Tejas postponed to 2005, prescott >120W

[Guest]

Where is AMDmeltdown when you need him ?

Long story short: Prescott not really available until Q2, hot hotter hottest (120W!), only scales to 3.8, maybe 4 GHz by Q4, Tejas not until 2005.

Even if we assume prescott performs on par per clock with the P4 EE (intel even claimed the EE will be faster overall), intel will only manage a 18% performance increase over the not even released EE over the next 12 months. Given hammers (much) better scaling with frequency (cf Aces overclocking article), AMD would not even need a 2.6 GHz part to match that. Looks like AMD can screw up their .09 transition and still claim the performance crown for pretty much all of 2004 without much effort.

<A HREF="http://www.xbitlabs.com/news/cpu/display/20031011084615.html" target="_new"> Xbit article </A>

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[Guest]

Oh wait, that is without taking 64 bit OS and software into account, nor a non registered RAM socket 939 FX.

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[Spitfire_x86]

I was just about to post that link

----------------
<b><A HREF="http://geocities.com/spitfire_x86" target="_new">My Website</A></b>

<b><A HREF="http://geocities.com/spitfire_x86/myrig.html" target="_new">My Rig & 3DMark score</A></b>

 

[Spitfire_x86]

popey's claims doesn't seem very incorrect at this moment

----------------
<b><A HREF="http://geocities.com/spitfire_x86" target="_new">My Website</A></b>

<b><A HREF="http://geocities.com/spitfire_x86/myrig.html" target="_new">My Rig & 3DMark score</A></b>

 

[papasmurf]

wow this will give amd a good shot with their 64bit stuff, if they do a good enough job promoting it intel will find that their 32bit cpus are not gonna be easily sold to the computer illiterates...

Join the TomsHardware IRC channel <A HREF="http://skulls.sytes.net/tom/ " target="_new">http://skulls.sytes.net/tom/ </A>

 

[Mephistopheles]

That is if they can't work around the heat problems at all... But seriously, Scotty was supposed to reach 5Ghz. How could they have been so off? Maybe the physical challenges associated with 90nm manufacturing are harder than initially thought... They must be running into lots of quantum theory...

Anyway, I think mass-availability of Scotty by 2004 is nothing new at all... It will still be launched this year. After all, even A64 is only expected to be readily available in quantity in 2004... FX in particular - neither P4EE nor FX are expected to sell a lot...

And Tejas being available in late 2004 still hasn't changed - but it's that same story - massive availability only in early 2005. I hope they don't run into too many problems on the 65nm front!

And I also hope they develop some sort of 64-bit extensions of their own (or support x86-64, for all I care). Looking for less conventional means to increase performance (as opposed to increasing clock and cache) is always a good choice.

<font color=red><b>M</b></font color=red>ephistopheles

 

[Parhelia]

If intel will continue rising the clock speed like this, they will fail, cos as the clock speed get higher the problems with it gets higher too, and the most important is the heat issue (more than 120watt), so intel to keep with AMD they should first work on the ways to increase speed withought increasing clock speed. amd has increased its 2.2ghz athlonxp speed more than twice with the FX runing at 2.2 GHZ also. So like i said the road for intel is blocked if they continue with this. for me the simpliest way and the most foolish way to increase performance is by clockspeed and by cache size, for me those are only used with the need of it, like the end of a cpu architercture, so they puch it to the max, and finish they swicth to a new design.
Intel will fail like this, and its not good for the computer market, cos they will be no more competition, and if intel thinks that theire marketing on seling processors, are based on the uninformed ones, so soon they will be.

 

[HardWareBoss]

Tejas Postponed to 1st Qtr 2005 was original release date.

 

[flamethrower205]

Way u pose it, seems like Intel's in serious trouble!!
Hehe

The one and only "Monstrous BULLgarian!"

 

[Mephistopheles]

amd has increased its 2.2ghz athlonxp speed more than twice with the FX runing at 2.2 GHZ also.

Increased, OK. More than twice? Didn't see that... That's a tad extreme... A 3200+ is slow, but not more than half as fast as FX-51!

