Interesting stock option exercises by Intel execs last week. Check out Yahoo for some news releases today about Otellini and other Intel execs exercising options at a strike price of $17.42 and selling them for a market price of somewhere just shy of $22.

The reason this is significant is that these stock options expire after 10 years, and 1997 was the first year that Intel expanded its stock option program to all employees. So for the first time I can verify that these options are exactly the same strike price as what I received back at that time when I was at Intel, and reflect the actual stock market value at the time of grant (of course, they got about 2 orders of magnitude more grants than me ;).

I think it's worth pointing out that for the next few years, the stock option gravy train should finally be drying up for Intel execs as we near the 10 year anniversary of the tech bubble. Here are the INTC employee stock option strike prices for the next 5 years:

Granted 1998, expire 2008: $19.00
Granted 1999, expire 2009: $30.70
Granted 2000, expire 2010: $61.19
Granted 2001, expire 2011: $25.69
Granted 2002, expire 2012: $29.33

I remember back around 5 years ago, there was a lot of shareholder unrest over on the SI board because of how well Intel management was making out on their $4 options after the stock had tumbled from $70 to $20. Some sane voices tried reminding them that these execs were being rewarded for the increase in share price from 10 years earlier, and the execs actually lost out bigtime because they didn't exercise earlier, but emotion won the day.

If any such people still hold shares, the good news for them is that Intel management interests should now be fully aligned with shareholder interests, because after next year there won't be any big paydays unless INTC stock is significantly higher than it is now...

Has anyone bought any puts?

I bought some Jan08 12.5 puts right before Thursday's close.

http://www.investorshub.com/boards/read_msg.asp?message_id=18948352

Actually, I've been following that thread the last few days (first time in a long time), and the tone there is sounding very much like this thread. Even Dan3 has been wondering why AMD's stock isn't in the single digits.

Almost makes me wish AMD management were shown the door to bring some fun back to this little rivalry. Right now it's like pissing on a corpse...

(Then again, given my positions in INTC and AMD, I think I like the current situation just fine, thank you very much...)

Thanks for the replies.

Like Elmer, I don't know much about packaging, but I would hesitate to proclaim that an endeavor to create a Barc MCM would be terribly expensive, at least initially on the processor side. You get a couple packaging experts and a few architects/designers together, and you see what you can do. I know that the Intel folks who first put Smithfield together was a small team, almost skunkworks in nature. And I remember they got some division-level awards for taking a chance and going from concept to production in something like 7-8 months. Intel was desperate to respond to AMD's dual-core threat, and pulled out all the stops they could. I can't imagine how desperate AMD must be right about now...

If AMD has working Barc silicon to play with right now (I know, that's an IF we're not sure about), they could start working on such a design now, and still have 8 months to get something out the door and beat Intel by 6+ months. Yes, it would be huge, bulky, expensive, likely boutique, and not very "elegant" (I'd say that describes Smithfield to a "T"), but I don't see how we can criticize that if it provides a measurable significant benefit for customers.

The bigger question is probably on the motherboard/chipset/memory controller side. A few months ago I probably would have dismissed this scenario out of hand because of all this necessary collateral, but with ATI's resources at their disposal, they are suddenly better equipped to create their own ecosystem.

I think I can envision a best-case scenario where AMD pushes something out the door around the beginning of 2008, but AMD is obviously not executing to best-case scenarios, so this is probably a moot point from an investment perspective.

But if I were AMD, and I knew that Barcelona wasn't going to substantially leapfrog Penryn, I would at least be investigating this angle to more effectively compete with Intel next year.

Comments welcome on this potential risk factor for Intel in 2008 (as well as a potential sign of life for AMD):

Market acceptance of quad-core processors in the server space is much higher than I ever anticipated. It is very possible that we are entering the era of the "core-wars" in the server space, in a bid to pack more and more transaction power into the same or even lower price points to feed the Googles of the world. I think Intel's drastic gains in server market share is in-part attributable to this phenomenon of having a product that delivers ~twice the performance at a given price point.

We already know from Pat Gelsinger's last presentation on Nehalem that they are planning to release 8-core versions of that design, which implies a 2-die MCM of the "native" quad-core design.

The opening I see for AMD is that Nehalem isn't slated for release until H2 2008. Even if AMD has a quarter delay over their latest estimate and doesn't sell Barcelona until Q4, this still provides them a minimum of 6 months to put 2 Barcs together in an MCM and claim a substantial lead over Intel in providing an 8-core solution. Recent comments by AMD that they regret not designing an MCM with Opteron lead me to believe this is a likely possibility.

My fear is that 6-9 months from now, we will be looking at server performance graphs that show AMD performance almost 2x that of Intel, solely due to multi-core advantages, essentially reversing the huge advantage that Intel has enjoyed the last 6 months.

Any thoughts/rebuttals?

Half a billion. Poof. Gone. Damn, these guys are good....

This is the company Richard Pryor should have turned to in the movie Brewster's Millions. AMD would have solved his "problem" in 5 days...

I woke up in a gambling mood this morning.

Bought some AMD Jan08 12.5 put contracts about 10 minutes before the close.

I have no idea what will happen during the conference call or in the immediate days after, but what it comes down to in my mind is I don't think anybody can do anything to keep AMD's stock from being significantly lower than $14 by the end of next quarter...

