10.15.2007

The Fundamental Law of Progess: NEW > OLD

We once claimed that major OEMs are ignoring Barcelona and the report today from ZDNET confirms this to be true:
"AMD may have a bigger problem since you can’t even order these systems from any of the major server makers and none of them can tell you when their Barcelona servers will be available."

None of the major OEMs seemed to be rushing out with their new K10 systems. The fact that the Big 5 cannot commit on server availability can mean several things mostly which can only be bad for AMD. Normally when you have a ground breaking product every OEM wants to be the first to market. But when AMD's new product violates the fundamental law of progress that demands all NEW PRODUCTS MUST BE BETTER THAN OLD PRODUCTS, it’s no surprise many are willing to wait, if not too afraid to try.

It is increasingly alarming how AMD seemed to be moving backwards. We've already given up hope that AMD would once again beat the competition (Intel, Nvidia) but at the very least we expect them to beat their older generation products. The introduction of slower 65nm CPUs and now K10's inferior performance to K8 are just some of the bad habits AMD seemed to be developing.

The article also tackles the question about Tigerton’s availability. The result of George's investigation showed that anyone can order Intel’s new server MP but with a product shipment of within a month or so. Whereas compared to Barcelona, orders are not allowed without even a promise when one can order. In the business world there is a big difference between an order taken and an order declined. Guess which one involves making money.

30 comments:

Tonus said...

It seems to be less a problem of Barcelona being worse than previous CPUs, and more about Barcelona not being available. Which is a problem AMD has struggled with often in the past.

Anonymous said...

The channel seems to be under the impression that the Tier 1 OEMs are all getting the parts, as availablility there is scant... only one or two of AMD's closest channel partners appear to be reporting getting any product.

http://www.crn.com/white-box/202402138

SPARKS said...

Well, Doc, this report of yours coincides exactly with what happened in the market today. On the whole, the market pretty much took a bath, 10-15-07.

These market corrections and fluctuations have been occurring for the past few weeks. The Feds announcements, housing starts, mortgage problems, overzealous semiconductors downgrades have contributed to this wild ride. Intel, for the most part, as a big player, rides the trends, typically.

Today, however, was a big exception. Where the DOW dropped 108, and NASDAQ dumped 25, INTC moved UP 20 cents. Even pipsqueak AMD gained a couple of cents. (This contrary to what those morons were saying last week by flatly downgrading the entire semiconductor sector.)

AMD, generally, rides the coattails of INTC, as their volume in sales are tied together by total market volume. This, however, will change very soon. AMD’s Barcelona failure has not reached the numbers crunchers. AMD’s expectation’s, or lack of, hasn’t been factored in. The demand is there, as you wrote, but fulfilling that demand hasn’t affected the bean counters numbers, yet.

The key to this mess is comparative trend lines. Look for INTC pulling away from AMD in the coming months, as it has been all year long. I predict, as the market comes to realize AMD cannot fulfill its demand to ALL its customers, in conjunction with INTC’s powerful execution at 45nM, and as they factor in Pheromones lackluster performance, with more market share loss, the trend line separation will get much wider. How long can most wait and others denied?

GURU’s, In The Know’s and Jumping Jack’s, excellent, and invaluable technical analysis on the previous page is one facet of an emerging 2nd “Perfect Storm” for AMD. They are bogged down in the mud with SOI, CTI at 65nM, at 2.6, with cripple core. Your report substantiates the volume production and speed problems that AMD is faced with, as of late. Further, a fourth consecutive quarterly loss, as expected, will only make a bad picture even worse. There is simply NO money coming in, now, or in the foreseeable future.

What is extremely clear, this is a very, very ugly year for AMD. I don’t see it getting any better next year, not with this report confirming multi faceted problems.

Then there is the Nehalem, Nvidia 2008 Tsunami’s off on the horizon.

SPARKS

sparks said...

And more.....


http://www.fudzilla.com/index.php?
option=com_content&task
=view&id=3627&Itemid=1

SPARKS

pointer said...

George Ou is not the only one noticing it. Some one in the amdzone already grunted about it and likely go for Intel's Clovertown offering instead of AMD's Barcelona because of that.

http://www.amdzone.com/index.php?name=PNphpBB2&file=viewtopic&t=11326&sid=046cfda6cd589c17e9aa2d09950068e1

InTheKnow said...

