AMD .13ì Problems?

[WizardRGN]

Rat over at X-Bit labs has pointed out some very genuine concerns about AMD's ability to migrate Desktop CPUs to a .13ì core. Rat points out that the currently released AthlonXP Mobile CPUs based on the Thoroughbred core show some very odd and troubleing issues such as low clock frequencies high voltage usage and a few other concerns.

Do you feel that AMD will be able to migrate the Desktop AthlonXP CPUs to .13ì with the same success that they had with the Palomino core?

Read the article from X-Bit labs here: Article on AMD .13ì Problems

 

[aka1nas]

<LI>The first and the most suspicious thing is the extremely low working frequencies of Athlon XP processors built on 0.13micron core. Look here: the fastest model in the today?s Athlon XP family makes 1.73GHz, and later on Athlon XP 2200+ is to come, with the core clock of 1.8GHz. And the first 0.13micron Athlon XP should also work at 1.8GHz core clock! Moreover, by the end of the year its frequency will get only 266MHz higher (Athlon XP 2600+ will work at 2.06GHz actual clock frequency). For a better comparison try to recall 0.13micron Celeron which frequency grew almost 1.5 times higher when they shifted from 0.18micron to 0.13micron and the Vcore got considerably lower at the same time: from 1.7V to 1.5V. And here comes the second reason...
<LI>The second thing witnessing that AMD has some problems with 0.13micron manufacturing technology is a too high processor Vcore. As is known, it makes 1.65V, which is only 0.1V lower than that of 0.18micron Palomino core. Of course, it wouldn?t be quite correct to compare it with Intel?s manufacturing technology, however, together with all other evidence, high Vcore signals about not very high chip yields. By the way, the relative rawness of the manufacturing technology is a very poor excuse: the Vcore of the first 0.13micron Celeron CPUs was 1.475V and the chips could work at 1.4-1.5GHz core clock (that is almost 1.5times higher than the core clock of 0.18micron processors).
<LI>And now let?s have a look at the yesterday?s announcement of the mobile Thoroughbred (see this news story). AMD started shipping 1400+ and 1500+ processors working at 1.26GHz and 1.33GHz correspondingly. Again, the clock frequencies are very low (even taking into account that these are mobile solutions).

The initial yields will of course be low for a new core and especially for a new process. This is why we are only just now seeing T-Bred mobile offerings and at relatively low speeds. This also explains Vcore etc. As far as the first .13 Celerys performing better than T-Bred, that is an entirely different chip and architecture. A Celeron is a much simpler chip than an Athlon. More complexity = more steps in "debugging" the CPU design and more steppings until performance ramps fully.

And the last thing. According to a new AMD roadmap (see this news story), 0.13micron Barton core has changed. Now it will be made not from SOI substrates (which used to be a direct mention of the joint AMD-Motorola technology) but from regular substrates of the silicon dioxide (UMC technologies do not know to use SOI). Here we can?t help recalling that in the end of Q4 UMC was going to start 0.13micron CPUs for AMD. No wonder that Barton core has been changed: it is most likely to be now oriented for UMC?s production lines. And these production lines seem to be pretty good, we should say, as AMD has increased Barton?s cache recently. I think that UMC has already made some trial Thoroughbred processors. And the outcome may have turned out so good that AMD decided to double the L2 cache size (up to 512KB).

They dumped SOI because it has proven to be very costly to implement, which is why Intel seems to not be interested in it. It doesn't make sense to spend more money to produce what is going to be a value processor line for AMD. The reduced cost of wafers and other materials will mean they can sell Bartons for less and make more profit, and hopefully make their CPU's more desirable to OEMs.

 

[DDad]

I have no way of knowing this for sure, just going with my "gut" thought. Is it possible that AMD is throwing more resources towards Hammer, and setting things up so that the Dresden plant will be ready to go 100% on that- transferring all Thoroughbred/Barton to UMC?

 

[lessell]

The production ramp for the tbred is not differing all that much from the intro of the palamino, everyone seems to forget the amount of testing and validation that a new chip must undergo from both mb manufacturers and oem companies before the release of a new chip. AMD has also announced that Dresden will fab only the hammer series of chips so they are actually ramping production at a new shared fab for the tbred which makes things even more interesting. The SOI process will not be used for Barton because it will be produced at the new shared fab by UMC. It makes no sense to spend the capital investment for the technology there to produce a processor that will be relatively short lived and targeted for the value market in less than seven months. Laptop processor speeds always lag behind desktop speeds and their designs are a result of differant concerns and needs. The truth is the introduction of tbred and barton are a short term project needed as a market bridge and are a secondary priority, AMD sold a record number of processors in the first quarter, have had no supply problems, and will probably exceed those sales this quarter whether they introduce tbred or not. It really won't impact their volume until quarter 3 and so it really doesn't matter if they introduce it April, May, or early June, what is important that at introduction the platforms are ready for the transition and the supply pipeline is full. Actually considering the major transitions AMD is making will make the balance of this year it is really remarkable that they are anywhere near their roadmap projections

 

[Mikewarrior2]

ouch, even the mobile Tbred chips are at 1.65V?

