Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
I believe that Intel may be using Phase Cange RAM.
Definitely not for SSD.
wbmw,
According to my calculations, and based on the score of 2200 on a system with 8 cores, the clock frequency would have had to been as low as 1.6GHz. This is well below even the HE Opteron, which clocks as high as 2.6GHz.
Good point, but if Barcelona only clocks at 1.6 GHz at 65W, and they wanted to make a comparison at the same clock speed, that's what probably lead to the selection.
But it's all a little weird, because if they wanted to make the comparison at the same power, they should have used the fastest Opteron at 65W, which, as you say, is 2.6 GHz.
Or, if they wanted to make the comparison at the same clock speed, why choose much slower speed, compared to expected launch speed?
What is concerning though for AMD IMO, is limited to non-existant IPC gain over K8...
Joe
wbmw,
I think AMD does not want to show how poor the current clock frequency is with Barcelona, which has got to be pretty bad at this point. So they underclock dual core Opteron to make Barcelona look better
They tested the HE edition, which is 65W. 65W QC is obviouly going to have much lower frequency than highest clocked, 3.0 GHz DC Opteron with power consumption of some ~120W. Each core consumes roughly 1/4 of the power of the highest clocked DC Opteron.
Why they chose to test low power version, with less than maximum expected performance? I have no idea...
Joe
From Intel CC:
The answer to the PCM question is that the new company will get the PCM IP, but the parent companies will retain the IP as well.
That leaves us with the qeustion of who will continue the research to bring this technology to market. It was not answered, but my guess is that the NewCo would be doing that. But Intel retains the option to jump back in throught the ownership of IP.
Joe
Alan,
I think the purchased output from Intel and STM factories get replaced by the 300mm factories when they are brought on-line. 200mm factories are at 65nm and moving to 45nm.
I think that is the plan, but it seems one of the 300mm fabs is for NAND, and the other is empty, and they mentioned there was no plan to equip it in 2007. So the NewCo will need to raise some money near the end of 2007 / beginning of 2008, probably from IPO.
Spansion is a bit behind in nodes (65nm production is only about to start now, but on 300mm) but will likely be at parity node parity at 45nm, with Spansion having 300mm and Mirrorbit trump cards.
I am not very optimistic about this new entity. There was a rumor that the deal would include cash to equip M6, which would have made it more formidable, but the deal does not include that. Intel and STM chose to take the cash instead.
Good for Intel, STM (and Spansion) bad for NewCo.
Joe
Alan,
Here is my take on the fabs of the "newco"
http://www.siliconinvestor.com/readmsg.aspx?msgid=23564265
Did I get it right?
Joe
wbmw,
I didn't mean to nitpick of start an argument. I am just looking for the correct answer.
Joe
wbmw,
re: phase change
I am not sure I agree with your interpretation. It seems to me that it will be in the new entity:
With STMicroelectronics and Intel contributing more than 40 years of combined experience in non-volatile memory technology development, including next-generation phase-change memory, the company will be well positioned to both serve its customers with complete memory solutions and accelerate the move to future non-volatile memory technologies.
http://www.intel.com/pressroom/archive/releases/20070522corp.htm?iid=pr1_releasepri_20070522r
Joe
alan,
I suspect in a similar arrangement to the way AMD owned spansion... however, there has been no announcement the new company will be brought public.
I suspect that the PE firm is doing this in order to prepare this new venture for IPO. Why else would Francisco Partners be involved? Road to profitable IPO (for FP) is much shorter that the road to profitability of the company.
BTW, does this company have a name yet?
Joe
smooth2o,
No, STM sells its NOR assets, as well as much smaller NAND assets into this new venture. The new venture is mainly a NOR company.
Joe
gb,
intel is selling their phase change to this as well.
still makes me wonder about the micron jv however.
intel will have ~50% of two companies competing on nand.
I agree that it is a little weird. The explanation could be that STM is just dumping its NAND stuff into this venture, and it may just be discontinued. Somehow, I doubt that these 2 Intel JVs will be competing against each other.
There were rumors that STM will throw in a fab. I guess it may have happened, since STM has higher stake, even though it started from smaller market share.
Joe
wbmw,
Re: Flash -> HALLELUJAH! eom
It certainly leaves Intel much stronger in its core business.