Sparks, thanks for the overclocker's link (and the article it linked to).

One interesting thing is that the authors identified 3 main opportunities early in the design phase: "1) targeting of transistor and interconnect characteristics, 2) definition of layout design rules, and 3) definition and development of extrapolatory process models used by design during subsequent technology phases."

They then proceeded to focus on items 2 and 3 as "they play the most critical roles in creating a robust DFM strategy.

I found this interesting because item 1 sounds to me like surprisingly like CTI.

So CTI concerns are deferred until development, and I think rightly so. Unanticipated process interactions are to be expected during this phase of the products life (and I'm certain Intel sees this too).

The issue, as Guru has pointed out, is what does your designer target for transistor performance. And what if the designers assume a mature transistor and the process doesn't deliver it? My guess is an under performing device. Which kind of sounds like Barcelona.

So I find myself wondering, why does AMD's focus seem to be on Brisbane improvements? Wouldn't it make sense to try and crank up performance on Barcelona instead?

I'm thinking that they might be focusing their 65nm CTI efforts on an established architecture (Brisbane) to reduce the number of variables they are playing with. If that is the case I suspect they will be able to apply what they learn to Barcelona fairly quickly. They will still find some new surprises I'm sure, but far fewer than if they were to try and do CTI on a new architecture.

InTheKnow said...

Scientia's blog seems to be a bit slow lately. I wonder why? :)

But that didn't keep Abinstein from failing his spelling test. He started out well but faded down the stretch. It went something like this...

I - good so far
n - still looking good
t - three for three
e - um shouldn't that be an h?
l - this should be that pesky e
- a blank?
f - okay we've completely lost it
a - getting worse
n - and worse
b - and worse
o - and worse
i - and worse

nope, that just doesn't spell intheknow. Nice try though. :)

Tonus said...

Pointer:
Some one in the amdzone already grunted about it and likely go for Intel's Clovertown offering instead of AMD's Barcelona because of that.

And he is being tarred and feathered for it. Some people just don't get it.

pointer said...

Tonus said...

Pointer:
Some one in the amdzone already grunted about it and likely go for Intel's Clovertown offering instead of AMD's Barcelona because of that.

And he is being tarred and feathered for it. Some people just don't get it.


when I read it yesterday, the threads there were still well behaved. Then you commented Gilboa being attacked and I actually thought you some how read it wrong it. Then i go and read the amdzone thread again and realize there is an idiot there that most of you all know of, was attacking him, accusing the poster being intel_fanboy under cover and use his favorite terms again shooting the poster for being posting FUDs ... hahahaha

and the funniest thing is that the poster is a real AMD supporter ... hahahah

Anonymous said...

Check out this thread

http://www.amdzone.com/index.php?name=PNphpBB2&file=viewtopic&t=11326&start=30&sid=32a74cf5a2483f419c6306628a3bfb7c

abinstein, the impartial, insults gilboa, one of AMDZone's members.

How lame can he be?

Anonymous said...

AMD is f@*&$(@

http://arstechnica.com/journals/hardware.ars/2007/10/16/amd-reportedly-planning-triple-launches-in-november

Tonus said...

Pointer:
and the funniest thing is that the poster is a real AMD supporter ... hahahah

Yeah, now the other long-time members there are defending him, and the guy ripped abinstein a new asshole as well.

Man, when AMD fans are calling you an AMD fanboi, you know you have gone off the deep end.

Anonymous said...

Looks like Abistein, the AMDZONE noob has been put right back in his box:

Abistein said to Gilboa: "Maybe AMD is facing a very tiny up-hill battle precisely because there are companies like yours and stupid moronic people like you making FUDs about its products."

Mechanix replied: " [gilboa] is one of the most cherished forum members, simply states his frustration of the unavailability of his favored platform, and he gets mercilessly attacked.."

The Ghost said: "Gil is a good man and a good AMD supporter, but, and it is a big BUT, his job as anyone elses comes first ..

Gilboa replies: "I'll make it brief. (Wasting time on idiots is not my favorite past-time)
I've been a member of AMDZone since July 1999...


uh-oh.

When did Abistein join AMDFORUM?
June 2005.
PWNAGE!