I plan to hold these for at least two months (unless a best-case scenario develops and negative reaction is swift and deep).

Enjoy the show everyone! <g>

Things had seemed a little less fun over on the SI AMD board now that niceguy has turned on AMD, but it's nice to know that some true believers still remain to bring some laughter to our day:

http://www.siliconinvestor.com/readmsg.aspx?msgid=23470848

I previously stated on this board that my AMD put-buying days were over. But dammit, at $14 my resolve sure is weakening.

Anybody thinking of pulling the trigger prior to AMD's earnings conference call? My fear is that this quarter's stunning losses might have actually gotten it through Hector's thick head that their current business plan is broken, and we might see some announcements on lowered capacity expansion. If they announce any drastic steps like that, it could put a limit on the stock's downside.

On the other hand, if they stubbornly maintain that all they need is Barcelona and some cost reductions, and will figure out their cash needs later, I will probably go in for another round of puts.

I just can't help wondering, though, whether we will ever see $14 again in the foreseeable future regardless of what they do...

Paul is seeing an overall trend away from MP and towards DP in servers due to the competitiveness in performance of quad core processors, and exceptional price and power characteristics.

I have been wondering for some time whether this might happen as the core race proceeds. I think I asked on here some time ago about whether Nehalem's CSI interface would facilitate putting two 4-core die in one package for an 8-core chip (it now appears Intel is planning on this based on the Nehalem details released a couple weeks ago).

If you can build a 16-core server in a 2-socket platform, do you really gain much by the added system complexity of 4 sockets?

I wonder if the more profitable way of thinking of 'SoC' is "Server on a Chip". I see the same dynamics in play in terms of system integration, lower system cost, lower power, etc.

Of course, Paul might just be saying this because Intel's 4-socket solutions aren't nearly as competitive right now as the 1S and 2S products.

Sounds like analyst estimates were for $9 billion in revenue and .22 EPS. Intel came in at $8.9 billion revenue, and after taking out the tax benefit hit .22 EPS. Almost right on the money of expectations. Even their 50.1% margins were nearly exact with their 50% estimate. Yawn, what boring execution! <g> Unfortunately I'm afraid Wall street likes a little flash (not to mention growth!).

The bad news is that Q2 revenue guidance of $8.2 to $8.8 billion means the analysts' estimates of $8.8 billion are likely high. And margins are guided to fall further from Q1 to 48%.

One plus about Q2 is that last year's Q2 had revenues of $8.0 billion, so the comparable quarterly revenue should finally show a little growth.

Generally, Intel is holding its own in this all-out war, as us bulls felt they would. But I'm left thinking that the investment thesis from here for much stock appreciation this year is if Q3 rolls around and Intel's product lineup, pricing, and competitive position are so compelling that a higher-than-seasonal uptick in demand takes place.

If Q3 revenue can approach $9.5 billion, and Q4 rise from there to say, $10.5 billion, maybe, just maybe Wall Street might start seeing signs of growth and reward the stock a bit.

If that doesn't happen, and we continue looking at no-growth forecasts, I will be inclined to write the BoD demanding that they triple the dividend and accept reality that growth in x86 is dead and to at least reward investors with a 5% yield from the cash cow they control....

I guess I'm just impatient....it's been a long 9 years. ;p

One concern I have is that on the slide describing the Enhanced Dynamic Acceleration Technology (I sure wish they would call this something simple like turbo mode), there's a little bubble that says, "Applicable in power-constrained mobile environments."

I hope this feature isn't confined to mobile SKUs. The idea of giving one core a boost when other cores are idle seems like a no-brainer to me as we begin the long transition to multi-threaded apps. I don't think there was ever a more clear-cut opportunity for a marketing message that can promise (and deliver) both improved performance now and in the future as new software comes online.

Duke, if you look at last quarter's earnings release, Intel breaks down their revenue into 4 global regions. Of their $9.7 billion in revenue last quarter, the breakdown was as follows:

Asia-Pacific: $4.9 billion (50.5%)
Americas: $2 billion (20.6%)
Europe: $1.9 billion (19.6%)
Japan: $936 million (9.6%)

So all of North and South America combined was only 20% of revenue. If you assign SWAG estimates for US versus Canada/Mexico (80%?), retail versus corporate (40%?), and retail desktop versus notebook (60%?), US desktop retail is likely much less than 10%, and I wouldn't be surprised if it's less than 5%.

Of course, one way to read those marketshare gains is that their revenue from US retail desktop has almost doubled from a year ago, so we could once again start to approach that 10% number, and boost overall revenue by a few percentage points over last year due to this one segment.

One interesting item from last quarter is that the $2 billion revenue in the Americas was a 6% increase over Q3, so we might be starting to see this effect.

About the only other number to offer in answer to your question about segments is that I remember reading somewhere that 50% of all corporate purchases these days are notebooks, and that number will probably continue to rise.

I find it striking that over half of Intel's revenue is in the Asia-Pacific region. I don't think I had fully internalized that before typing this post....

I'm personally not looking for much impact to the stock price either way due to Q1 earnings. If anything, keeping within past guidance should be a positive in the midst of AMD's failures.