<font color=red><b>M</b></font color=red>ephistopheles

 

[traviss187]

I wonder if they could adopt their pentium M technology to desktops, they preform incredably well per-mhz. If they increased the voltage a little and dumped some more ram they'd still have a low heat-highly efficient processor.

But i'm pretty much just speculating and could be way-off,

 

[guesswhosback]

Kill the messenger

how did i know prescott would be delayed..

simple

i look back to every major intel launch... and they all were consistant...

people had fast samples early, there were never any rumours of delays...there was never a hint of fab problems... intel hit its dates in stride

this time everything was the opposite...

i said PRESSY was stillborn i was called an idiot and a buffoon.....

who are the buffoons now....

my saying is ... never have too much faith in a BRAND NEW PROCESS....if Hammer was launched at .09 i would have said the same thing about it..

ask Nvidia with its .13 geforces

 

[eden]

Um dude, that proves nothing about nVidia. ATi released the 9600PRO, brand new process on their own cores and had absolutely NO PROBLEMS. They are not even tied to nVidia's manufacturing, so they probably had no ideas how to make it work right.
Though I'm just pointing out that the nVidia vs ATi process transition issue does not necessarily correlate with Intel and AMD's issues.
I still think AMD will run into a lot of 0.09m problems. Especially migrating SOI to it.
--
<A HREF="http://www.lochel.com/THGC/album.html" target="_new"><font color=blue><b>This just in, over 56 no-lifers have their pics up on THGC's Photo Album! </b></font color=blue></A> :lol:

 

[eden]

You know, I have a hard time even concluding anything from that article. They have no real claims from Intel themselves. All of it comes from them and some Japanese site.

Big deal, this is nowhere to really state the truth. No one even knows anymore what Prescott will feature. From rumors of SOI to x86-64, it will be a surprise if it's delayed.

You're taking this article's claims way out of proportions.

--
<A HREF="http://www.lochel.com/THGC/album.html" target="_new"><font color=blue><b>This just in, over 56 no-lifers have their pics up on THGC's Photo Album! </b></font color=blue></A> :lol:

 

[Guest]

> it will be a surprise if it's delayed.

Only for you and the most faithfull intel apologists I think. For everyone else, it is pretty much a given prescott will be late, hot and underwhelming. Anything else would be a surprise.

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[Guest]

>Way u pose it, seems like Intel's in serious trouble!!
>Hehe

Not really. It does look like they will lose some marketshare and mindshare in the highend, but it takes a LOT more to get intel into serious problems. They where facing a similar scenario in the early Tbird days, but i don't recall that getting them into any serious trouble. Intel surely can live a year without having the faster/best desktop cpu out there, and if tejas supports some 64 bit ISA, all bets are off. The most significant deduction you could make, is that AMD might perhaps finally be making some money again in 2004.

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[Guest]

Roadmaps change all the time, almost by definition, but last time I checked, tejas was supposed to roll out by Q404. Even prescott was at some point slated for Q203.

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[Guest]

>But seriously, Scotty was supposed to reach 5Ghz.

Yes, and by 2005 or so we where supposed to be seeing 10 GHz chips, reaching 5 GHz looks like a minor miracle now. I have little doubt intel's netburst can achieve those speeds, if only they can control power consumption. Looks like 0.09 is a lot leakier than they thought, and they will need a SOI like solution urgently.

>Anyway, I think mass-availability of Scotty by 2004 is
>nothing new at all... It will still be launched this year.>

Define 'mass availability'. Everytime a new cpu is launched, it is only availble in relative small quantities for at least a full quarter, so in a way, yes prescott was not to reach mainstream quantities this year. However, I don't believe it will now reach mainstream quantities in Q104 either, looking at those roapmaps and reading all the leaks. If they would paper launch it this year, it would only help the intel apologists on forums like these to be able to claim "see, its not delayed !", but unless real quantities hit the channel, it won't do intel any good.