Roborat, Ph.D said...

So Abistein is now claiming that the reason why AMD is having problems with 65nm and K10 is because honest AMD supporters are ranting?

Sounds similar to quotes saying that "rampant adultery is the cause of earthquakes" or "liberal women and pagans were the cause of 9/11". Interesting.

Giant said...

This is just getting hilarious:

http://www.theinquirer.net/gb/inquirer/news/2007/10/16/idf-taipei-nehalem-real-big

AMD hasn't even caught up with Clovertown or Kentsfield yet and Intel has Yorkfield and Harpertown coming out next month. These CPUs will scale up-to speeds of ~4GHz.

But the real killer is Nehalem.


These - obviously - are fast parts and, with a few more steppings coming along over the next six months, Intel got that frequency headroom to something like 4GHz under normal operating conditions, something confirmed to us this morning by a very large Taiwan OEM.

So, how much will the Nehalem be faster than the already oh-so-fast Penryn, focusing specifically on the core IPC single thread performance? I asked Kirk Skaugen, Intel's Digital Enterprise Group VP and GM of Server Platforms Group.

Kirk is always friendly, even to the point of having had a fun chat during the previous Taipei IDF on how great it would have been if Nvidia actually did come out with a dual socket Xeon Nforce SLI chipset - which it didn't up to now - so I expected a fun answer this time too.

Well, no fun here - he didn't want to comment on the CPU core-specific performance expectations beyond the well known integrated memory controllers and interconnects. But he did say that the CPU core performance jump from the same process Core 2 (Penryn) to Nehalem would be higher than the jump for Netburst to Core 2 itself.

Since that last jump was pretty substantial - big enough to turn around the market situation and put AMD in a lot of further trouble, I'm curious how far would Nehalem go then.


This will be insane. Penryn -> Nehalem will be a bigger leap than Netburst -> Conroe!

I previously thought that Intel would not make significant changes to the core outside of SMT support - I was very wrong!

It goes without saying that AMD is royally screwed now.

JumpingJack said...

So Abistein is now claiming that the reason why AMD is having problems with 65nm and K10 is because honest AMD supporters are ranting?

Sounds similar to quotes saying that "rampant adultery is the cause of earthquakes" or "liberal women and pagans were the cause of 9/11". Interesting.


lol ... where did you read this? This is too funny.

Tonus said...

Jack,

See the link in Pointer's first post here.

re: early Nehalem claims - I try not to get too excited over rumor-mill comments, especially early ones like that. But if Nehalem can indeed scale to 4GHz (non-OC) and is showing a significant performance leap per-clock, that sounds pretty damned good.

I don't bother overclocking these days, but a CPU like that would make me want to try it again...

Anonymous said...

And the hammer falls: Intel Q3 earnings

Revenue $10.1 billion, up 15% year-over-year
Operating income $2.2 billion, up 64% year-over-year
Record microprocessor, chipset and flash unit shipments
Net income $1.9 Billion
EPS 31 cents

Anonymous said...

Earnings are out and it looks as though the Q2'08 BK prediction may not happen (shockingly enough)

Revenue +0.5Bil over expectations (10.1 vs 9.6B expected), gross margins up to 52.7%, EPS @$0.31/share vs expectation of 0.30, net income was 1.9Bil...

and the kicker.... gross margin for Q4 expected to go to 57% (man this price war is absolutely killing Intel's margin's) and expected Q4 revenue was forecasted at 10.5-11.1 Bil. The AVERAGE analyst revenue expectation was 10.4Bil.

Intel indicated strong demand in Q3 and it will be interesting if AMD's #'s reflect this too.

Anonymous said...

"Let's say that Barcelona is running about 5% of volume right now. I think AMD is doing about 1.6 Million chips a week. That would be about 80K Barcelonas a week. From what we've seen, the channel would soak up 80K a week without ever noticing."

Why Scientia should not be writing a blog -> Barcelona = server chip.... AMD's weekly production? Is that all server chips or I don't know >90% (unit volume) desktop and mobile?

5% volume would mean almost EVERY server chip AMD is producing right now is Barcelona? Anyone want to take any bets on the probability of that?

To put this in perspective, 80K chips per week would be ~1million over the next quarter (it would presumably be higher than 1 million in Dementia's world as Barcelona should still be ramping up, no?)

