edited by Eden on 11/18/02 04:21 PM.</EM></FONT></P>
November 18, 2002 7:16:57 PM

It ain't really good news.

Think about it, NW 3.06GHZ HT is currently drawing a whopping 82W of power and probably runs quite hot compared to before. It would seem to me that the NW will hit a limit soon due to the power constraints, and the fact few have the feel to get a new motherboard, as WELL as possibly a new PSU to accomodate any power usage over 100W, which would greatly hurt the value in that segment.
It seems Intel will have to scale NW further, but I doubt it will be any competitive as we first thought. Reach 3.6GHZ and you'll already have lots of heat and power problems. I am guess a maximum of 600MHZ will be extracted off 0.13m after 3.06GHZ P4s. Will that allow AMD any actual benefit?
It all depends on the ClawHammer strategy. If it launches in Q2 and can be available, and has competitive performance which beats the top of the line from Intel, AND Intel has problems ramping the NWs with their new power demands, then yes AMD can, for some months actually regain the market share. But this seems like such a weak chance event, that it would surprise me to see Intel let go for a little.

--
*You can do anything you set your mind to man. -Eminem
November 19, 2002 12:29:27 AM

I agree, i think Intel's stepping procedures have been incredibly beneficial. I think this is mainly not because of luck, but since they are on top, they don't have to push the technology. They can focus on getting excellent line yields instead of shrinking the gates (i'm obviously talking about gate width variance in the process of making the same line of chips, not .13 to .09, etc.). AMD, on the other hand, is a bit behind, so they are more than likely trying to push their technology as hard as they can. It's engineering trade offs, performance vs cost (to the company).

Since Intel does have those great yields, it's obvious that they can and probably will continue to cut prices, leaving AMD with little ground to stand on. As stated previously many times from many people, the next 6 months are going to be very important to the chip field.

Athlons and Pentiums are just melted rock. Who’s rock is better? Who cares, let’s play some games
November 19, 2002 12:53:29 PM

1H03: intel "Springdale" northbridge + ICH5:
- 4x133/200 FSB support
- Dual channel DDR400 (DDRII)
- AGP 8x
- Serial RAID ATA-150
- 8x USB 2.0
- LAN

2H03: intel "Canterwood" northbridge + ICH5:
- SMP version of above :) 

- JW

[edit]
DDR-II is expected to be made available as DDR400, DDR533 and DDR600 modules, with the initial DDR400 modules offering a memory bandwidth of 3.2Gb/s. Unlike standard SDRAM and DDR SDRAM, which use TSOP packaging and run at 3.3 or 2.5v, DDR-II modules will ship in BGA or TSOP-II packages and will require 1.8v. DDR-II is backwardly compatible with DDR, allowing the use of DDR modules in DDR-II motherboards.
[/edit]<P ID="edit"><FONT SIZE=-1><EM>Edited by JCLW on 11/19/02 09:56 AM.</EM></FONT></P>
November 19, 2002 2:52:27 PM

nice to know that, i thought springdale was only going to support ddr 333,

but wasn't it also going to support dual channel ddr

it's a shame imo that they delayed preston its going to be such a bump (4Ghz and dual ddr 2 on 6,4 Gb of memory bandwidth, who woudln't wanna have a machine like that :-p

Remember the time You used 20Mb harddrives
November 19, 2002 4:16:51 PM

I did mention dual channel memory :) 

E7205 (Granite Bay) supports dual channel DDR, expect a dual channel DDR333 version out Dec/Jan. That and a 3.2ghz Northwood would keep you going for a while I bet.

- JW
November 20, 2002 3:10:49 AM

Actually with rdram you can pull it off with 4 channel 32bit rdram, with modules made up of 800mhz devices. But I see nothing like that in intels roadmap.
400mhz ram is avaialble but it's 3.5-2-2 which is a lot slower than 2-2-2. Also I wouldn't even buy 400mhz before the specs are release for dual channel DDR, cause of there are lots of variables. To me the 3.5-2-2 just doesn't seem attractive for 400 ddr.
November 20, 2002 3:35:42 AM

But what about heat output... Without a die shrink thinga are gonna get pretty toasty!

the 3.06 is allready putting out 80+ watts of heat, using a new stock cooler to combat it.

<font color=red><b><i>"Humans are just overclocked Monkeys!"</i></b></font color=red>
<b>Serving THGC for over 2 years.</b>
November 20, 2002 10:49:28 AM

Dude, DDR400 can be obtained at CAS 2 from some extreme timing brands.
Now if the first number in your 3 timings is the CAS, in no way does DDR400 run at CAS 3.5!

As for RDRAM, well it's greatly Rambus' fault that they are not doing anything to continue the RDRAM support. We barely find PC1066, and in Canada they are expensive as hell, how would Rambus start rushing 64-bit RDRAM or 2*32-bit?
Yellowstone is not slated for desktop so far, so there goes another worthy technology. It's all about "dragging ass".

--
*You can do anything you set your mind to man. -Eminem
November 20, 2002 2:57:55 PM

Quote:
A dieshrink by Intel always means a new socket...


not true: what about the williamette .18 socket 478 dieshrink to northwood .13 socket478?
November 20, 2002 4:38:34 PM

P3 Coppermine had a slot 1 version as well.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.
November 20, 2002 6:53:17 PM

DDR 400 from mushkin one of the best places to get high quality ddr doesn't even have ddr 400 at 222. Mushkin has it at 2.5-3-3-3, for 129US a stick not exactly all that cheap.
But when it comes to DDR I don't trust buying it from just anyone or you could end up with bad ram.

Anyway there is a lot of talk that DDR400 may not even be finalized by 2nd quarter next year, so intel may not be able to pull it off, we'll see.
November 21, 2002 6:43:49 AM

but you're missing the point, a die shrink doesn't MEAN that you HAVE to change the socket.
!