That does NOT bode well for desktop Tbred chips. that means a very, very high wattage/sq mm for future tbred's.




Mike

 

[pm]

<< ...but from regular substrates of the silicon dioxide. >>

This is a mistake, by the way, regular substrates ("bulk" substrates) are made from silicon, not silicon dioxide. If they were made of silicon dioxide they would be some form of SOI since SiO2 is the typical insulator used in SOI.

 

[BD231]

They must be having trouble lowering the voltage because of the high operations per second, I can't believe that the vCore is still above 1.6. The high voltage is what will keep the T-Breds running hot. Was the T-Bird on a 25m process?, I remember when I upgraded to a 1800+ from T-Bird 1.33 and my temps went down a good 6 to 8 C. I guess we wont be seeing anything near that with the 13m process?

 

[Mikewarrior2]

not only is it over 1.6V (its 1.65), and this is on lower-clocked mobile chips that typically have lower voltage than desktop chips. So, it doesn't look too good for tbreds to be cool running chips in a desktop setup.


Mike

P.S. Athlon XP and T-bird temps cannot be compared, even on the same motherboard, due to different CPU pcb's. If you compare actual CPU wattage, with a top notch heatsink, an XP1600+ versus T-bird 1.4ghz is no more than 3-4C in DIE temp change (other factors play a role, too, but an 8C+ Drop in DIE temp isn't feasible, save with inaccurate temperature monitoring).

 

[WizardRGN]

Definately the Voltage issue and the clock speed issue put a negative outlook on AMD's attempt @ .13 microns

I think the worst thing right now about AMD is how incredibly stupid Sanders is acting.

First off was that terrible witness stand for MS
AMD's Sanders knows nothing about anything, claim States
Then he's being sued for using racist slang and descrimination
Senior AMD executive claims Sanders, Ruiz humiliated him

This is a guy that doesn't know how to retire with style. Here's to hoping that AMD gets the Hammer working before Sanders is completely out.

 

[Mingon]

somehow I doubt the mobile version @0.13 micron will be 1.65vcore, the current athlon4 works at 1.4v.

 

[Degenerate]

hmm.... They will also need something faster to compete with P4 Ms. I wonder how the power useage compaers with P4M and P3M?

 

[Mingon]

The athlon 4's are very competitive e.g. the 1300 at 1.4 vcore according to amd's figures produces 25 watts of heat, so you would hope the 0.13 athlon will improve on this.

 

[Mikewarrior2]

Mingon,

One would think that Tbred mobile's would be low Vcore, but the truth is, they are not at this point. Not only that, they aren't much of a MHZ improvement over current mobile Athlon Chips.

There are NO voltage numbers available for Tbred Mobile chips at AMD.com. Very curious. Going by the xbit lab's numbers, this explains why there are only low-speed Tbred mobile's at this point.



Mike

 

[Mingon]

I just cant understand how they would have a higher vcore, it would give no advantage (except lower cost) than the palimino. Personally I cant understand why all the samples of t'bred that have been seen have not had the same square die.

 

[BD231]

I just cant understand how they would have a higher vcore, it would give no advantage (except lower cost) than the palimino. Personally I cant understand why all the samples of t'bred that have been seen have not had the same square die.

AMD proc's have always run well undervolted for me, which must mean AMD is just playing it safe.

 

[aka1nas]

The first batch is of course gonna suck with a new process. AMD just prob decided to still make money of these chips by selling them at a lower speed grade. Remember, we probably won't be able to actually buy these processors in a desktop form factor for another 2 months. Thats a long time to get the kinks out in that industry.

 

[Mikewarrior2]

The problem is, if lower end (1.33ghz and lower Tbreds) are already running at 1.65,

There is virtually no chance of desktop chips running at lower than 1.65V. Especially clocked at 1.8ghz and up. Mobile chips are almost always lower voltage than desktop chips.

There isn't a problem with a 1.65V 80sq mm Tbred except that it is 1.6X hotter in terms of watts/sq mm versus a comparable Xp chip.


