4.08.2008

AMD's Annual Layoff Exercise

"Advanced Micro Devices, reporting a slump in sales, said Monday it will cut 1,680 employees, or about 10 percent of its worldwide workforce, between now and September. An AMD spokesman said the move was part of a cost-cutting plan to help return the company to profitability. He said the company doesn't know how many will be affected at its two Silicon Valley campuses. AMD said that the revenue drop was caused by lower than expected sales across all business segments, which it attributed to an anticipated seasonal decline". -siliconvalley.com

Expect every horrible announcement to be beautifully written and without fail stated in a positive manner. It is amusing that AMD calls a sudden announcement of workforce reduction as a cost-cutting "plan". Intel announced a 10% workforce reduction over the course of several quarters to save $2B annually. Now that is what you call a plan.

It's not surprising that AMD's sales are off by around $100M. AMD's customers are left to choose between an aging Athlon or a Phenom that does not have guaranteed delivery dates. Large OEM planners are having problems booking production plans for Phenom based systems - or so I heard. To most companies with tiny margins, on-time delivery is key:

Citigroup wrote about the issues with Dell this morning, saying: "Our analysis suggests that a source of AMD's shortfall is Dell's decision to reduce their AMD exposure. Consistent with speculation throughout the quarter, AMD's shortcomings in two flagship products (Barcelona and Phenom) was the likely sore spot for Dell. - cnbc

58 comments:

SPARKS said...

Since we beat the life out of this topic on the last post, I thought a little trip down Memory Lane was in order here.

“Today marks a historic day for our employees, our partners and our customers as we officially welcome ATI into the AMD family,” said AMD Chairman and CEO Hector Ruiz. “On day one, we are delivering a winning set of complementary technologies, igniting a new level of innovation and continuing to champion choice for the industry. Thanks to the strength of our talented employees, the new AMD now has a full range of intellectual property (IP) in microprocessors, graphics, chipsets and consumer electronics to deliver open platforms and integrated solutions. In the near term, customers gain a new level of choice, and in the long term, we believe the possibilities for innovation are truly limitless.”

No, I think today marks a historic day for AMD employees, et al. This was the Wrecktor Ruinz pipe dream power play on INTC back in 2006. What a stupid bastard.

“talented employees”

SPARKS

http://www.amd.com/us-en/Corporate/VirtualPressRoom
/0,,51_104_543_13744~113741,00.html

Randy Allen said...
This comment has been removed by the author.
Hector de J. Ruiz, Ph.D said...

What?! I'm not a stupid bastard, honest. I'm not taking AMD out to the shitter! It's all Randy's fault, honest it is!

If anything I deserve a hefty raise, wouldn't you agree? They still haven't finished my castle made from bricks of gold yet! I only made a puny 7M last year, that's not enough to finish off the castle. I think I deserve AT LEAST 10M this year!

Sincerely,

Hector.

Anonymous said...

Oh Paid Intel PumPtrasters!!

HEre are set undeniable factitudes:

A. HectoR RUIz is t3h God and will FOREVER Run AMD in all perfection. Just look at the perfection of financial results: AMD is making $500TRILLION in moneys after Intel is made to be paying!
Do not listen to Intel PUmper liars like Scientia and Randy Allen who talk about "revenue" and "profits". These are all fanboy lies! All true AMD peoples know that Every K10 is an individual and unique snowflake and that like Snow, AMD will destroy Intel!!

4. AMD is the most prosperous company in all the histories! There has never been a "lay off". All the news sites are LIARS. Instead the perfect AMD is eliminating INTEL SPIES!
This is 100% with the Logics. AMD is perfect, and Hector is God. Therefore anything Hector does not approve of must not exist. Therefore the supposed "core 2" is a COMPLETE PUMPER LIE from Intel. Intel can only steal REAL chips from the AMD!!
So... Intel is hiring of the SPIES to sneak out super-hyperclocked AMD chips in lunch boxes! Hector used his superpowers to personally identify and ELIMINATE these evil spies!

AMD will be Purged of t3h Evils!!! AMD is Perfect!! K10 Is perfect!! There is no "core" there is no "Nailhomer" there is Only Hector!!

Anonymous said...

Anyone know what was the last company that Hector worked at?

What happened to them?

Anonymous said...

Can't wait to see Scientia spin this one.

I told everyone quarters ago when AMD was making money hand over fist it was only a matter of time.

Once the sleeping giant was awaken their bigger bank account, superior technology, bigger manufacturing machine and army of designers and architects once pointed in the right direction would kick AMD's A$$ back to the dark ages.

Face it superior technology, superior manufacturing and huge econcomies of scale can't be beat. Any monkey who thinks a consortium can produce better technology, that one factory with APC can beat 3 factories, that asset lite can beat asset rich company doesn't know what it takes to make a CPU.

AMD is finished.. All 45nm aren't the same, anyone who believes that AMD Shanghai on 45nm is the turnaround vehicle is the same fool that believed AMD wasn't going to be beat two years ago.

AMD will never make it to volume 35nm..

Anonymous said...

So looking back on things was Dell going to AMD the beginning of the real end? (of course the ATI overpayment played a small role as well) Yes AMD had a couple of good quarters after it but here are some thoughts:

1) AMD did not have the capacity when they did the Dell deal, as a result the channel and lower priority OEM's got shorted on parts. Intel exploited this and this is something folks don't quickly forget. Now come around with the Barcy issues and OEM's start seeing a pattern as opposed to a one time event.

2) Dell was likely getting far better pricing then everyone else due to volume expectations (thus beginning the eventual erosion of ASP's and margins). AMD was already selling everything out when they did the Dell deal, so all they effectively did was lower the amount of money they were getting.

3) Dell required capacity guarantees up front which significantly limited AMD's flexibility - how may times did you hear poor product mixes during an earnings call? This also led to shortages in certain areas and ties in to point #1.