Joe
spaarky,
I wonder what it means for NAND and phase change (the AP story says Intel is only selling NOR)?
I think Intel is selling phase change into this JV, but NAND stays where it is.
We could only speculate here, but if phase change really was the next big thing, why would Intel be selling more than half the ownership of it? I can only conclude that phase change is far less than a sure thing.
Joe
Alan,
But again, I did hear a rumor that much of that was NAND start-up and R&D costs.
Hard to say without breakdown of revenue and costs. But a couple of data points with YoY data, IMO, show that it is most likely NOR, unless there were some drastic cuts in that division:
NAND+NOR Q1 2007 Q1 2008 NOR%
revenue 469 544 ~0%
loss 283 125 ?
costs 752 669
NOR Q1 2007 Q1 2008 YoY
revenue 351 544 -35%
loss ? 125
costs ? 669
Alan,
I have not checked, but someone posted that flash costs Intel 5 cents per share. I bet most of it is NOR, and NOR is not a strategic product for Intel any more. NAND losses are smaller, but NAND may become more strategic product for Intel if the silicon storage takes off...
Joe
sdwalks,
Hi there, I do remember you from Raging Bull...
The anti-trust cases against all players has removed the ability to protect IP.
Interesting take. Has there been any precedent of striping the company of its IP? I would really doubt that.
Joe
chipguy,
The droids obsession with unit share
"Ok, I didn't follow the thread correctly" would have been a better response, IMO.
Joe
chipguy,
6523/5760 = 1.132 = +13.2%
You are right.
You're right. Intel could have priced its MPUs to generate
ASPs of $200, $500, or $1000+.
In turn its units sales might *fallen* 10%, 50%, or 90%
which would have caused fab under utilization charges to
skyrocket.
AMD couldn't have taken any of those sales, since they were capacity limited throughout most of the quarter. The below par pricing of Intel CPUs had the sole result of taking billions of dollars of profit out of the CPU market.
There was a period of time when AMD did not sell out its capacity - late Q1 and Q2. AMD was signaling to Intel that AMD did not want to go into a price war, and for some time, AMD held prices at more "normal" levels. But Intel did not want to hear any of that, and continued to slash prices.
The x86 MPU market has too much capacity
right now largely because of AMD's megalomaniacal
expansionist plans.
Intel has capacity for >100% of total CPU market, AMD has capacity for 25% of the market. Tell me who has too much capacity.
even setting aside the gargantuan blunder of buying ATI.
On that, I tend to agree. In particular, buying ATI for the amount that AMD paid. It would have been a good eal at 1/2 of the price...
AMD rolled the dice on a bet-the-company strategy and
in Q4 it came up snake eyes. You and mas go cry on
each shoulders over your folie a deux fantasy world that
isn't.
Mas and I? As far as I know, we are both out of the CPU stocks. It is AMD and Intel shareholders who should be crying on each others shoulders.
And FYI, I flirted with Intel too in 2006. I had some LEAPS, but I didn't see the light at the end of the tunnel (and stock fluctuation offered me to make a quick profit) so I dumped the LEAPs.
Joe
Sarmad,
your blatherings would have some credibility with me if they had some credibility with you. The stock is down 60% from its recent high. Yet you can't bring yourself to put a penny into it.
Gee, I would think that avoiding most of the share price decline would gain me some credibility....
Joe
chipguy,
I was talking about unit share. Intel lost unit share of x86 market in Q3, while claiming in the CC that they gained share.
Joe
Samrad,
what is this bullshit you are going on about ad nauseum.
I am talking about unit market share that Intel has been losing steadily. After slowing down the market share losses in Q2 and Q3, unit market share losses accelerated in Q4 again.
AMD had zero share in any profitable pc market. That is Z-E-R-O. Intel had 100% share of pc processor profit. You are on the wrong thread. The morons are at the moderated AMD thread on SI.
LOL. I think you are getting a little too emotional (for your own good). And a bit over your head - again.
Joe
chipguy,
From Q3 to Q4 Intel increased its MPU revenues by 13.2%
compared to 4.2% for AMD.
It was 12.3, IIRC.
An extra $2B represents a further 31% increase for Intel.
You think Intel could have raised MPU ASP by 31% plus
another sizeable chunk to counter the resultant loss in unit
sales?