1mil @ at an average ASP of say 350 would be 0.35Bil on next quarter's revenue (ASP could actually be higher for Barcy if you factor in some of the 8XXX pricing). AMD's TOTAL CPU REVENUE in Q2 was ~1.4Bil, so now he thinks Barcelona will account for ~30% (or more) of AMD's CPU revenues in Q4? Did not the CEO himself say Barcelona will have a negligible impact on 2007 revenue? Does 30% sound negligible to you?

Of course the fanboys will look at his statement as gospel and think, wow, that's pretty good...ignorance is bliss, eh? The numbers all seem so reasonable, but the assumptions are TERRIBLE and unsupported (specifically the 5% #)!

JumpingJack said...

Earnings are out and it looks as though the Q2'08 BK prediction may not happen (shockingly enough)

Wholly cow, they blew it out of the water this quarter... it would appear that strength in the industry played right into Intel's hands... this will likely bouy AMD as well and Q3 will not be as bad as some are expecting.... however, if it comes in worse than expected I am wishing I would have shorted the stock.

:)

Anonymous said...

"Since NewEgg has a lower customer priority than, say, HP or IBM we can safely say that these vendors also have stock. If HP has stock but isn't shipping then the issue is not availability from a soft launch as George implies. The issue is certification."

More FLAWED Dementia logic... perhaps the large OEM's like HP, IBM, etc. have already pre-sold most of their initial stock to preferred customers. Some average Joe calling in will likely not get the same crack at new products as say a fortune 500 company buying bulk.

Again Scientia has taken observations he has made and is trying to conrom them to some preconceived solution. (i.e AMD doesn't do soft launched --> must be lots of supply --> must mean lots of demand --> Newegg has some product so availability isn't an issue).

How many flaws are there in this logic train? (Rhetorical question)

Supply is SHORT, triple core announcement out of the blue (this was NEVER on any roadmaps) means what? 65nm process is doing what? quad clockspeed is what vs initial roadmap? Let's see shortage of quad core chips --> let's make some cheaper tri-cores (mind you we are not having yield issues, customers are DEMANDING tri-cores) --> who wants to sell them expensive quads anyway?

This is rather pitiful - I'm all for being a fan, but there has to be a point at which you say I can't spin some of this data with a straight face. AMD Q3 earnings will be out later this week, and I'm certain Dementia will slurp up the forward looking statements, ignore the losses and underlying financial technicals and write a blog like "Q3 earnings a good start", "2008 preview", "Intel's 2008 processors: Mist in the morning sun" (he already wrote that article for 2007, so he simply needs to change the dates!)

Anonymous said...

"So, 2007 is now shaping up to be pretty much even in terms of performance between Intel and AMD"

"And, 2008 shouldn't be much different since IBM/AMD also have high-K and will use it on 45nm just as Intel will."

(Intel = 2007, AMD = 2009, maybe 2010?)

"This puts AMD in a much stronger position in 2007 than it was in 2006 and it gains a new architecture on top of that. This makes for a very strong offering by AMD in 2007. However, if we take Intel's bag of tricks for 2007, turn it upside down, and shake it hard, not a lot falls out."

"They will introduce K8L Barcelona in Q2 which should make 2007 K8L's year in the same way that 2006 was for C2D. AMD's margins should be up in Q1 and their channel distribution should be fine."

"Perhaps because of the lack of gains by Intel there has been a flurry of negative rumors about AMD lately. This includes ideas that AMD is behind in 65nm, is having problems with 65nm, and won't deliver K8L on the desktop until Q4 07."

"The earliest that it looks like Intel could release CSI is early 2009."

Just a reminder of the "analysis" being done - don't confuse 'hope' and 'wishful thinking' with analysis - especially there is no underlying funadamental knowledge supporting these, except for web rumors and extrapolation of past performance to future performance.

In virtually every AMD prediction best case is used, as well as coneviently overlooking fine details like mixing in production or shipping with actual product availability.

In most Intel predictions a conservative case is used and the standard is significantly different --> well Intel won't have any VOLUME on 45nm until H1'08 so we'll compare that to AMD's "starting production" (reads no product available) in mid-2008.