Mike

 

[KF]

>The first and the most suspicious thing ...

It may seem odd, but really it is the norm to have problems in a new process. Intel's Celeron intros are really the oddities. In case no one noticed <g>, AMDs Athlon chips have always run hot in comparison to PII/PIIIs. AMD needed the speed edge to play in Intel's league, and high current is a proven way to get speed. AMD just built the chips to take the heat, like all the expensive, low-volume RISC processors always have. Intel is the company that worries about hot chips. Intel tries to gain speed by running cool. That's the hard way to do it, and that is why Intels chips are so spectacularly overclockable. You have to refrigerate AMD chips to get comparable over-clockability.

Introing a new core first in the mobile segment is obviously done because the volume in that market is so low and the price premium so high. That means the yeilds can be low while the process gets debugged, and the margins are high enough that you can afford to make mistakes.

AMD is playing out of its league, and it remains to be seen if it can survive this round of competition. Intel just barely missed crushing AMD out of existence a few years ago, and is intent on not making that mistake again. Intel must have at least 10-20 times the production capability of AMD and a similar depth in engineering. AMD pretty much has to use process technology proven out by others, while Intel has the ability to originate its own.

Intros have been pushed back because the economy has been slow and semi-conductors are in oversupply. while demand for high-end is low. But this gives engineers time to get things right before they are put into production.

Two years ago, one would have had to say AMDs survival to this day was unlikely.

 

[Degenerate]

KF, Both companies worry about heat problems. Both Cpu's are capable of running at high temps. INtel has better hat management. Both Cpu's over clock just as well - P3 and Athlons

 

[Mikewarrior2]

.13 micron intel chips likely will overclock better than tbreds, if for no other reason the drastically higher Wattage/SQ mm numbers from the new Tbred chips. Which puts a huge demand on a quality heatsink and paste



Mike

 

[Mingon]

Its a wierd situation, look at the athlon versus pIII, the athlon managed 1.73 ghz on 0.18 whereas the pIII topped at 1ghz. And yet the 0.13 PIII hits 1.4ghz with ease and would have gone onto around 1.8, so you would think that the athlon should hit some high speeds. I know the PIII-m's have an intergrated heat spreader but I am not convinced this helps that much.

 

[ST4RCUTTER]

I remember when speculation was running rampant about the Palomino.

Rumors were hopping in these forums about SSE2, integrated heat spreaders, on-die thermal transistors, 1MB of L2 cache, double-pumped registers...pretty much anything new in the field of processing technology was at one time or another going to be part of Palomino. I regard this opinion based on unofficial information in the same light as those earlier post on the Palomino. The truth is that no one really knows, with any degree of certainty, the specifications of the Thoroughbred. I will reserve judgment until then...May 2nd can't come soon enough.

 

[Mikewarrior2]

Except that VOLTAGE information is derived from MULTIPLE sources.

ZDnet Germany posted 1.65 in an article several months ago. If it was too high, I highly doubt AMD would stand pat while this number was floating around. They made a huge deal about Palomino's being far cooler than T-birds, which was false. The end result of the palamino line is ~12% cooler than Tbirds.

The general voltage numbers vary from 1.6 to 1.65. Even at 1.6, its not enough of a voltage drop to not drastically increase the Wattage/SQ mm. That isn't speculation, its truth. They'd need to drop to 1.4 or lower to match current Wattage/Sq mm.

Muropaketti. These guys had a Tbred sample. 1.65 is teh voltage they reported.

One thing you'd expect, if indeed Tbred was going to be soo much cooler than Palomino, that AMD PR would be all over the place bragging up numbers about it (like the 20% from the Palomino's). Yet, AMD is silent regarding Tbred heat advantages.

EVEN on the Tbred Mobile press release, amd said nothing regarding heat. Really interesting....


Mike

 

[Skawttey]

What's May 2nd?

 

[WizardRGN]

The Heat issue is very interesting. AMD has been having problems with .13micron since they first started developing it. TSMC could not provide a stable .13micron CPU Manufacturing process also other problems have slowed down the Thoroughbred reaching market.

Rasing voltage is often used to create a more stable CPU, the same way raising voltage helps with OCing. Seems as though AMD will have a bit of a problem scaling Thoroughbred if they are already at 1.65v. Further more to those who are saying "this is the first batch and shouldn't be expected to be perfect" I should remind you that these are mobile CPUs which require lower energy use as well as lower temps because of the cramped enviorment in the laptop. If anything the current Thoroughbreds are probably as cool and as low voltage as AMD is able to make.