4) The AMD deal was the end of Intel rebates/preferred pricing for Dell (or maybe it was the lawsuit)- this gave Intel MORE revenue and allowed them to cut prices elsewhere with less pain (or spend the increased revenue on other priorities).

Some might say losing some business at Dell was the best thing that ever happened to Intel and the worst thing that could've happened to AMD.

It is unfortunate that employees now have to pay for high level mis-management while that management shares no pain... how about some leadership from the top and Ruiz accepting

Anonymous said...

Anyone know what was the last company that Hector worked at?

I believe it was Motorola. They're still recovering.

SPARKS said...

“………….. but here are some thoughts:”

Excellent summation, brilliant one thru four, however, I believe there was another. Shall we say a fifth that you may have overlooked? This is pure speculation on my part. It’s up there with secret societies and corporate conspiracy theories. Give it whirl.

5) DELL was made a non combatant in the AMD v INTC antitrust lawsuit.

This is a HUGE win for INTC. You alluded to it in item four, “(or maybe it was the lawsuit)”, but I don’t think you went on to reveal its full impact. The substance is there. The evidence, however, isn’t quite so clear, ------yet.

“Some might say losing some business at Dell was the best thing that ever happened to Intel and the worst thing that could've happened to AMD.”

This is gospel, designed, composed, and orchestrated by none other than WRECTOR RUINZ himself, entitled, “The Perfect Storm”.

SPARKS

SPARKS said...

GURU! I'm DYING!

http://news.moneycentral.msn.com/ticker/article.aspx?Feed=AP&Date=20080409&ID=8461922&Symbol=AMD

"ALBANY, N.Y. (AP) - State officials have budgeted $46 million to help Advanced Micro Devices Inc. build new factories in upstate New York"

and.............

"At this point in time it's full bore ahead ... with the expectation that sometime around next January they'll be able to start construction," said Malta Town Supervisor Paul Sausville."

We are in 4.4B in debt!!!!!!!!!!!!

"AMD's moving forward and developing and implementing the necessary measures to ensure the plant will be one of the largest economic development projects in New York state history," Thompson said. "We have no reason to believe it will be sidetracked."

DO I SMELL ANOTHER SHOREHAM!?!?!

Their killing me.

SPARKS

SPARKS said...

Sorry, incorrect grammar.

THEY’RE KILLING ME!

SPARKS

Anonymous said...

A fab costs a couple billion, and NY is throwing in a few tens of million.

Does anyone in the AMD camp appreciate how dire things are?

Takes a good billion to two to develop next generation processor design and an equal amount to develop the process. Is it not wonder that the current 65nm and the upcoming 45nm won't be optimized to AMDs particular needs and thus why there were surprises during the Barcelona validation. Expect those to be worst on 45nm and AMD to be out of the silicon business by 32nm.

Tonus said...

"A fab costs a couple billion, and NY is throwing in a few tens of million."

This is how you bleed the taxpayers dry. Spend millions of dollars on a billion dollar project that has a good chance of not being developed. Keep spending a few more million each year because shucks, you're in for a penny, in for a pound.

New York City is finally developing the Second Avenue Subway line, after spending millions of dollars a year for THIRTY FOUR years to maintain the tunnel while they waited for an opportune time to build.

Thirty... four... years.

Anonymous said...

AMD needs to sell off its fabs and go "asset-lite". That's the only way the company can survive. Look at what AMD is doing now, wasting hundreds of millions on an inferior process. If the x86 license agreement prohibits outsourcing to TSMC then AMD needs to come together with Intel to negotiate. Maybe take all anti-trust crap off the table in exchange for a free and clear x86 license.

The only thing standing in the way of "asset-lite" for AMD is pride. The whole world knows that AMD is a second-class company to Intel but Hector can never admit it to himself. Going "asset-lite" would mean that AMD would be second best forever with no hope of ever catching Intel. But if bankruptcy is the alternative, second best doesn't seem so bad.

Anonymous said...

"DO I SMELL ANOTHER SHOREHAM!?!?!"

Has that been taking apart yet? Or is it just sitting there mocking people?

I'm surprised you didn't mention the award of a 33Mil contract for roads that was recently done for the plant as well.... death by a thousand paper cuts! 33mil here, 46mil there and pretty soon we're talking real money!

Anonymous said...

"Maybe take all anti-trust crap off the table in exchange for a free and clear x86 license"

I couldn't see Intel doing this - do you know how much that license is potentially worth? If this was ever done, AMD would likely just sell it (and could do this to as many companies who'd be willing to buy it) and Ruiz would (golden) parachute out.

Whoever wrote the original contract surely had some foresight - probably seemed like not so big a deal at the time when it was agreed to but this outsourcing restriction is like a cement block dragging AMD down. I think the license is up for renegotiation in 2010, so AMD needs to (just) hold out until then and hope they can negotiate a better deal.

Maybe they could negotiate that Intel will be the sole foundry for AMD chips!

Tonus said...

Would it be in Intel's best interests to give AMD some breathing room vis-a-vis the x86 license just to keep trust-busters at bay?

Let's assume that AMD is going to die off, or perhaps just fade into irrelevancy. If Intel pulls the license agreement at any point (assuming the proper clauses are triggered) they would face accusations that they did it to finish off AMD and cement their own monopoly position. If AMD dies off/fades away and Intel didn't pull the license agreement, they could shrug their shoulders and say "see, they did it all themselves."

Does that make sense, or is it better for Intel to pull the license should the opportunity present itself?

Anonymous said...

"Does that make sense, or is it better for Intel to pull the license should the opportunity present itself?"

This isn't a question of 'pulling' the license, it would be a question of AMD violating it. The free and clear part won't happen as I don't see Intel wanting AMD to offer the license to the highest bidder (or bidders as you can license IP to as many parties as you like), but I could see some terms being renegotiating in advance of 2010 which would allow for oustourcing (but perhaps not allowing the license to be sold, or survive a sitauation where AMD is acquired by a larger company)

And let's be straight here, if AMD dies, AMD finished off AMD, not Intel. Just 2 years ago this was a healthy company on the rise - the inability to outsource is not the cause of AMD's current issues.