The "waiting for the wind to blow" blog is simply an inability to do more pro-AMD spin and an inability to put together good #'s for the K10 savior... You know it's just a benchmarking issue by those paid Intel pumper websites.... no wait it's the compiler... no wait it's that people don't understand the architectural details... no wait it's the low clockspeeds confusing the #'s... no wait...

Anonymous said...

"Wholly cow, they blew it out of the water this quarter... "

JJ - the Q4 forecasts are the key (look at the stock afterhours). Intel's bottom range of the Q4 revenue (10.5B) was higher then the AVERAGE analyst projected revenue (10.4B). Also keep in mind that Intel is generally conservative with these ranges and will likely either tighten the range later in the quarter or shift it up.

Also look at the Q4 projected margins while they are ramping 45nm equipment in the fabs... 57% is just about the good old days and a 1% increase in the overall 2007 margin this late in the year?

The only question now is will the stock peak somewhere mid-quarter with these high expectations or is it best to sell just prior or after Q4 earnings release in January?

Don't short AMD - if nothing else continued buyout rumors will continue to prop the stock up. Also there are now plenty of institutions with convertibles who have a vested interest in getting that stock price up. Finally, the overall strength in the PC market will likely cascade down to AMD too (as you indicated). It won't be pretty for AMD, but it will likely be "less ugly".

If you are looking for stock speculation, buy VMWare or some Internet high flier like Baidu, Google, et... or move into industries that are absolutely killing right now (dry bulk shipping, energy/oil)

When I'm not doing process technology I dabble in the market...

-GURU

JumpingJack said...

"JJ - the Q4 forecasts are the key (look at the stock afterhours). Intel's bottom range of the Q4 revenue (10.5B) was higher then the AVERAGE analyst projected revenue (10.4B)."

Yeah, I know... I did not mention that, this is a particularly strong guidance, and seems to break Intel's trend of erroring on the conservative side (unless this is actually conservative, then Q4 will be an industry wide record quarter).

Jack

SPARKS said...

Remember when I said “back to the traditional 10B earnings”?

10.1B


Remember when I said AMD’s Barcelona will be a non event?

Cam someone please pass the chips?



Remember when I said “INTC stock price target @ 27 after the 3Q earning report”?

Well, INTC jumped $1.52 during after-hours trading today 10/16/07.



Remember what I said about the comparative trend lines between INTC and AMD?

You just watch, tomorrow’s opening bell!



Remember what I said about the IDIOT bean counters NOT factoring in INTC execution as to AMD’s failure with Barcelona?

Those pinheads on Wall Street just got a wakeup call. They are going to take a good hard look at the reality of the two company’s market position and execution.



Remember when I said Nehalem will be Ottelini’s and company’s coupe de gras?

The above mentioned INQ article! 2008 WILL be INTC’s great server assault!


Oh yeah, Nvidia has been very quiet lately.

Too, quiet.



Think AMD’s second consecutive “Perfect Storm” after its 3Q earning report. It gives a new meaning to the “Second Coming”

By the way, would anyone like to take bets on 37 cent earnings per share for INTC?
THIS IS HUGE!



SPARKS

SPARKS said...

HOLY COW GURU DABBLES IN THE MARKET!

The way I see it now, his speculations a worth more than 200 IDIOT anal-ists!

Energy stocks and supply chain infrastructure!

HAIL GURU!

SPARKS

Giant said...

I was joking.... the dear Dr. has actually changed his prediction for Intel BK to 2009... He keeps pushing it out actually.

First he predicted Intel would be BK in Q4'07 (now). Then it was the infamous Q2'08, and now it's BK in 2009.

Tonus said...

I think he is sticking to his BK Q208 prediction, but now the 08 refers to 3008.

Eddie said...

Hello, Roborat64. I have been a reader of your blog for a while, I hope you will keep meeting your objectives because at least for me this blog is a good resource that I would like to see prosper.

I wrote a post, 65nm hurts AMD, where I link to this article of yours. I was about to let you know of this link, but before I got to do it I saw that you linked to my blog instead, in consequence, I reciprocated in kind by linking to yours.

Just allow me to tell you that my blog is purely a place to exercise free thinking of both Howling2929 and myself, where we also allow the interested people to read and comment, in that regard, I sometimes "think at loud" and long articles without too much cohesion are published.