If you re-read my post, I was comparing YoY results. Q4 results are what they are not only a consequence of actions taken in Q4, but they are consequences of actions take in prior quarters.
What I mean in particular were price cuts of Q1 and Q2. Intel's ASPs did not need to be lower in Q4 2006 while Intel was selling high percentage of CPUs dual core, some of them market leading Conroes and Woodcrest - compared to Q4 of 2005 when Intel was selling single core and donkeys - at higher ASPs.
Joe
tecate,
hmm, I'm not sure I believe this yet. I'll wait.
If you are going to wait to hear it from Otellini or Bryant, it may be a long wait.
Last quarter, for example, Otellini said:
"Tim. It was a record, we believe we gained some overall share this quarter."
but in fact, Intel lost share in that quarter, again. This time, Otellini was more cautious...
Joe
snowrider,
Intel's CPU revenue was up 12.3% while Intel claimed ASPs to be up. That means that Intel's units went up less than 12.3%, while AMD units were up 19%.
Joe
Alan,
So if I understand this correctly...
Intel gained revenue share but lost unit share,
while AMD gained unit share at the expense of revenue share.
Now who is engaging in a price war here?
AMD is reciprocating what Intel unilaterally started.
For comparison, from Q4 2005 to Q4 2006, Intel has gone from weak product portfolio to strong, PC market is surging, but Intel revenues are down 5%, Operiating Income is down 55%, Net Income is down 39%, EPS are down 35%.
Intel chose much lower level of profitability for the CPU industry. It didn't have to be that way, and still, it didn't stop AMD from selling out its production for the year anyway.
Intel left probably $2B on the table in Q4 alone to fight this price war. That, in order to deny its competitor $200M...
Joe
tecate,
AMD units went up by 19% in Q4 (vs. Q3). Intel units went up by less than 12.3%. (Info from quarterly releases and conference calls).
Joe
sgolds,
Any guesses on how competitive Barcelona will be?
That's a million dollar question... Especially the clock speeds it will be released at, and when.
Joe
mmoy,
If you're really dreaming of a 6000+, buy a 5600+ and overclock
I think Keith is talking more in terms of AMD's ASPs. For AMD to have at least stable ASPs, performance needs to steadily improve. AMD released 4800 in Q2 of 2005 (IIRC). Since then, AMD released 5200 only in Q4 of 2006, and barely released 5600 in Q1 2007. 6000 looks like a Q2 product now. So it is 3 speed grades in 2 years. That's clearly not acceptable, especially while your competitor is not shooting himself in the foot...
Joe
chipgeek,
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.
I think that's how generally a margin account works. IRA is considered to be a cash account.
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.
I was under the impression that the brokers requires 100% of assigned cash value in an IRA, but I am not positive on that. It may be, as you say, 50%. But in your example, having $10,000 in the account and 50% cash requirement, you could write puts with value (when assigned) of $20,000 (if you wanted no safety margin).
Joe
Elmer,
Long puts and calls are fine in IRA. All you need is appropriate level of approval for option trading.
As I said, everything except naked calls.
Joe
Willco,
I think the rule on buying puts is that you must have the money in the IRA to make good if the put gets exercised. This rule is a relaxation of the old rules which only allowed covered call writing. Will
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.
Joe
Keith,
From the article:
The company hopes Barcelona will allow AMD to finally break into the low-end server segment, including one-chip and two-chip computers and tower PCs as well as rack mounted servers.
"We're not yet in the SMB segment; it represents for us a great volume opportunity since we have so little market share now," said Kevin Knox, vice president of AMD's commercial business, on Wednesday.
Overall, I believe quad core single socket (with plenty of memory slots) is all that's needed for the vast majority of the server buyers. It is not, however, what Intel and AMD want buyers to "discover" since 2 and 4 socket servers are much more profitable.
BTW, single socket quad core servers can, curiously, be faster in lightly loaded environments than > 1S, since there are no delay for memory coherency checks.
It is only when you have extremely high load (loading all the cores) when gains from processing power of extra cores outweighs memory access penalty.
You may have noticed that when AMD released FX, on many (most) of Anand's tests, single FX outperformed dual FX...
Joe
Elmer,
Etrade lets me do all option transactions in IRA except naked calls.