My eye is on the Q2 earnings report, and more specifically, the outlook for Q3 and H2. Remember that margins are expected to rise in the second half, and if that expectation is maintained in the midst of AMD's low point in profitability just before Barcelona, I think we could start to see that magical sentiment turning you mentioned earlier.

wmbw, weren't those calls you purchased Jan'08 22.5s? Personally, I would hold on to those through at least Q2 earnings, as I think this run-up from $19 will offset any time degradation you'll see until then. Plus I think the odds are very, very good that they will be in-the-money by Q3 at the latest, providing a nice pop in their value.

Those were the exact same calls I was contemplating buying, and was looking to sell them around August-September with the stock in the $23-25 range. Of course, I chickened out and didn't buy them, so feel free to ignore the geek in the cheap seats...

Here's a fun factoid: I sold my last AMD puts shortly after Q1 earnings when the stock dropped below $17, and bragged on here about my 178% gain. Those puts actually expire April 21st, and if I had held on to them I could now sell them for a 320% gain and almost doubled my money again. Oh well, such is the fun we have following options. ;)

It won't last. Investors are encouraged that AMD management is taking things seriously by restructuring their business model, but in the long term, things are really going to suck. As this dawns on investors, the stock will head down to single digits. I think AMD puts are starting to look more interesting. I'd like to see how high the market will take it, and maybe I'll buy some.

Yes, I have to admit I would likely re-think my stance if the stock somehow gets back up near the $15 range.

The problem with proclaiming that single digits are inevitable is that the closer we get to that point and get closer to tangible book value, the more players enter the picture that are unknowns to me. I'm talking about entities like hedge funds, private equity investors, etc. Once the buzzards start circling, I just feel like it's time for the small fries like me to step away from the carcass...

Interesting price action in AMD stock today. Revenue is coming in a whopping $500 million less than original projections (a 30% miss), and $300 million less than the consensus estimate after their first warning, yet the stock is up 5% so far today.

I had been toying with the idea of buying another round of AMD puts prior to their earnings release, but have been worried that the mental and/or emotional support level was too much at these levels. Today validates those concerns, and I am inclined to continue watching this from the sidelines.

And yet, things sure are looking horrible for AMD. We now know Q1 was really bad, but I don't see how Q2 can be anything but downright ghastly. Could they actually drop under a billion in revenue during a seasonally weak Q2 as Intel looks to complete their Core 2 ramp? I think that's in the realm of possibility. But does the stock price care? I just can't get a read on that.

My inclination is to resist my urges to further cash in on AMD's woes, and now look for Intel's stock to finally reap the rewards of their competitive advantage this year. Beamer's INTC call purchase back at $19 was a smart move, IMO, although I just couldn't bring myself to do the same given my substantial long position.

Good luck to all INTC longs, I think we will finally be in for some financial reward by the end of 2007.

I have a hard time getting worked up over anything that guy says (even if it's positive for Intel) after he wrote this laugher:

http://www.247wallst.com/2007/03/maybe_intel_sho.html

Simply amazing.

How blind must somebody be to think that Henri Richard wouldn't give his right nut to be in the position that Intel is in today. I can't believe people actually buy this BS about AMD only wanting their fair share of the pie, and once they get it then everybody will be in nirvana.

If AMD somehow turns things around and gets to 30% market-share, you can bet their next goal will be 40%, and then 50%. And you can be damn sure that AMD will leverage their newfound market strength to increase ASPs and profit margins. This crying about the plight of PC makers is hilarious.

What's humorous to me is that AMD's stated long-term goal is to achieve 50-55% gross margins. At the same time, they are accusing Intel of squeezing all the profits out of PCs. Yet Intel estimated their gross margin this quarter to be 50%. This does not compute!

One more random observation: Although I think the term "price war" doesn't actually apply in this current market, one way that it looks like a war is how the cheerleaders on each "side" are viewing their competitor with regards to a measly 5-10% share. All the 'droids go on and on about how Intel is needlessly hurting AMD, and if they would merely raise prices and give up an additional 5% MSS to AMD to get to their magical 30%, all would be well for everybody and profits would abound. At the same time, the Intel folks moan and groan about how AMD should simply give up and go back to a niche market with 20% MSS, where they can have sustained profitability.

I just find the sight of each side yelling "just give up!" while pressing on the fight somehow amusing. In that respect, I guess we really are in a war of sorts.

(Unfortunately for AMD, they brought a knife to a gun-fight. And their gun (Barcelona) should arrive just in time for Intel's cannon (Penryn), and next year Intel breaks out the nukes (Nehalem). Maybe AMD really doesn't have any other alternative than to go crying to sympathetic "journalists").

Tick Tock, Tick Tock....

Today marks the second datapoint in Intel's tick-tock strategy of releasing a new chip design every year.

The first data-point was the jaw-dropping performance claims of Merom that were presented around this time last year at IDF. Such claims were met with skepticism, and explanations that AMD would have until June to catch up, and cries that "anything can happen in 6 months!"

We all know what happened in the year that followed.

We now have the second datapoint to measure this strategy by over the next year. Already these claims are being met with skepticism, and a hilarious claim by Petz over on SI that FP divides aren't used in real applications.

While AMD has been preparing their best shot to respond to last year's products from Intel, Intel has just upped the ante and given them a bigger bar to shoot for before their first response is even ready (and I can only imagine how motivated the Penryn team must be to nail everything down in time to launch even before Barcelona does....how big of a coup would that be?!?).

Now AMD will be forced to somehow respond to this higher standard. Yet, one has to wonder what surprises might be in store when specific details about Nehalem are released around this time next year. How will they respond? How can they?

