AMD going to 9 micron?

[UHR]

I was browsing the AMD website today and I looked at their Roadmap. It shows that the first half of next year they are planning to go to 90nm with the Opteron and Athlon-64 chips...... I took a look around the web and found plenty of rumors about what the chips specs will be. I am just wondering if anyone out there (Maybe even Tom) could shed some light on this. With the Idea that they would be moving to yet another new process brings up alot of questions.

Will it require another Socket?
Is this the reason Why it has taken so long to release the New Opteron/A-64 chips? Was AMD attempting to get the R&D done fast enough (or hold out ling enough) to Make an immediate Jump instead of having to release a 13nm version of the Opteron/A-64?

Are the Rumors of using a DDRII Memory controller within the Chip True?

There are Many more Questions as well as Speculations to this..... My Main Question though is :
Is this Why AMD has been holding out for so long. Are they doing this as a Strategy against Intel?

One of the best ways to win is to allow them to get comfortable....Then Strike! We have Seen AMD do this before.... Will it be deja vu?

There is no Spoon....

 

[marneus]

another new socket, quite possibly not as the Athlon went from around .18 to .13 with the same socket, it will depend on the silicon & stuff like that... same goes for onboard ddr2 controller...



Hmmm, wonder if I can get a valid page fault ???
these invalid ones are far too commonplace...

 

[Mephistopheles]

No, that can´t be the reason why AMD is taking so long. That doesn´t make sense...

They´ve been overrating and overpricing their products, as if hampered by some engineering (like yield/stability or whatever) in their products...

The first A64 will not be built using 90nm. Nor will it incorporate a DDRII memory controller - far from it. The first A64, according to current official data, will just use one channel of DDR333/400.

One of the best ways to win is to allow them to get comfortable....Then Strike! We have Seen AMD do this before.... Will it be deja vu?

Intel doesn´t seem comfortable at all. Their newest chipsets are impressive, and they´ve strengthened their portfolio considerably with 800Mhz FSB and hyperthreading. If indeed they´re waiting for their pray to "be comfortable", it´s a very dumb move and they´re paying a hefty price for it. It´s a chance they wouldn´t lose, if they could. This is, of course, just my opinion... And given the characteristics of x86-64, they need to get that product out as soon as possible to get people to support their new extensions... or Prescott will eat them for lunch. At least now Opteron is on the move... hopefully.

 

[UHR]

I know that The First A-64 chips are not going to be 90nm chips or use DDRII. DDRII has a ways to go before it will be used beyond the Video Card Market. I am Just curious as to why a company like AMD would release a 90nm chip or even plan to release a chip so close to the release of the A-64?
As far as intel with the 800FSB? Now lets look this for a second. When Intel was caught with their pants down when the T-bird hit the Market, They Knew they were in trouble.
And thus sparked Their drive to Produce some great product.
AMD has now been around long enough to know the game and know how to play it. I highly doubt that AMD is in so much trouble that they have not come up with a plan of their own.
they have had nearly a year now to develop something that would be comparable. Intel has pretty much leveled off as far as releasing new tech. The last of which is 800FSB.... Now granted That kind of rig will scream compared to what is out there. But I do not think AMD is just sitting back ignoring what is being put out there.

But Really All I am asking is if anyone knows anything about the 90nm chips, some sort of solid evidence or something so I could better understand why AMD would do something as strange as attempting to put out a totally new chip so close to release the infamous A-64 series

There is no Spoon....

 

[tabytha77]

I don't think AMD will be going .09 micron anytime soon. They have trouble enough with .13 micron. Intel surely will win the .09 micron race with Prescott. AMD must have it in their plans though!

 

[rebturtle]

The first Opteron chips can already be bought! Check out prices <A HREF="http://www.newegg.com/app/ViewProduct.asp?catalog=343&DEPA=1&submit=property&mfrcode=0&propertycodevalue=4777" target="_new"> here</A>. I'd really like to think that AMD can do as much their new socket design (socket 970) as the last one, and that it will last for a long time before the next change in design. It was one of their best selling points against Intel, and they'd be very sorry to lose that edge.