Joe
wbmw,
Your analysis gives a range, and I was going for more of an actual figure. Either way, with your other points noted, I don't think our estimates are inconsistent.
Yes. It is a range because the only information that we have from Intel is that their ASPs increased. We don't know by how much, but it must be at least 1%. Otherwise, I doubt Intel would broadcast an ASP increase that is a fraction of a percentage. That would be under "flat" category.
Hector has all of Fab36 and most of Fab30, and yet he is still paying money to Chartered for more wafers. That's at least one expense he could cut.
I agree there (on Chartered). But apparently, AMD still needed Chartered in Q4. I would expect the Chartered orders to fall until possibly Q3/Q4, when Fab 30 goes to conversion and AMD will have large die K8L to ramp.
Not really. As part of their Analyst Day, they had yet another large charge to post for the ATI acquisition. I believe it was around $430M. Add in additional losses, and you're back to a half billion in red ink.
Those may be write-offs for paying ATI higher price than the book value, not operational losses. While there will likely be operational losses in ATI division at least until R600 ramps, and small operational profit of CPU division may swing to operational loss is slow quarters (Q1 to Q2). But half billions of red ink in operational losses would only happen if Intel becomes extremely aggressive in low end of the market. And that is outside of AMD control.
Joe
wbmw,
AMD claimed a 19% sequential increase in processor unit shipments, and 26% YoY. What did we assume for last year's production capacity?
To arrive at how much unit share AMD gained, you don't need to know the unit shipments. You start with last Q unit share and add the unit growth figures (which we got from the CCs). here is my summary:
http://www.siliconinvestor.com/readmsg.aspx?msgid=23215731
As far as unit, Mercury research figures are the only ones that can be used, since they track microprocessor units in all segments (server, mobile, desktop). AMD had 23.3% in Q3 and now it should be 24.5% or more. That may be an all time high.
BTW, I think you are a little low with 239M units for the whole market. That may be because Gartner figure track PC shipments and not processor shipments. I guess it must exclude server shipments (with > 1 CPU per server) and people who build their own PCs.
But if you want to put me down for a prediction, Joe, I will say that I believe 2006 to be the apex of AMD's market share percentage.
That's quite possible, I can't predict the future. But if it is the case, it will not be because AMD lacked the capacity to gained market share. AMD capacity will soon reach (and exceed) 30%.
Given that ATI is already weighing down AMD to quarterly losses, just how much do you think AMD will lower prices to gain share?
I don't know, but we do know that AMD was capacity constrained in 2 out of 3 months of Q4. So AMD could have sold even more units, theoretically.
Then again, maybe they will. Hector Ruiz sounded pretty desperate and fanatical during the earnings conference, and it sounded like he would be willing to swim in all kinds of red ink before he slows down his production ramp.
You are assuming that if Hector slowed down the production ramp, AMD would be more profitable. I disagree. Letting expensive equipment sit idle is not a recipe for profit.
So Hector can at this point (considering competitive situation of both companies respective products) do only one thing, which is to go for more unit market share to try to keep its factories full. In my book, when you have only one option and you take it, you are not fanatical.
So maybe they will gain another point or 2, and post more half billion dollar losses.
Now you are getting silly.
Joe
wbmw,
The most likely product we'll see from Fusion is a combined CPU and GPU that has had to derate the performance of both in order to fit in the same power envelopes that current mainstream processors currently exist. For example, a 65W TDP Fusion processor might include 45W of CPU and 20W of GPU. A 95W TDP might include 65W of CPU and 30W of GPU. Essentially, they both have to share the resources available to them.
That's the ticket - mainstream to low end market. High end will continue to be discrete.
BTW, what I meant by net basis is the system power. If you eliminate some of the external buses, you save power. In addition to saving power, you save on mobo costs.
Joe
tecate,
In what category, pointer please.
AMD gained unit market share of total market (servers + mobile + desktop).
Joe
Joey,
Paul O. was quoted as saying (right after earnings) that their strategy is to regain any share they lost to AMD. I thought that was a pretty strong statement.
Intel appears to have lost another 1.5% to 2% of market share in the same quarter he said that...
Joe
gb,
i thought henri's response to the question of "what is your sales pitch for 1p servers?" to be embarrasing. i think it was "better product."
I seem to recall him saying better value. I guess we will see it in the transcript soon.
Joe