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.
wbmw,
They seem a bit off to me, too, but there may be a grain of truth to it.
Whatever sustains your faith:
http://www.theonion.com/content/node/39512
On the other hand, one set of figures I do not trust on the subject are yours.
Do so at your own risk...
Joe
chipguy,
Xeon MP (1M L3) - $22
Xeon (2M L2) - $24
Opteron (SC) - $79
Opteron (DC) - $85
LOL, good one.
Microprocessor Report is turning into a publication that is a lot of fun. But I think The Onion still beats them. Both on performance and price performance:
http://www.theonion.com/content/index
Joe
gollem,
I am still a little puzzled about the differences (in chipset support) between Socket AM2 and all of the previous sockets.
Prior to the Rev F, one chipset could support anything from the cheapest Socket 754 to the most expensive Opteron multiway, with the only minor variation being 800 and 1000 HT.
There was a rumor somewhere (PC Watch?) that the HT speed would remain at 1000 MHz for desktops, which makes me even more puzzled about what is there to support (by chipset providers)...
Joe
fpg,
I was kind of assuming that the crossbar is somewhere in the region that looks kind of blank, in the upper-middle portion. Just because of proximity.
The 2 DDR2 interfaces at the bottom are not attached to Crossbar. They attach to the memory controller in the upper half, just right of center. So we have the 3 HT ports, MC, and SRQ attached to the cross bar, all of which are in the upper portion of the die.
BTW, it is interesting how the cores shrink, and the I/O does not. AMD die has a bigger portion of the die dedicated to I/O then Intel CPUs. So on one hand, AMD is using up more capacity on it. On the other hand, the rest of chipset infrastructure has much lower complexity (and so do the motherboards) making it easier to avoid chipset / mobo caused shortages.
Joe
Hmm...:
We cannot assure that any acquisition or other change in control transaction will occur, or that if such a transaction does occur that it would result in stockholders realizing value equal to or greater than the current trading price of our common stock.
http://www.channelregister.co.uk/2006/02/08/sgi_warns/
Joe
cruzbay,
I may by a few shares of Spansion (not calls).
Spansion profitability and share price rise would certainly help AMD stock...
Joe
Paul,
It is interesting that only the tiny area on the top is labeled Hypertransport pads. COmpare it to this one where it wraps around:
Joe
wbmw,
Chipdesigner. From hence forth, I recommend a sell on AMD and buy on INTC for at least one year.
It seems that instead, you are trying to sweep under the rug the atrocious performance of your last pick, zero out the losses and start fresh. Sorry...
Joe
chipguy,
January, IBM's chief architect Frank Soltis, who heads up the company's core iSeries and pSeries divisions, hinted that IBM might be pursuing the Sun business aggressively.
"Sun will run on one of these platforms — AMD, Itanium or Power," Soltis told journalists. "Really it is going to be Power or Itanium. It's just too expensive to develop these things now."
Seems like someone in serious denial about x86. He doesn't even mention Xeon...
Joe
chipguy,
To your credit, you didn't address my point (testing for "GenuineIntel")
Joe
wbmw,
Wrong. Intel tests for 'Genuine Intel' to prevent mislabeled 3rd party CPUs from vendors like Via or Rise from breaking the code.
Nobody with a clue and an ounce of integrity can defend testing for "Genuine Intel".
Joe
chipguy,
I would be shocked if AMD didn't move to a design with
two independent memory controllers, one for each DIMM
channel, for better memory level parallelism.
What's the advantage of 2 independent parallel memory controllers vs. 1, more complex one?
Joe
X2 today, anyone gonna sell their amd and buy intel?
waiting for X4... <g>
Joe
I think Ediot had 2 choices:
a) continue following his "insights" about AMD production and capacity, continue making ass of himself
b) cheat, and read this BBS and SI
Ediot decided to go with b).
Joe
re: Rackable Systems seems to be doing well
That was noticable in Sun's slide on the Opteron market in
its last earnings release presentation.
Sun's slides only had Q3 data on market share etc. The Rackable financial results are based on Q4, meaning newer figures. From the Sun and Rackable results, it seems that Sun is not (yet) scoring any breakthroughs (as far as gaining market share) in the Opteron part of the market (mainly due to product delays and poor execution).
Joe
wbmw,
You're a genius, Joe. According to you, anyone who bought AMD before it was $24 (i.e. in the $5-20 range) and sold it at $24 would have subsequently "lost money". It sounds like you have a pretty strong grasp of the way trading stocks work, eh?