SPARKS said...

“And let's be straight here, if AMD dies, AMD finished off AMD, not Intel. Just 2 years ago this was a healthy company on the rise”

You know, that’s so good let’s do it again.

“And let's be straight here, if AMD dies, AMD finished off AMD, not Intel. Just 2 years ago this was a healthy company on the rise”



See, the inverse of this precise and direct statement/concept is entitlement. Where on this earth does it say in corporate law when a company’s upper management makes fatal errors in judgment totaling BILLIONS of dollars, they are entitled to special privileges and considerations by political, legal, or governing bodies?

In fact, if the early 2006 cash rich AMD exercised caution and prudence and licensed ATI technologies as opposed to buying them outright, they wouldn’t be in this calamitous situation today.

Ironically, if they at the time, focused their efforts on their most valuable asset, the x86 license, they wouldn’t be in a position to defend it by violating the agreement out of dire necessity.

Look, their energies and goals were focused on everything but. They arrogantly set the company in a different path and direction and catastrophically geared the company toward the entire platform solution. They needlessly bought in to the graphics arena, a technology they were completely unfamiliar with, thereby diverting vast amounts of money and resources from their main product PROCESSORS! Further, they blindly failed to recognize the market dynamics of buying into two very POWERFUL competitors the very moment they purchased ATI. Consolidating and spending billions while taking on INTC and NVDA is exactly why private capital investment, pouring tons of capital into the current AMD/ATI, is so ugly presently! AMD paid 5.4 B for the privilege, incredible!

“do you know how much that license is potentially worth”

This statement is so ridiculously obvious it completely underscores the mindset and lack of vision by AMD’s top management. At the time, holding the x86 wasn’t an issue, now in complete and utter 20/20 hindsight they are scheming everyway possible just to hold on to it, including government and/or legal intervention. Only now do they realize the importance of a powerful processor.
Sorry too late.

Is INTC supposed to be the nice guy and give away the company for IBM’s 20 year old second source which INTC selected from the onset?

Or perhaps, they should give AMD complete dominion over the license so the ENITIRE FUCKING WORLD, SOLD TO THE HIGHEST BIDDER, COULD BRING DOWN INTC! This way AMD could survive and prosper, and INTC could fade away and DIE!?!?!

This is absolutely, insanely preposterous! Why is it so hard for anyone to believe and/or accept AMD had it all and simply BLEW IT???????

“do you know how much that license is potentially worth”

AMD IS NOT ENTITLED TO MULTIBILLION DOLLAR fiascos and failures, and they can’t be given a free pass simple because they are…………….

“The Scrappy Little Company”


SPARKS

SPARKS said...

HA HA HA HA HAAA!


http://www.theinquirer.net/gb/
inquirer/news/2008/04/10/intel-
atom-beats-via-isaiah


SPARKS

SPARKS said...

HA HA HA HA HAAA!


http://www.extremetech.com/
article2/0,1697,2282395,00.asp

Lowly Q9300 kicks the shit of AMDRIODS best.

SPARKS

InTheKnow said...

Okay, let's take a step back and look at AMD's predicament two years ago.

The writing was on the wall, that the desktop was no longer going to be driving growth.

Severs are good money, but are inherently a low volume business. If there is one thing you need to keep a fab from turning into the money pit, it is volume. So servers alone aren't the answer.

With growth in the desktop was drying up and knowing severs will never be really big volume, where does AMD have left to turn in order to generate the volume needed to keep their fabs profitable. The only thing left is to get into laptops where the growth was clearly shifting, or find a way to drive x86 into new markets.

To drive x86 into new markets, you need deep pockets, because many of your efforts are going to net a loss. If you can't shrug off a bunch of lost cash, that isn't the way to go.

In order to get into laptops, AMD needed a graphics solution. From that perspective, you can see why the ATI acquisition was attractive to AMD. It gave them what they needed to sustain long term growth.

I think the big mistake was in not taking care of the short term first. AMD arrogantly assumed that they would retain market share (and even be able to grow it) while digesting the ATI merger. 2-3 years down the road, they would be in much better shape.

Unfortunately (for AMD), Intel didn't just lay down and die. Instead, they got up and showed what made them one of the top companies in the world. This left AMD with shrinking market share at a time when the had left themselves with very little cash in hand.

Rather than cutting his losses and running, Hector beat the Barcelona drum and kept going after market share at any cost. Now he is having to divest himself of his most valuable resource (talented employees) in order to stay afloat.

I really think that Sparks was on the right track with his suggestion of licensing the needed graphics capability. As another option perhaps they could have entered into a joint venture.

Imagine AMD only plunking down $2B (max) in cash for a joint venture with ATI instead of $5.4B to buy them. They would not be carrying the crushing debt they have now and would have been able to stay at least on the fringes of profitability. That scenario would have given Intel a run for their money.

S said...
This comment has been removed by the author.
S said...

I think AMD is picking up pieces and starting to look forward again. Phenom is out performing reasonably well. 45nm Samples are out. There have been some good products out of the ATI stable. Intel is reaching out to ATI again to fight nVidia.

They need to productionalise 45nm quickly as selling products with 1.5x die size at 2/3rd the price as compared to Intel effectively makes their margin wafer thin.

Tonus said...

"And let's be straight here, if AMD dies, AMD finished off AMD, not Intel."

I agree, but that wasn't what I was referring to. AMD is the architect of it's own current situation, but that hasn't stopped the EU (and the US DoJ, in the past) from investigating Intel and promising to keep a sharp regulatory eye on it.

Making sure your hands are clean is a political expedient. Since when have politicians cared about accuracy when they were on a crusade against a particular target?