And how can they respond in early 2009 as the 32nm Nehalem shrink (Westmere) prepares for launch?

And how can they respond in early 2010 when further details on the next-generation Gesher design are released?

This steamroller is rollin', anybody that's in the way better look out....

More details from anandtech:

http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=2955

Some excerpts:

Intel EDAT: the End of the Multi-core Clock Speed Disadvantage?

Intel also talked about its "Enhanced Dynamic Acceleration Technology" which is effectively integrated overclocking based on load. If you are running a single threaded application (or a multi-threaded application that's predominantly using a single thread), Intel's EDAT can power down the second core and increase the frequency of the working core to maintain the same thermal envelope at all times.

Intel's EDAT could spell the end of the clock speed differential between single and multi-core processors. With all cores running workloads, the multi-core system would be clocked lower, but when some cores are idle the chip could potentially run at the same speed as a single core solution would. Single core designs have pretty much disappeared from roadmaps already, but considering there are still applications that are single threaded in nature and benefit more from clock speed improvements, future processors will offer both options in a single package.

.....

Performance

Comparing a 3.2GHz Penryn (1.6GHz FSB) to a 3.0GHz Conroe (1.33GHz FSB), Intel has measured more than 20% increase in gaming performance (with no code changes). For video encoding applications, if SSE4 is utilized, the same Penryn vs. Conroe comparison can offer more than a 40% increase in performance.

Finally, Intel mentioned that in the server space, the fastest quad core Penryn available (>3GHz) vs. a 2.67GHz quad core Xeon resulted in a greater than 45% increase in performance in "bandwidth and FP intensive applications". It's incredibly vague (and oddly similar to AMD's claims of Barcelona vs. Xeon performance), but Pat mentioned that STREAM and certain benchmarks in SpecFP could be considered to be "bandwidth and FP intensive".

.....

Nehalem Micro Architecture: Intel Embraces the IMC and IGP

Surprisingly, Intel gave away quite a few details about Nehalem. Although Nehalem is still based on the 4-issue Core architecture, it takes "multithreading" to a whole new level. First of all, Nehalem can contain up to eight cores per die. Combined with 2-way Simultaneous Multi-Threading (SMT or Hyper-Threading), you'll have the ability to execute up to 16 threads on one chip!

Nehalem will no longer use a FSB but a serial point to point interconnect. Even more revolutionary is the fact that Nehalem will have an integrated memory controller (IMC) and that the number of serial interconnects is variable (Intel's version of "HyperTransport"). Another potentially groundbreaking move is that some Nehalem CPUs will have a GPU integrated (Intel's version of "Fusion"). With an integrated memory controller, new interconnect, and potentially integrated graphics, Nehalem will obviously require a new socket.

Intel would not give any more detail, but it is clear that the GPU will not be high-end (that would require too much power); more likely it will be a kind of midrange (or even low-end depending on your perspective) solution. Intel would not confirm this, but it seems pretty clear to us that Xeon DP and desktop products will probably have an IMC that supports DDR3. Xeon MPs will probably have an IMC that supports registered FB-DIMMs with DDR3. Nehalem should first be available in the second half of 2008 as Intel talked about "production ramping in 2008, with full production in 2009".

In the spirit of piling on to all this AMD doom-n-gloom talk, I am curious if anybody thinks that AMD might need to write-off any of that multi-billion-dollar goodwill that got put on their balance sheet from the ATI acquisition, when we hit the one-year anniversary of that purchase in Q4.

My (limited) understanding of the financial treatment here is that in times past, goodwill could be amortized down over time. However, now companies are required to keep that goodwill on their balance sheet in its entirety, and review that value annually. If the value is deemed less than what is on the balance sheet, then the difference needs to get written off as a hit to earnings.

I welcome any corrections from those more financially astute than I.

Although I tend to join with those who think single digits are a distinct possibility, I don't consider it a slam-dunk like I did when I previously bought puts when the stock was at $29 and again at $24. I think it's because I fear the psychology of the market may have created a concrete support level, and am afraid that the only ones left holding stock are the true believers who will weather whatever storm is coming, no matter how bad it gets. This fear was intensified when AMD warned for Q1 and the stock did basically nothing.

I keep waiting for a little spike in price to the $15-$16 range, at which point I may take the plunge with some puts. But the stock isn't cooperating, and if it keeps inching down I may sit this one out...

Intel did not dominate X86-silicon out of the starting-blocks.

Oh really? In case the other laughing (and well-deserved) responses aren't educating you, allow me to spell it out for you:

Intel invented the x86 market, and supplied 100% of all IBM PC systems, as well as all clones that soon followed from Compaq and others. It wasn't until the 386 generation that AMD reverse-engineered Intel's design and sold a copycat product (or was it the 286 generation? Getting old, memory foggy).

The tale of how Intel and Microsoft inked deals with IBM that allowed them to sell to other suppliers has gone down in technology lore as enabling the creation of the entire PC industry as we know it. To be a follower of this industry and not be aware of this history is truly mind-boggling...

Thanks for the additional differences, and you're right on all counts.

I still think that this is a race that's going to be too close to call, and the fact that AMD still hasn't shown real performance data (like Intel did 6 months before Core 2 was launched) is very telling to me....