<A HREF="http://rebturtle.com" target="_new">rebturtle</A>
<A HREF="http://www.geocities.com/rebturtle/TurtleTech.html" target="_new">My System</A>

 

[flashpan]

I couldn’t help but feel I might add my two cents on this matter……

Suppose that AMD might very well have nothing at all. Suppose that because of money or ideas or both AMD maybe spent. I know that this is going to be a very unpopular thought but please hear me out.

Let us take for a moment a couple of things that we know. First AMD is late. Opteron was late and it is anyone’s guess when we are going to see Athlon 64. (If you thing you know when it is coming or believe AMD, I hear the Brooklyn bridge is for sale.) No one knows why it’s so late. That it will first start on a .13 die and then move to a 90nm later. Many questions, no answers. From AMD smoke and mirrors

Second, AMD has had to make a 3-year deal with IBM to “co-develop” its next ideas. This in my mind is the biggie. The way I read it IBM is calling the shots. They say when a 90nm chip is coming and AMD seem to follow right behind. IBM says that the roadmap goes 65nm and 45nm on 300 mm wafers. AMD follows right along. So when I see a new AMD chip coming, I see an IBM.

For those na-sayers to this thought I ask you to think about this. Opteron is a bridge proc between 32 and 64 bit, basically to make Xeon look real bad (which I think it dose). Athlon 64 will be a desktop proc aimed for Intel’s Pentium line (Good luck, they’ll need it). Why is there no main stream “Enterprise” level chip to compete with the Itanium line of procs? I believe that this is because IBM wants an open field when they start to really push their own 900 line (true 64 bit) of processors.

I personally think only one of three things is the possible.

A. Athlon64 is so good, so state if the art, that it could wait over a year to release. That AMD has convinced enough software companies to move to a 64-bit platform all at once so as to cut off Intel’s market. And that they think that Prescott, Tejas and Nehalem are a myth. No way Intel will make 10ghz by Q4 2005. Intel is completely unprepared to counter such a thing that the AMD brass think they have it all under control.

B. IBM is their lifeline.

C. AMD PR rating system will save them! And God will smite those Yankee infidel dogs of the great satan Intel (ha ha just kidding. I hope)

FYI: Yes. I am pro Intel. I’m form Portland OR. Intel’s second home. Lots of friends that work there.

 

[rcj187]

the best information i have is from and amd techie.. so here it is:
1: A64 WILL be released on time with current roadmaps (september)
2: A64 will release at 2.0GHz to 2.2GHz
3: A64 will be competitive with prescott (he had a very smug grin on his face when he said that so im guessing its more than competitive..)

 

[Mephistopheles]

Just curious here, but how can he even be so sure that it´ll compete against prescott? Does he have insider details on prescott to compare A64 to the prescott p4s?...
Or vice-versa, for that matter: do those two companies actually <i>know</i> what they´ll be up against in those cases?...

(best case scenario: they don´t)

 

[flashpan]

September huh? And your bid on the bridge?

Honestly the problem behind the Prescott/A-64 question is Intel in their own cool sort of way. Folks (Intel folks) that I know that have been working with it for almost 2 years and say it’s a wiz bang chip, cool as all get out, but won’t utter a peep other wise. Intel has that employee fear thing down. AS far as the company line goes they share almost nothing with the general public until the last minute. Even after release don’t tell us everything about the chip i.e. Hyper-threading.

I guess it’s a difference in company style. Intel says “Here you go, new chip, it’ll work for ya”. AMD says “With our own testing we have the fastest chip in the world! We’ve taken the competition completely by surprise. However we will be putting off its release date another quarter. All hale Big Blue!

(and with my own smug grin)
Do you really think Intel doesn’t have it’s own 64 bit desktop solution? After as much time as they have had to prepare for it?