Sure, if AMD violates the terms of the agreement, Intel is entirely within its rights to terminate it. But my question was more along those political lines. Do you feel that the government here would consider leaning on Intel if they did that, or that the EU would ratchet up their own case (well, I suppose that's a given)? Neither would go after Intel over the actual act of terminating a contract that was breached. But they may start (or continue) making noise about anti-trust. That is what I am wondering about and am looking for opinions on.

Tonus said...

Ed at Overclockers thinks thatAMD might make this all moot.

I know that he hasn't been the most AMD-friendly writer out there, but he does have a point-- as negative as his forecasts for AMD have been, the reality has often been worse than he was predicting.

Anonymous said...

Now Phil Hester has left! Soon AMD will have no executives other than Hector!

SPARKS said...

"Now Phil Hester has left! Soon AMD will have no executives other than Hector!"

Isn't the captain the last one to leave a sinking ship. Or, are the rats the first ones to bail?

I guess it really doesn't matter at this point.


"JMS
Oct 23, 2006 3:29 AM GMT

Let's see, 5 billion ATI buyout, no 65nM process, newer chips with small speed ramps, second-tier vendors screaming for chips, no visable answer to Conroe, and Christmas is eight weeks away. Sure, AMD looks great, so did the Titanic just before it hit an iceberg. 4Q will be AMD's (and investors') worst nightmare."

I wrote that in Business Week, 10 '06.

http://app.businessweek.com/
UserComments/combo_review?
action=all&style=wide&productId=
11253&pageIndex=1

SPARKS

SPARKS said...

YO! Giant, check this out! NVDA"s JEN SUNG HUNG is having temper tantrums---------- publicly.

INTC is starting to go to work on him. And oh boy, did they get under his skin! Did I, or did I not, tell you this “daytime drama” was going to slowly unfold? It’s heating up, real good, and it’s gonna get better.

If there was any doubts how much they hate each others guts, let this dispel them. That said, NVDA took a pounding in the market today.

The INQ wasn’t too pleased with his performance either.

INTEL CORPORATION, oh the joy!

I hope they kick his “ass back to the dark ages”, too.

http://www.theinquirer.net/gb/inquirer/news/2008/04/11/nvidia-chief-gets-graphic-intel

SPARKS

InTheKnow said...

Goldman Sachs doesn't seem to be too high on AMD. Surprisingly, they seem to think hemorrhaging money isn't a good business plan. This quote from the NY Sun.

"Our view is that AMD's model remains structurally broken and we continue to recommend that investors sell/short the stock. We are increasingly concerned about AMD's cash burn (we estimate a $550mn cash burn in Q1'08 alone) and expect the company to try to come to market later this year given its $5bn in debt and less than $1bn in cash pro forma for 1H'08 losses," an April 7 analyst report from Goldman Sachs stated.

Anonymous said...

"Goldman Sachs doesn't seem to be too high on AMD."

What's scary is that Goldman RARELY recommends short selling - they will give a stock a 'sell' rating, but rarely go to the sell short extreme. This is a bit surprising given where AMD stock price is - a merger or private buyout could possibly blow those shorts out of the water.

(Of course there were folks on this blog talking about how AMD is so much more attractive at these price levels not too long ago and were bashing analyst downgrades to sell when the stock was at ~$8)

SPARKS said...

I thought this might be a interest read for anyone who doubted INTC's superiority, and is still bullish on AMD.

http://www.uscourts.gov/
bankruptcycourts/bankruptcybasics/
chapter11.html

SPARKS

Anonymous said...

AMD going going going

Its of no suprise at all that the rats are all abandoing the ship. ANyone that has to work for another 10 years, or wants a career will get out of AMD is its the Titanic and its already hit the iceberg. Once they bought ATI and started talking fusion and other nonese was the end.

They can't even get their 65nm designs to yield as they shortchange the required investment in process and design. How the hell can they be expected to ramp a successfuly high perfomring 45nm process and design. Without adequate investment years ago these things can't be shortcut. I expect 45nm to be poor yielding and have the same surprises as AMD had on 65nm. As to 32nm, forget it, they never had the money nor time ot invest as they needed to make such complex process and design work togather.

I told Rick and Scientia this over and over but they are too stupid to understand the business.

LOL The big guy Lex

SPARKS said...

GURU, ITK, do me a favor; see if I have this binning thing right. First, check out these numbers

.
4 core 2.5GHz 2MB 2MB 2.0GHz 1.2-1.3V 125W


3 core 2.4GHz 1.5MB 2MB 1.8GHz 1.05V-1.2V 95W

3 core 2.1GHz 1.5MB 2MB 1.8GHz 1.05V-1.2V 95W


And, last but most ugly

3 core 2.1GHz 1.5MB 2MB 1.8GHz 1.05V-1.2V 95W

I know certain chips are qualified for speed, but what I don’t get is why nothing changes with the power and thermals.

From the HEXUS article they have a total of 8 binned models from the top dog 4 core 2.5 to the mutt 2.1G 3 core. Does this mean for every seven mutts they get a good one? Are the thermals so bad they bin them on power usage at certain voltages, as apposed to binning them at certain speeds? Why is it no mater what they do with the voltage the thermals never change, even at lower speeds

What I really don’t get the 2.5G 4 core with larger cache, higher memory controller speeds, and higher voltages, only burns an additional 30 watts.

1 additional core @ 2.5 gig, 25% more cache, with 20% more voltage, equals only 30 watts more than the most ugly at only 400 MHz difference.

Sorry, I don’t get it. What am I missing here? Is AMD trying to sell anything they can get off the wafer, or what?

http://www.hexus.net/content/item.php?item=12761

SPARKS

Anonymous said...