I'm still in disbelief that they could have a product only 3-6 months away

I am giving them the end of the month until I put my money where my mouth is and buy more AMD puts. If there is no substantial news before then regarding performance or demo parts or whatever, then any news that comes out after that can be only bad.

I see that they are presenting at a Goldman Sachs Technology Conference this coming Monday, so perhaps we won't have much longer to wait.

Then again, we just might. But I'll wait only so long before I place my bets against the company. Maybe their presentation will cause another baseless jump like last December, and provide a good entry point.

This article gives good, balanced insight into the differences between Opteron and Barcelona. I don't really see anything in it that I disagree with.

Of more importance, however, is what the difference will be between Barcelona and Core 2. This isn't addressed directly, although you can glean hints from snippets of this article. Let's look at each of the key Barcelona improvements that are outlined:

SSE128:

The culmination of the SSE128 improvements is very similar to some of the changes made in the Yonah to Merom transition.

Branch prediction:

Barcelona adds a 512-entry indirect predictor which, believe it or not, predicts indirect branches.

The K8's branch predictor was quite good and very well optimized for its architecture, but there were some advancements Intel introduced in the Pentium M and Pentium 4 that AMD could stand to benefit from.

Stack management:

Intel's very first Pentium M introduced a feature Intel called its dedicated stack manager.

With Barcelona, AMD is introducing a similar technology it is calling a Sideband Stack Optimizer.

Load/Store ordering:

One major aspect of Intel's Core micro-architecture advantage is its ability to allow load instructions to bypass previous load and store instructions.

Core also allowed for loads to be moved ahead of stores, which was previously not allowed due to the possibility that an earlier store could invalidate the data that was just loaded.

Barcelona can now re-order loads ahead of other loads, just like Core 2 can. It can also execute loads ahead of other stores, assuming that the processor knows that the two don't share the same memory address. While Intel uses a predictor to determine whether or not the store aliases with the load, AMD takes a more conservative approach. Barcelona waits until the store address is calculated before determining whether or not the load can be processed ahead of it.

Divide latency

When Intel introduced its second Pentium M, codenamed Dothan, one of the enhancements made was a lower integer divide latency. Although details at the time are slim, AMD has indicated that it has moved to reduce integer divide latency in Barcelona as well.

Prefetching

More recently, when Intel introduced its Core 2 processor family it stressed the importance of its three prefetchers per core in drastically reducing perceived memory latency.

AMD's K8 core had two prefetchers per core - one instruction and one data. The Barcelona core still retains the same number of prefetchers, but improves on them.

Cache

These days two Core 2 cores share up to 4MB of L2 cache, while the fastest offerings from AMD weigh in at half that. The gap will continue to widen with Barcelona, as each of its four cores will only have a 512KB L2 cache. While a quad-core Barcelona chip will have 2MB of total L2 cache for all four cores, a quad-core Kentsfield currently has 8MB of L2 cache for all four cores. By the end of this year, Intel's Penryn is expected to have 12MB of L2 cache for all of its cores.

In order to keep die sizes manageable, AMD constructed its quad-core Barcelona out of four cores each with a 128KB L1 and 512KB L2, much like most mainstream K8 based products today.

With four cores sharing a single die, AMD didn't want to complicate its design by introducing a large unified L2 cache. Instead, it took the K8 cache hierarchy and added a third level of cache to the mix.

=============================

Other areas mentioned that I would expect to remain mostly neutral versus Core 2 are the TLB improvements, and power management features.

I see only two improvements mentioned in this article that aren't already addressed in current Core 2 processors: support for nested page tables to help with virtualization, and an improved memory controller. The memory controller advantage has always been there, though, and Intel has always kept up with its bigger caches.

Not mentioned in the article is the fact that Core 2 added an extra execution port to make it a 4-wide machine, while Barcelona is still a 3-wide machine.

Oh, and clock speed? Fuggedaboutit, Intel remains king here.

In my mind, this article confirms that Barcelona will simply allow AMD to catch up to where Intel is today. Any claims of a significant performance advantage are hogwash. I think AMD knows this as well. A recent interview said that the performance advantage would be "in the double digits", which to me means that AMD is projecting a performance advantage of about 10% versus Core 2. I think we could very well see some applications with such an advantage compared to Woodcrest, but I also think there will be a fair number of applications where Woodcrest pulls ahead by 10% or more.

Any AMD diehards who are looking for Barcelona to be their savior are going to be disappointed, IMO. And Penryn is right around the corner.

I think that slides 7 and 8 in this presentation give the clearest picture yet of what AMD is up against:

http://media.corporate-ir.net/media_files/irol/10/101302/2_21_07_presentation.pdf

The clueless droids see this long list of Barcelona improvements, and start applying a flat 1-2% performance improvement for each bullet item, and before you know it performance has doubled. It's laughable. The fact is, each architectural improvement will provide a varying performance boost on certain applications. You put enough of them together, and you end up with a modest overall performance improvement.

I have heard from folks who have closely investigated Barcelona's architecture that they are expecting about an overall 15% improvement in integer IPC over Opteron. Some focused FP applications (specFP_rate, linpak) could see significant increases, and you can bet AMD will hype this to no end.

Of course, nevermind that AMD will see minimal financial benefit from Barcelona until '08, and any practical performance comparisons will need to be made with Penryn and then Nehalem (and then Westmere, and then Gesher, and on it goes...)