 

[reever2]

why would Intel have a 64-bit dekstop chip? They have been saying for a long time that 64-bits isnt needed right now, Intel is not a company to say one thing and do another, where would a 64-bit desktop chip go in between Xeon, prescott, celeron, and Itanium, wherever it goes it would obliterate an already existing successful product line

 

[flashpan]

I’m assuming that you understand the concept of a free market. If the consumer goes ga ga for a 64- bit chip and your competitor says their making one right now and says they’re going to give it to then soon (give or take four quarters) you don’t just sit there and say well you really don’t need it right now. Intel isn’t stupid nor do they easily forget it when they take it in the teeth (the 1 GHz fiasco) They also are very aware of the IBM issue.

Just because Intel hasn’t said it’s working on a 64-bit desktop chip doesn’t mean they’re not.

But just for the sake of argument lets just say you’re right. Intel may not be looking at a 64-bit option for a few reasons.
1. They think that AMD won’t be able to deal with a 10GHz chip by Q4 2005. Just out run em.
2. The cost of the rest of a 64-bit system may be cost imperative to the end consumer. I here that the 64-bit Windows app will need a Gig of memory just to run out of the box. Although this may have been a server OS I don’t remember.
3. The gamer factor. While gamers will go to great lengths to acquire the latest and greatest at any cost. This doesn’t mean that the programmers of those games wouldn’t automatically switch over. Game code writers in the past have been one of the last groups of people to move to new technology. Hence if the game isn’t 64-bit why have a 64-bit box.

I dunno…chew on that for a while

 

[Mephistopheles]

Yes, there´s the 1Ghz fiasco... It´s been some time, though. Intel has played some very interesting cards...

For example, one nice day, one of their lab technicians (or more likely, one of their executives) decided it might just be a good idea to reveal their hidden SMT technologies on desktop processors...

And I can still remember the feeling I had at New year´s... As I remember, Canterwood and Springdale were "simply" 667Mhz FSB and dual-DDR333 parts... Then someone said, as if that was no big deal, <i>"hm... nonono, no, wait, make that 800Mhz FSB and dual DDR400"</i> As easy as that... And the damned thingies were still on schedule and produced the P4 "C" line we´ve just met - which is, as far as I´m concerned, one of the most interesting CPU families to come to the market for a loooong time. And how can anyone say these people aren´t careful enough to have 64-bit desktop plans of their own, just in any case? They´ve already got 64-bit experience... And they´ve got resources. It would be very naive to think they won´t be prepared to go that way...

I wonder what is actually on the prescott die, after all? Lots of hidden cards for Intel to play... if our assumptions are true. <i>(conspiracy theory #3: Intel´s got a "world domination" chip feature that just needs activating that noone knows about...)</i>

And one more note about the 1Ghz-Fiasco, I find that the Palomino thermal fiasco is much worse. Intel actually noted that there was something wrong and it took them some time to figure it out, but Palomino came without any thermal protection whatsoever (while the P4 always had a rather sophisticated thermal protection logic built-in) - and that wasn´t really called "a flaw" by AMD itself and was never corrected until the next generation of CPUs from AMD came. AMD processors still don´t have built-in thermal protection. The motherboard manufacturers have to worry about that. At least it isn´t the consumer anymore.

 

[papasmurf]

they aren't going to .9 so soon, they just switched to .13 at considerable cost mind you and their cpus have only gotten a minimal clock boost from that. They can't afford another long readjustment process like that, low yields would kill them, their prices are already very high.

Treat your body like a $600 car. God didn't intend it to last so use it. Run it into the ground!

 

[papasmurf]

they aren't going to .9 so soon, they just switched to .13 at considerable cost mind you and their cpus have only gotten a minimal clock boost from that. They can't afford another long readjustment process like that, low yields would kill them, their prices are already very high.

Treat your body like a $600 car. God didn't intend it to last so use it. Run it into the ground!

 

[HardWareBoss]

You some it up in a nut shell. They way AMD is going they will not have 90nm till the end of next year. By then Intel will be working on 60nm or will it be 45nm