Sparks - I think I wrote about this a bit while back. My suspicion is that some of the tri-cores are quads that couldn't hit the thermals (TDP's) for the quad SKU's - by disabling a quad this helps bring the thermals back down into an acceptable TDP range. This (in addition to a dead/defective core) is another reason why AMD is pushing tri-cores. It is symptomatic of running the process/design near a cliff - ideally you would have some margin for binning these chips but given the thermals and the lack of any higher speeds it is clear AMD has close to zero window on some of these bins. So AMD's only choices are:
1) Jack up the TDP rating, which would be a PR nightmare for the 'drop-in' marketing spin (cough, cough if you have the right MOBO that actually supports it)
2) Throw out the quads that are above the TDP specs (not a good thing for a company losing money)
3) Turn off a core and hope it's power drops enough to sell it as a tri-core - then claim the idea behind tri-core is market driven (Ding, ding, we have a winner!)

To me it's fairly obvious that the tri-core and quad-cores are not coming from the same bin (the tri's are from a 'worse' bin) and therefor the folks who kept theorizing that the tri-cores should be much better overclockers are in for what I think will be a harsh dose of reality.

...and in other news Fudzilla is reporting that the 9950 (2.6GHz chip) will be 140Watt TDP!!! Just how long ago was AMD the champion of green and power consumption - looks like that went out the window. Though I'm sure this chip will be an excellent overclocker(?). Apparently there is there is not truth to the rumors that this will be called the 'red edition' or 'white hot' edition.

I wonder if these are just 9850's with the Vcore jacked up a bit (essentially factory OC'd parts). But hey at least AMD managed to squeeze out 5 or 6 SKU's in the quad desktop space - it's a huge market you know (though when Intel was the only player it was of course a 'niche' market)

Anonymous said...

Oh and Sparks that Hexus article is now starting to show the folly of having so many SKU's and why I will still question whether it is perhaps better to simply throw the tri-cores away.

Look at the pricing:
- A 2.4GHz tricore is the same price as a 2.2GHz quad, with same TDP - so people will turn down the extra core in favor of a 10% clock increase? And if folks start preaching about SW not taking use of quads, then they should be buying an even faster DUAL core, not a 10% faster tri-core!
- The bottom tricore pricing (99E) will be putting pressure on dual cores (if they ever come out) - so you have to ask is this sweet 99E-122E range tricore market, the one customers are "demanding", really worth it? Would AMD be able to get a bit better pricing for the dual core chips if this segment didn't exist? Now factor in the relative volumes and an argument could be made, under the right circumstances, it is better for AMD to simply throw out the tri-cores.

While Hexus' argument that the tri-cores should be priced lower to reflect performance makes sense technocally and intuitively, this would absolutely CRUSH the dual core (high volume) pricing - and thus AMD is in this pricing predicament.

Intel was in a similar predicament - what did they do.... well you don't have 800 freakin products in such a tight range (generally 3 is good enough) and you keep the extreme editions purely in one top product bin (quad core, max speed). I'm sure Intel could put out a pretty kick-ass high clocked dual core, but it would completely mess up the whole pricing tree.

Ho Ho said...

Lol, I just got banned from amdzone about telling a mod to stop personal attacks and try to actually write something that adds to a thread instead of constant OT :D

Tonus said...

I think we're seeing the benefits of having the fastest CPU. Sure, most people are not going to pay $500 for a CPU, much less $1,000 or more. But when the high end of your price range is $1,400 you have a lot of room to create a varied product mix at lots of prices.

When your best CPU is about as fast as the competition's $250 CPU, your options become pretty limited. I think AMD is trying to have a product lineup that rivals Intel's, but they have neither the speed range nor the pricing freedom. Twenty SKUs from $60-$1,400 isn't so bad. Twenty SKUs from $60-$250 is bad, especially with the overlap that they've created with the tri-cores and Black Editions.

Sometimes you need to swallow your pride and simplify things. I think AMD needs to do this right now. Their product lineup and pricing scheme makes me wince.

S said...

There is space for a dual, quad core at the same price point. For example a Gamer will go for a fast dual core. A multi tasker will go for a Quad core. Would someone doing both benefit from a tri-core ? It is a glass half full situation. A tri-core may be slower for games than dual core, but will be faster than a quad core. Complicates decision making. A dumb user may just base their decision on $s per core.

I think, introducing tri-core will see AMD through this year, breaking even by Q4, before it is hit by Nehalem speed train.

Anonymous said...

"There is space for a dual, quad core at the same price point. For example a Gamer will go for a fast dual core."

You're naturally assuming that K10 dual cores will come out substantially faster than quads? I have seen no announcements... and the lag on these makes me wonder how much better they will be than K8 or how much faster (clock-wise) they will be over K10 quads.

Another place AMD has pigeoned-holed themselves into is the slash and burn pricing on K8... if the K10 dual cores are not substantially better than AMD is in serious trouble. They already lost all hope of higher margins via desktop quad cores and soon K10 dualies may be relegated to the 'it's efficient and good enough' scrap bin - yeah you may sell some volume, but AMD need higher ASP's more than volume at this point.

Also AMD's problema are not JUST performance - how many quad desktop products does Intel have? How many does AMD have? You simply don't need as mnay as AMD has (the market is just not that big yet). This appears to be a problem with binsplits and instead of someone with some business sense saying we don't need 100MHz steps, and asking why there are products with the same clock speeds but different powers AT THE SAME PRICE - they should just simplify the quad lineup and put out some dual cores which have to be far cheaper to produce.

And tricore should simply be eliminated - or down to 2 speed bins... (they could drop the top bin price down a bit, and raise the bottom bin a bit). Are they looking for the high end desktop user for tri-core, or the budget tri-core user, or the mid-range user? Do they think these all exist for tri-core? For folks who want clock - they would go dual core anyway, for those who want multi-threading/tasking, they will go for the quad.

Finally get the quad desktop to 3, maybe 4 products - one low end 'efficient' (cheap) part, 1 or 2 'mid range' parts with 200MHz steps (maybe 2.2 and 2.4) and then one 'extreme' part (2.6)... get rid of 2.3, 2.5 and the multi-TDP flavors at some clocks.

SPARKS said...