Pass the pipe, please....

Hey Duke,

I seem to remember that Intel had previously mentioned that this issue regarding tax credits for export sales had been resolved. A quick search yielded these results:

The Q3 2005 mid-quarter update:

http://www.intel.com/pressroom/archive/releases/20050908corp.htm

"Intel's tax rate forecast for the third quarter was approximately 30.5 percent. The third-quarter tax rate is now expected to be impacted by additional taxes of approximately $250 million, plus or minus $25 million, related to a potential repatriation of approximately $6.3 billion of accumulated income earned abroad. The company’s tax rate forecast for the fourth quarter is unchanged at approximately 30.5 percent."

And here is the Q3 2005 earnings report:

http://www.intel.com/pressroom/archive/releases/20051018corp.htm

"As discussed in the company’s Sept. 8 Mid-Quarter Business Update, Intel’s results for the third quarter included the impact of an increase in taxes of approximately $250 million, equivalent to 4 cents per share, related to the decision to repatriate foreign-earned income under the American Jobs Creation Act (Jobs Act). The company’s results for last year’s third quarter included a tax adjustment that increased earnings-per-share by 3 cents. Intel’s results for the previous quarter included a tax adjustment that increased earnings-per-share by approximately 2 cents."

Reading ahead in this same release, it now looks like Intel was continuing to assume that they would get some tax credits based on what they consider foreign export sales. Perhaps this really has gone against them:

"The fourth-quarter tax rate is expected to be approximately 31 percent. The tax rate expectation is based on current tax law and current expected income and assumes Intel continues to receive tax benefits for export sales. The tax rate may be affected by the closing of acquisitions or divestitures; the jurisdiction in which profits are determined to be earned and taxed; changes in the estimates of credits, benefits and deductions; as well as the taxes associated with repatriation of cash under the Jobs Act; the resolution of issues arising from tax audits with various tax authorities; and the ability to realize deferred tax assets."

I have no idea if this is related to this current issue. But at least here's some more reading for you....

Check out some investor sentiment at:

http://caps.fool.com/Ticker.aspx?ticker=AMD

This is a stock-picking game started by Motley Fool. In this case, 1109 people have picked AMD to outperform the market, while 239 currently have it picked to underperform. Most of the recent positive comments focus on rudimentary technical analysis (ie, surely it can't go any lower!). Here's a real doozy, which appears to use reasoning that is simply made up:

"All of AMD's currently available processors outperform Intel when used with the 64 bit Microsoft Vista OS. When certified clockspeeds are timed using Vista as the OS later this year, AMD will outscore Intel without changing anything. Lower cost, lower power use, better performance. The Vista roll out will take about one year. Within that time AMD will have finished its merger and produced its first ATI derived product. Also, Dell will probably buy its video processors from AMD after the merger settles. I have this at $30 in 1.5 to 2 years."

This comment got 2 recs. I love it...

Let's face it: the market is full of suckers.

Anyone have the latest institutional holding data?

When I bring up AMD on my Ameritrade account, it says that institutional holding is at 72.72%. I have no idea where it gets that number from.

Here's one question I have: as we're all sitting around having giddy discussions about the future doom-n-gloom of AMD, does anybody have an approximate share price that would represent asset value or liquidation value? Ameritrade reports the current price as having a price/book value of 1.27x. I'm assuming that includes their current cash hordes, which we all suspect are flying away rapidly. If the current competitive environment continues long-term, what's the theoretical bottom that the share price should never dip below? (obviously, any such estimate would assume that share count remains constant with current levels, which I don't think any of us believes is likely).

I'm still feeling greedy and eyeing more AMD puts, but am cautious after such a tremendous fall thus far. I can't help but wonder how much future bad news is already priced in to the stock. I might get back in if one of two things happens. Either:

1) The share price spikes up north of $16 again.
2) We get through the month of March with still no word about real Barcelona performance data.

About the only good thing to come from Transmeta was
the spotlight it put on low power x86 MPUs for mobile
computing. It basically forced Intel to hedge its bets
by developing Banias which laid down a lot of the early
groundwork for NGMA.

I think that "forced" is too strong a word. IIRC, Crusoe came out around the Pentium III timeframe, and I believe a low-voltage PIII did just fine against anything Transmeta could offer. The performance was just that bad.

What it did do is test the market, and show that it was worth spending Intel's resources to develop products for this segment of the market. I think Intel would have gone here anyway (guys like Perlmutter in the Mobility group were already pushing hard for this), but Transmeta reduced the risk of doing so, and made it more of a no-brainer, IMO.

In short, I agree with everything you say, but I wouldn't even give Transmeta as much credit as you.

Here's my take on that hack-job of an article...

AMD’s next-generation processor line, code-named Torrenza, has gone from a block diagram to living, breathing silicon. The first incarnation of AMD’s redesigned x86 CPU is Barcelona, that which your non-co-readers will call quad-core Opteron.

So the Torrenza code-name is for a processor line, which is now actual silicon. And that silicon is called Barcelona. Got it.

Barcelona is genius, a genuinely new CPU that frees itself entirely of the millstone of the Pentium legacy. It’ll do the same for you.

Oh give me a f***ing break....