> After all, even A64 is only expected to be readily
>available in quantity in 2004... FX in particular

Once again depends on your definition of being "readily available". All my favorourite online resellers have both the FX and the A64 in stock, I would call that readily available, even if we are of course not talking millions of cpu's. Just enough to meet demand. I doubt the same could be said of prescott if it is indeed launched in november.

> I hope they don't run into too many problems on the 65nm
>front!

I doubt it. Intel will use the same equipment for .065 as for .09, just like they used the same equipment for .18 and .13. The second transistion usually goes much smoother than the first. BTW, AMD uses the same equipment for .13 and .09, (like they did for .25 and .18) so that may go relatively smooth, but .065 will probably be much more challenging for AMD.

My best guess is that 2004 will be a very good year for AMD, 2005 may be intel's come back year.

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[guesswhosback]

i dont think .09 will not be as hard as people think for AMD

after all they didnt leap first ... intel did... they no doubt will pick up some tips... plus i susect IBM will be aiding

 

[flamethrower205]

Pretty sure IBM will be helping out too, given what they have in their testing labs .

The one and only "Monstrous BULLgarian!"

 

[eden]

I meant if the Prescott is delayed, that means some features will likely be added, so whatever the features are, they will be a surprise, as in, mysterious new thing.

Didn't mean it would surprise me if they delayed it.

Only for you and the most faithfull intel apologists I think.

Judge not lest ye be judged, ok?

--
<A HREF="http://www.lochel.com/THGC/album.html" target="_new"><font color=blue><b>This just in, over 56 no-lifers have their pics up on THGC's Photo Album! </b></font color=blue></A> :lol:

 

[Guest]

>I meant if the Prescott is delayed, that means some
>features will likely be added,

You know how long it takes from design to tape out ? How long just the final verification of a cpu (re)design takes ? Creating mask sets ? A wafer production run ? <i>You just don't add features to a 100 million transistor design in the last minute</i>.

>Judge not lest ye be judged, ok?

I judge every day, and I am judged every day. Its my job to judge, and its my management's job to judge my judgement. My judgements or opinions may not always be spot on, my crystal ball isnt failproof, but they are pretty much always backed by facts, knowledge, reason and experience so I can defend them when I get called on them, and not just by a fuzzy warm feeling or some vague personal preference overshadowing any logic.

OTOH, your opinions and analysis here consistanly seem to lack any susbstantition, or factual background other than wishfull thinking. Call it what you want, clueless, fanboy, apologist,.. but sure, feel free to call me arrogant or something else if you like. See if I care.

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[eden]

You know how long it takes from design to tape out ? How long just the final verification of a cpu (re)design takes ? Creating mask sets ? A wafer production run ? You just don't add features to a 100 million transistor design in the last minute.

It takes ~3 months for chips to be developped. Add about 2 weeks delivery.
Now add about 3 months of feature implementation.
Therefore, it would not surprise me if it came out in Q1 04 with extra features or a very tweaked core for scalability.

--
<A HREF="http://www.lochel.com/THGC/album.html" target="_new"><font color=blue><b>This just in, over 56 no-lifers have their pics up on THGC's Photo Album! </b></font color=blue></A> :lol:

 

[Guest]

AFAIK, it takes around 3 months to produce a single cpu (from waferstart to working silicon). Take into account time designing, validating design, making test samples, validating test samples before beginning actual commercial production, and you are talking a multiple of that. Tejas is supposed to tape out around now or the next few monts, that basically means a feature freeze for a cpu not expected for another 18 months. No way they would quickly add 'features' to prescott now. If they where tweaking prescott's feature set, it should have happened in the beginning of this year or even earlier, and the delay would have been known for a loooong time (at least to intel).

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[lhgpoobaa]

I dont look forward to 120W CPU's... or even 100W ones. thats just rediculous.
Power sucking roomheaters.


<b>I am not a AMD fanboy.
I am not a Via fanboy.
I am not a ATI fanboy.
I AM a performance fanboy.
And a low price fanboy.
Regards,
Mr no integrity coward.</b>