Guys, I didn’t mention the pricing scenario of the quad vs. tri thing. It is too pathetic, too ugly and too ridiculous. You all handled quite well, I might add, A couple of bucks separating these low performing power hogs left me beyond words. I was lucky to get past the binning thing (thanks GURU). The whole thing’s a mess designed to eek some money by selling anything they can.

“why I will still question whether it is perhaps better to simply throw the tri-cores away.”


The question is, why bother purchasing anything they make? I can buy a Q6600, or for few bucks more, a Q9300, and be done with it. Either of these two will clock to 3 Gig with nary a sweat, just for shits and giggles, with lower thermals to boot!

A 95 watt to 125 watt spread going from 2.1 to 2.5 GIG?!?! Call me bias, as I am, no doubt, but, these things couldn’t look any worse than they do, even if they planned it this way. Frankly, I think they're just winging just to have something to sell.


“I think, introducing tri-core will see AMD through this year, breaking even by Q4”

Ahhh----- Here’s some sound advice, don’t bet on it, and certainly, don’t take a mortgage out on the house to buy AMD stock.


“and thus AMD is in this pricing predicament.”

Man, you said it.


“Their product lineup and pricing scheme makes me wince.”

Tonus, my sentiments exactly, the whole thing reeks.

I’ve been building my machines for 18 years. I’ve NEVER seen anything as ugly as this, bar none. This is the final stepping on 65nM; it’s no wonder the rats are jumping ship.

SPARKS

SPARKS said...

Did I say the rats are bailing?

http://www.fudzilla.com/index.php?option=
com_content&task=view&id=6816&Itemid=1

SPARKS

SPARKS said...

Ok. Now factor this.

Huh?

http://www.tomshardware.com/2008/04/14/
amd_phenom_triple_cores_more_expensive_
than_quad_core/

Sorry, now I’m really confused.

SPARKS

Chuckula said...

I wouldn't read too much into the price inversion for triple core CPUs. The article you link even notes surprise that they are available for sale at all. This probably means the high price is due to very limited quantities of chips which are garnering more demand.
If the price inversion lasts more than a month or so then I'd be very confused, but once the triple-cores become as widely available as the quad cores the price is gonna drop like a rock (especially because they don't appear to be clocked any higher).

Anonymous said...

"The question is, why bother purchasing anything they make? I can buy a Q6600, or for few bucks more, a Q9300, and be done with it."

Well for some it may be worth it. Certainly if I had an AM2+ board (though I'm not sure why I would), it would be strongly considered. Also if you are making cheap the priority, pairing it with a 790G for a HTPC type application would not be absurd either. Bottom line is there are AMD fans who will buy it.

"A 95 watt to 125 watt spread going from 2.1 to 2.5"

Be careful with associating TDP with actual power consumption... if the chip was 100Watt at max they can't put it in the 95Watt bin so it might not be as large a bump as it seems on paper (I haven't looked into actual power measurements too carefully). You may ask why not make it a 100Watt or 110 Watt bin then -well you want to leave room for future clocks...though if Fudzilla is right the next 100MHz step is going to blow the 125Watt bin out of the water (to 140Watts). You also probably don't want to stick your OEM's with a lot of different TDP's to design toward. I'm still of the opinion though, that the majority of their problems are power-related on the quads and much of the problems, including the ridiculous over-segmentation of clocks and powers in the quad core space is a result of this.

Anonymous said...

"The whole thing’s a mess designed to eek some money by selling anything they can."

That's the most amusing thing - they are probably LOSING money by selling these. They are no doubt making money on the tri-core chip itself, but it's impact on quad core sales and both quad and dual core pricing will likely make this a negative ROI fairly quick for AMD. They are just too short sighted and cash strapped to see it (or choose to ignore it in a sense od desparation). This lack of foresight is yet another reason why Hector should not be running the company - instead of babbling about this being a 'unique', 'market driven' product - Hector should be questioning his sales and marketing teams on the sanity of this approach.

Think of it this way - a tri-core purchase is in all likelihood either a lost quad core sale or an upgraded dual core... and not really a 'new' market - but how much of the market will this be? Maybe 5%? Now will these have a 2.5% impact on dual core pricing? (i.e will the existence of these partspush the dual core pricing down by $4-$6?)

Surely I mean 5%, no? Well, the chip is a disabled quad so for every 1 tri-core sold, that's potentially 2 dual cores that could have been made for the same cost.

Well, what if it becomes more than 5% of the market? well then AMD has a REAL problem as they either have to hope their yields stay bad to support it, or they now need to start disabling working quads.

Lunacy, I say, lunacy...just throw the tri-cores away and bump up the pricing of the cheaper to produce dual cores to make up for it! I hope Intel doesn't follow suit when Nehalem comes out.

SPARKS said...

“Well for some it may be worth it. Certainly if I had an AM2+ board (though I'm not sure why I would), it would be strongly considered.”

How strong? Say, I give you a 5 to 1 shot? I’ll take the odds and won’t hit you with Vig for a month.


“The vast majority, eight out of ten motherboards, did not work with Phenom at all, which I found a very frustrating result.”


I don’t know if there is a difference between the AM2(which AMD claimed would work initially), or the AM2+. I’m not sure that I care, except it left AMD fans waiting for something that never really worked. Unless, of course, they spent money on a new motherboard that would. That’s if they hell bent on running Pheromones. Sure, right, the “The Smarter Choice”

Besides, someone’s going to compare a maximum overclocked (3.1 GIG) Pheromone ‘Black Edition’ with a maximum overclocked 1yr. + old Q6600 (3.2~3.4), and finally put all this bullshit to rest.

If the article below were INTC, THG would have had a field day, “frustrating result”, GMAFB. It was a nightmare.


http://www.tomshardware.com/2007/12/26/
phenom_motherboards/page15.html

SPARKS

SPARKS said...

For your amusement:

http://products.amd.com/en-us/RecommendedMBFilter.aspx

SPARKS

InTheKnow said...

Our friend with the rose colored glasses is at it again. Here is his latest.