Each of Barcelona’s four cores incorporates a new vector math unit referred to as SSE128 (128-bit streaming single-instruction-multiple-data extensions). I am aware that you only do quantum physics on weekends, but the potential for hardcore IT tasks such as encryption, compression, real-time analysis of high volumes of streaming business transactions, and wire-speed packet analysis is also the stuff of science fiction. Barcelona gives floating point operations their own schedulers (checkout lanes) and runs them twice as fast as 64-bit SSE did. AMD claims that Barcelona’s per-core floating point performance is more than 80 percent faster than the present Opteron. Benchmark that. And separating integer and floating-point schedulers also accelerates this thing called virtualization, which you may notice is a recurring theme for Barcelona.

No mention of the fact that Core 2 already does 128-bit SSE.

I'll give AMD one uarch advantage when it comes to a separate FP issue pipeline. Does anyone know if Opteron already has separate FP schedulers? Prescott had that, and is one way the IPC was maintained versus Northwood despite the much longer pipeline. Unfortunately, going back to a P6 derivative for Core caused this feature to get dropped (for now).

Nested paging tables is a per-core feature that will light the afterburners on x86 hardware virtualization. A paging table holds the map that translates virtual memory addresses to physical memory addresses, and each CPU core has only one. Virtual machines have to load and store their page tables as they get and lose their slice of the CPU. AMD solved the problem with nested paging tables. Simplified, each VM maintains its own paging table that stays fixed in place. Instead of loading and saving paging tables as your system flips from VM to VM, your system just supplies Barcelona with the ID of the virtual machine being activated. The CPU core flips page tables automatically and transparently. This is another feature that’s implemented for each core.

Will nested page tables really have that much of an impact on the everyday user? I admit I am unfamiliar with much of the hype surrounding virtualization.

Much fuss has been made about power efficiency, but the best of x86 power saving schemes is crude. They adjust the clock speed and the operating voltage of the entire CPU, and the selection of set points is small. Barcelona keeps this technique, but builds on it with inspiration from IBM and Transmeta. Barcelona blacks out power to individual portions of the chip that are idled, from in-core execution units to on-die bus controllers. This hasn’t made it into PCs before because it’s very difficult to manage light switches for several “rooms” individually and to make sure that, like a refrigerator light, whenever a door is opened, the light is on as if it’s been burning the whole time. Power savings from these schemes are dramatic. If Barcelona lacked this feature, it would still be a green CPU.

So Barcelona is the first x86 CPU to implement heavy clock-gating? Pure, unabashed fabrication. What a crock of sh*t.

I do admit, one concern I have is that AMD may one day expend the engineering resources to implement as much clock-gating as Intel has been doing since Willamette. If they ever do, they could greatly close the existing power gap.

Unlike Intel’s Core, Barcelona gives each core dedicated L2 cache

There is nothing inherently better about a shared L2 cache versus a separate L2 cache for each core. It's simply a tradeoff between optimizing for multi-core operation versus single-core. A shared L2 cache will allow any single-core application to have more cache available, thus increasing performance. Now consider that Intel's manufacturing advantage allows it to have a shared cache that is almost as big as AMD's 4 separate caches combined, and I'd say Intel is sitting pretty here.

Barcelona incorporates a redesign that reduces cache latency (access delays).

Wasn't the cache latency increased on Brisbane, to account for future larger cache sizes? Why do this if it won't be the same on Barcelona?

Barcelona adds Level 3 cache, a newcomer to the x86 and a page out of IBM’s POWER playbook. All four CPU cores in a Barcelona socket will share a single master catalog of recently-retrieved data. A three-level cache is a must-have for a multicore CPU, and that becomes obvious when you get a demo that switches L3 on and off.

With only 2M shared between all 4 cores, don't expect this to have much impact on multi-thread workloads.

Barcelona is a new CPU, not a doubling of cores and not extensions strapped on here and there. Get ready to be blown away long before its release, which is scheduled for midyear.

Core 2 still has more advanced Load/Store reordering techniques. Core 2 is still 4-wide issue versus Barcelona's 3-wide. Core 2 is available now, for a very reasonable price. Barcelona is still a powerpoint slide. That is the reality.

Have you ever done a lead vehicle?

I was on the design teams for Northwood and Prescott. Both times, I remember other DEs getting pulled off to work on the process test chip. These test chips not only had SRAMs, but also had adders, muxes, rotators, latches, buffers, etc, etc.

You do make a good point about clock skew. The demos may have been done before the skew recipe could be hashed out, and that could be significant.

We know it can run more than taskmanager....

As I said, if this wasn't the case, it would be unusual by Intel's standards, especially for a shrink (primarily) of a design already in production. The only Intel project I've been on where A-0 didn't boot multiple OS's was on Prescott, where a bug in the equivalency checker allowed some data pins to be swapped on a block boundary. A quick dash stepping fixed the problem, but that resulted in several weeks delay in making significant forward progress with post-silicon analysis.

What process files? From SRAMs? That's not enough.

Every design uses a process file that they get from the fab guys. Even if the process won't exist for 2 more years, there will still be a process file to characterize device delays, capacitance, resistance, etc. At that point, it will be just a guess and won't be worth crap, but it will still be there.

As the process is refined, so is the overall process file, and new revisions are continuously fed to the design team. Typically, projects I've been on migrate to a new version of the process file every 3-6 months or so, and everybody is instructed to refresh all timing data, and then the design reacts to any major path changes. This is especially critical (and painful) on a new process that hasn't gone into production yet. By the time a chip tapes out, the pre-silicon timing models are pretty well in sync with the latest incarnation of the process (maybe just a few months behind). Maybe Penryn did it differently than I've experienced, but I doubt it...