We can finally settle the question of whether AMD's 65nm process is broken. AMD's fastest 65 watt, 90nm K8 runs 2.6Ghz while the fastest 65 watt, 65nm K8 runs 2.7Ghz. So, the 65nm process is at least a little better than the old 90nm process.

So we got just under a 4% speed bump by moving to the next process node at the same power. It took them ~4 months to achieve this stunning milestone.

By way of comparison, AnandTech reported Smithfield at full load drew 233W at 2.8GHz. Presler, with twice the cache (2x2M vs 2x1M), runs at 3.4GHz at 235W. I make that out to be a 21% increase in clock speed.

So here is my conclusion. In order to match the improvement Intel saw in clock speed from what was essentially a dumb shrink moving from 90nm to 65nm AMDs processor would need to run at 3.15GHz. I'd settle for 3.1GHz.

So while the statement that 65nm is slightly better than 90mn for AMD is technically true, it isn't showing the expected power/speed improvement.

So is AMD's 65nm process broken? That depends on your definition. The data point presented leads me to believe that the drop in voltage that comes with the shrink is almost completely offset by the increase in leakage.

I remain unimpressed.

Anonymous said...

"So we got just under a 4% speed bump by moving to the next process node at the same power. It took them ~4 months to achieve this stunning milestone."

And to pour further salt on that wound, the power on 65nm seems to go up far faster than on 90nm, which is why we don't see higher clocks (and for reasons intheknow alludes to). Scientia is also comparing the TDP BIN and not ACTUAL POWER USAGE, so is it really better/ There was some data on the various 65nm and 90nm steppings (I think it may have been THG) which show 65nm equal to the last 90nm stepping at idle and actually slightly worse under load!

If my Kia goes 0-5mph slightly faster than someone's Porsche, is it better?(conveniently ignoring above 5mph where things may not compare so well!) What if I just go by what the cars were rated to do (they both claim they can go 0-5mph in under 2 seconds), instead of looking at the actuals!

"we can at least give AMD a small amount of praise for getting the B3 stepping out the door."

We can praise AMD for fixing an issue which those folks claimed was not significant to begin with? We can praise AMD for releasing a B3 part which is STILL BELOW THE EXPECTED 2007 speeds! The only praise here is to the person at AMD finally woke up and said we are can squeeze no more blood out of the 65nm K10 stone and said let's end the craziness here and move to 45nm. Whoever made that decision should seriously be promoted (and I'm not being sarcastic here - this was actually a very smart business and strategic decision).

Remember "K10: A good start"? now it's K10 "a small step on a long road". He's been talking about future products for over a year now... 2007 should be interesting, Q1'08 should be interesting when K10 comes in volume... when you are CONSTANTLY talking about the potential future products may have and spend no time on the current actual products, that's a strong sign.

He can talk about SSE5, AVX, GM2, but Bulldozer ain't coming until at least 2010, so what is AMD to do in the meantime (ride it out for 2 years!?!) The other problem is these features are sitting on a weak foundation of a leaky process with low clockspeeds. AMD needs to fix the foundation (the process and design iinteraction health), before any of these features can amount to a hill of beans.

Although Kudos for getting the superpi benchmarking in, but points off for failing to discuss the benchmark riggin via use of Intel optimized compilers (he's getting sloppy!)

"I think a better strategy would be for AMD to get the 2.6Ghz 9950 out the door as soon as possible "

And in other ridiculously insigtful news, I'm thinking a better strategy would be for AMD to start making money and stop the losses. I also think they should migrate to 45nm quickly, release Bulldozer ASAP, release fusion ASAP, heck they might as well even release 32nm as soon as possible.

He makes it sound like AMD is just sitting aroung taking their time on releasing a 9950... does he really think they are doing this by choice!?!

Anonymous said...

"I think, introducing tri-core will see AMD through this year, breaking even by Q4, before it is hit by Nehalem speed train."

Arguments can be made both ways, personally, I see tri-core hurting more than helping.

As most informed people understand, AMD's tri-core is a means of increasing the earning potential of a 283 mm^2 die that would otherwise go to waste. If 3 or 4 cores are functional and one is defected, then disabling the defective core takes a non-revenue generating die and turns it into a revenue generating die.

But the caveat is this.... say AMD gets an ASP for their dual cores of 65 bucks on Brisbane -- (this was their ASP for Q4 07 for mostly dual/single core -- roughly). Then say they get 135 ASP for tri core and 185 for quads. 65 / 126 mm^2 vs 135 / 283 mm^2 vs 185 / 283 mm^2 -- or 0.67 vs 0.47 vs 0.65

The revenue per mm is much less where AMD must shoe-horn the tri-cores between quads and duals. This is advantageous to AMD so long as they are always selling defective cores ... but the risk comes in the form that the moment they generate market demand for a product that they 'cannot make' they could shoot themselves in the foot. Say they improve the yields such that to satisfy tri-core demand they must start intentionally disabling a functional quad core ... point blank, they are taking money off the table.

Add to this that the market will ultimately adjust to the performance of dual-tri-quad, and now AMD has put price pressure on it's own dual core (smaller die).

Not that tri core is not a good idea, but it is risky and can end up backfiring on them they way they chose to bring it to market.

Ho Ho said...

anonymous
"He makes it sound like AMD is just sitting aroung taking their time on releasing a 9950... does he really think they are doing this by choice!?!"

What if customers aren't demanding them and AMD chooses to produce lots of low-clocked CPUs to flood the market instead of making some of them clock higher to raise ASP?

Tonus said...

"when you are CONSTANTLY talking about the potential future products may have and spend no time on the current actual products, that's a strong sign."

"He makes it sound like AMD is just sitting aroung taking their time on releasing a 9950... does he really think they are doing this by choice!?!"


Both of those struck me as well, and seem to indicate the growing frustration that many are feeling with AMD (fanatics moreso than the rest, but I'd still rather see AMD producing good CPUs than foundering like this). The pattern lately has been 'wait until AMD releases [item]!!!' and then 'well okay, so [item] wasn't all that great, but wait for the next [stepping/speed bump/technology/etc]!!!'

Given that, I think that statements like 'so now, all AMD has to do is release a 5GHz/10 Watt chip and they're back in business!' are mostly a way of venting at a company that has been over-promising and under-performing of late. And that's probably a lot nicer than what is being said by people who have held on to their AMD stock this long...

Anonymous said...

Intel earnings:
Revenue up 9% YOY, Gross Margin up 4 points YOY, Operating Income up 23% YOY, Net Income $1.4 Billion; EPS 25 Cents

Unknown said...

9.7Bn revenue on the quarter Intel was supposed to BK! Sharikou, what happened?!

Roborat, Ph.D said...

...we can at least give AMD a small amount of praise for getting the B3 stepping out the door. It's a small step on a long road.

AMD doesn't deserve anything for taking 3 steppings to make something work properly the first time. AMD's B3 stepping is about as equal as Intel's Rev -1 engineering silicon. You don't reward delays and mistakes especially when you're behind at least 2 years in overall performance.


We can finally settle the question of whether AMD's 65nm process is broken. AMD's fastest 65 watt, 90nm K8 runs 2.6Ghz while the fastest 65 watt, 65nm K8 runs 2.7Ghz. So, the 65nm process is at least a little better than the old 90nm process.

normally it is incorrect to judge process health based on speed alone without considering yield spread, but if a desperate AMD could only muster a 2.7Ghz in a time where a higher bin gives better ASP, that says a lot about how bad their process is.

People still keep clamouring for Ruiz to step down. Frankly, I doubt Ruiz had any direct involvement with K10's design or development ...

Of the $Billion other reasons why Ruiz is being pushed to go, he chooses the one that is least relevant. Compelling argument.

I think a better strategy would be for AMD to get the 2.6Ghz 9950 out the door as soon as possible and try hard to deliver at least a 2.6Ghz Shanghai in Q3...

this is a negative information paragraph. this paragraph is a step backwards, it states the obvious, it doesn't provide any real answers and a waste of time. when our minds are at the point of looking for ideas on what AMD can do to minimise loses, this paragraph suggests fantasies as if it was an alternative solution. incredible.

Anonymous said...

"Revenue up 9% YOY, Gross Margin up 4 points YOY,"

What was good about the margins is that they included the rather large hammering Intel took on flash price erosion. If they can spin off that damn flash biz - the margins will be back to the hay days.

At least the street after hours realized that AMD's miss was not representative of the entire industry. Intel's guidance was also fairly positive. All in all a pretty good quarter given the state of the US economy (though this is now less than 40% of Intel's CPU biz)

SPARKS said...

“I'm long past hoping that these people will see these two companies as separate entities in very different stages of health.”

As an INTC shareholder for over 10 years I can tell you without reservation Wall Street has, and always will, look at both companies as permanently glued at the hips.

The seesaw market share battle between the two companies during 2006 is a case in point. The street, while buying into AMD’s success, also bought into the notion that INTC was fat, over bloated, past it’s prime, and incapable of presenting anything new and innovative to the market. They truly believed that INTC was out of fresh ideas and AMD would continue to surpass them technologically.

The street didn’t believe the big hulking giant could turn direction so swiftly and so nimbly. Market share was the key indicator. Back then. AMD was gaining AND making money. (This is why Wrector Ruinz tried so desperately hold on to market share AT ALL COST during FY 2007.)

Basically, the street sold INTC short and thought AMD would come back with a product that would break the INTC market share strangle hold. That was Barcelona. It was bad idea, simultaneous process and architecture change, demonstrated by the current 3 core conundrum.

In fact, the belief in Core architectures performance superiority didn’t gain traction, well past its introduction. This is why AMD spun Barcelona so hard and for so long. They, those on the inside of AMD, (and most of us on this site) knew the real deal. The industry insiders on this site (read: geniuses) read through it like a goddamned cheap novel.

More to the point, AMD has always been riding on INTC’s coattails simply because of the market share ratio, hence, if INTC does well, so will AMD just by the numbers and volume alone.

Today is no exception, as INTC was up $1.22 while AMD rose 29 cents for NO apparent reason other than INTC performing well! Go figure. This flies directly in the face of what the analysts are calling AMD’s “broken” business model, merely two days ago.

Historically, AMD has been IBM’s second source solution to INTC which was why AMD was given the license initially. It glued them at the hips from the start.

However, times have changed. In the past AMD BATI (before ATI), INTC could never arbitrarily pull AMD’s x86 license without swift and immediate monopoly legal intervention. INTC would be DIRECTLY responsible for putting AMD out of business.

But, what anonymous said was absolute fucking GOLD and irrefutable.

“And let's be straight here, if AMD dies, AMD finished off AMD, not Intel. Just 2 years ago this was a healthy company on the rise”

This is the difference between AMD now and AMD BATI. I truly believe AMD has opened a huge gape in the monopolistic dynamics of their 20 plus year relationship with INTC. Like some failed Siamese twin operation, AMD clearly tried to separate itself from INTC with a VERY expensive and risky procedure. The operation failed so miserably that one patient is dying and hemorrhaging without its monopolist life support care givers. The other is recovering quite well without its weaker anchor of a partner.

The evidence will be clear under forensic financial examination. The weaker partner made terrible mistakes by voluntarily going it alone and separating itself from its larger more powerful partner.

This is precisely what the street has not taken into account with the AMD/INTC dynamics as of late. After all, you can’t blame INTC for the 5.4 billion mistakes AMD made outright, entirely on their own. INTC could and will LET THEM DIE without legal prejudice. The law cannot ask INTC to subsidize AMD for its stupid and fatal errors in judgment. It would be historically unprecedented. Now your point……

“I'm long past hoping that these people will see these two companies as separate entities in very different stages of health.”

……will be taken quite seriously.

INTC has waited a long time for this and AMD did it themselves.

What fools.

SPARKS