SRAMs aren't the only thing that are put on a test chip. You can bet that inverters, NAND gates, NOR gates, etc are constantly being fabricated and then measured for their performance, and reflected in newer versions of the process file.

Fine, so it's much too early to draw any conclusions on 45nm from A0 silicon of the lead vehicle.

Agreed. Which is why I don't think those anticipating a schedule pull-in for Penryn should get too carried away. We just don't know...

I have never experienced any process (even brand new ones) where actual device performance was 30-50% below what was expected on pre-silicon (at least not at Intel). If the worst pre-silicon paths were expected to limit the part to, say, 3GHz, and those paths ended up being 2GHz on actual silicon, then I would say that the 45nm process is in dire straits, indeed...

You can bet that Penryn received a steady cadence of updated process files every few months right up until tapeout. Yes, you can expect some variance, and re-ordering of all the worst paths once you get silicon back. But not such a huge difference between pre- and post-silicon timing.

More likely, I think, is that some path somewhere received a stray override through a poor use of a wildcard (*), or some ignore_path command that slipped through the cracks. All it takes is for one path out of millions to suffer from user error in a tool configuration file, and your entire frequency is tanked. I've seen it happen before, which is why I suspect that now.

One factor that I think could play a role in Penryn's schedule is the low frequencies that are being reported from A-0. Frequencies so below the target indicate to me that an "oops" might have happened on some pre-silicon timing path, like a bad override or something. While likely easily fixed, I suspect that this greatly limits the amount of post-silicon speedpath debug that can be done on A-0, and they may have had to rush out a dash A-1 or even a full B-0 stepping to make further forward progress on post-silicon timing analysis.

I wouldn't expect this to result in an overall schedule slip, however. Often there is an internal "best-case" schedule, along with an external commit schedule that accounts for the inevitable unexpected to occur. When you see a schedule pull-in, you are often seeing a case where everything went right between the design team, the debug lab, and the fab on all the silicon steppings from A-0 to production. I would bet that Intel's committed 2H07 schedule accounts for the possibility of one or two hiccups like this. But in my mind, it reduces the chances of seeing any significant schedule pull-in. Then again, you never know....

(I would also point out that A-0's successful boot of several OS's isn't necessarily a good indicator of production schedule. My experience is that A-0 silicon is functional more often than not at Intel, and I know that functional A-0 is the expectation from the design and validation teams. It's certainly one less hiccup to worry about, though. I have often contended that no other MPU vendor can hold a candle to Intel's extremely thorough and paranoid validation processes. The FDIV bug has done 100 times more good than harm, IMO).

I was talking about short puts. <shrug> Shows what I know. Thanks for the correction.

Yes, I believe you are right. Selling a put ties up quite a bit of cash in the account. But, on the plus side, at least the cash is still making some money in the money market.

Actually, I don't think it's even this stringent. I think (but am not sure) that as long as your potential put obligation is less than a certain percentage of your account balance, then you can have that balance invested in whatever you want. So you can have all your money tied up in stocks working for you, and if the put gets exercised for some reason then the broker may sell some of your stocks to cover your obligated purchase.

So if your broker's required ratio is 50%, and you have a $10,000 account balance, you can write puts that are worth up to $5,000 of the underlying stock. And I'm pretty sure that you can continue to have your entire account balance fully invested if you want. I have not actually done this, so I'm not 100% sure, but I believe this is how it works.

I think (but haven't really checked) that most brokers allow option trading now in just about any type of account. I use TD Ameritrade, and at the time I formed my IRA account last year, I was asked what sort of trades I was going to do, and selected stocks and options. I remember some disclaimer warnings about option risk, and that approval might be required, etc, but I was approved pretty much straight away.

Trading options seems to really be catching on. I've heard a lot of buzz about a brokerage named OptionsXpress, which specializes in options. In fact, the buzz I hear is that OXPS is under-valued and poised for some pretty good growth.

Well, the ride is finally over for me (for now). My limit order was hit today, and I sold all my AMD April 22.50 puts that I bought about 3.5 months ago when the stock was north of $24. Final gain was 178% (minus commissions). (Btw, the ticker on those was AMDPU. Quite a appropriate, I think). The downside is that I invested a pretty paltry sum of money, but the upside is that this is in a Roth IRA, so Uncle Sam won't get any of it.

I'm definitely not looking to stay out for long. I've got my eye on some Jan 15.00 leap puts, so I can continue riding the wave through the year. Will probably pull the trigger on any dead-cat bounce that gets the stock closer to $17. I probably won't wait long, though, if that doesn't happen. I think a $12 price target by Q3 is a very realistic scenario.

I would like to take this opportunity to extend my sincere gratitude to AMD management. <g>

I was referring to these posts you made:

http://www.investorshub.com/boards/read_msg.asp?message_id=16279892

http://www.investorshub.com/boards/read_msg.asp?message_id=16283102

I only have time to follow this site, and from this vantage point it sounded like you raised a trial balloon, then later referred to it as a rumor.

Regarding integrated graphics on Nehalem, I will only say this: I wouldn't bet on it. But don't be surprised if the chip has another trick or two up its sleeve....