The time of purchase is irrelevant. Whether it was $5, $21, or $40 to cover a short. The time of sale is, and your recommendation had just about the worst timing.
And, after several times of getting it explained to you, you still choose to misportray my recommendation, which was to "look for a place to sell"
LOL. Retroactively, it turns out, that it was not the best time to start looking...
Come on, Joe. You were one of the chumps holding the bag when Jerry Sanders did his pump and dump, and AMD took the long, hard road from 2000 highs to fears of bankruptcy. I'm quite sure you've already rivaled by far the worst investment decision I've ever made.
I see, I made a mistake by not selling for a loss near the lows (like you, IIRC), but held, bought more to ride back to the highs.
Thanks for reminding me of another one of your great investment decisions.
Joe
Gordon,
You still have some time on those...
Joe
AG,
Anyone who has stayed the course with Intel for the last 9 years has nothing to show for it. Anyone who has listened to you for the last few years has even less.
I think there is an accelerating trend to how following wbmw would lose you money. We have his recommendation to sell AMD in September when AMD was at $24.
If there were to be contest on how quickly you could lose money, I think it would be tough to rival wbmw.
BTW, his "staying the course" on Intel is no picnic either...
Joe
chipguy,
What is the colour of the elephant you think you see?
Green. You? Red?
chipguy,
The French nuke folks got their 50 TFLOP/s system ordered,
delivered, and up and running to spec in about a year using
existing commercial server components. The French obviously
learned a lesson from NASA's experience with Columbia.
In contrast the Sandia folks paid ~$100m Cray to develop
a proprietary 40 TFLOP/s system based on Opteron.
x86 is proprietary now, and Itanium is "existing commercial server component"?
You must be missing something. Itanium is a leap of faith, therefore needs to be heavily discounted. Did Intel chip in by giving the Montecito samples for free?
Joe
Ahh, the legendary French military is going to be "armed" by Montecito.
http://www.albinoblacksheep.com/text/victories.html
Joe
chipguy,
Pardon me but I must have missed AMD showing off working
45 nm devices. Do you have a link?
I don't have any of those, but I have a link to some 90nm devices showed off nearly 2 years ago:
Intel Demos Dual-Core Itanium 2 Processors.
Next Year's IA64 Chips on Track
Category: CPU
by Anton Shilov
[ 06/19/2004 | 06:13 PM ]
http://www.xbitlabs.com/news/cpu/display/20040619180753.html
Looking back, those were probably not "working devices". Shouldn't Intel first get their 90nm devices to work before showing off 45nm ones?
Joe
dh,
I will take a look at the site when I have some free time on my hands. I have been kind of busy lately...
Joe
mas,
my take is that even though Intel may gain a small lead in performance, AMD answer will primarily be in lower power consumption, allowing higher clock speeds.
If the info (that recently appeared on MSI roadmap) on 25/35W Turion is accurate, F stepping is going to make good progress, still on 90nm.
65nm should bring further gains. I am not sure if we can expect huge gains, like we saw from Newcastle to Winchester, but there should be some gains.
Joe
2 spots are open for assistant moderators, if mas chooses to nominate someone...
Hey guys, stay on topic. You are making me work on this moderating stuff. The other thread is more free-wheeling than this one.
Joe
bobs,
Here is my stab at trying to figure out costs of manufacturing CPUs between AMD and Intel:
http://www.investorshub.com/boards/read_msg.asp?message_id=9395966
My conclusion is that they are at parity in high 30s.
Joe
chipguy,
As more functionality migrates from chipsets into future IPF MPUs that fraction will probably increase somewhat.
Do you mean CSI, or something else?
Joe
Rink,
Galaxy was announced, but initial supply was very low, not enough to meet the demand (that's what I seem to remember).
There could be a good jump in Q1, and Sun will start to realize some return on the Opteron investment, but hopefully, this whole thing has been a very long (and probably expensive) learning lesson, that on x86 end, the development cycles are a lot shorter, and you have to be right there, from beginning, with your product. Otherwise, someone else will take the sales.
Sun is late, and HP is taking the bulk of Tier-1 Opteron sales.
Hopefully, Sun will have learned the lesson, and they will be right there when Socket 1207 Opterons launch.
Joe
smooth,
Intel has shown Intel data on successive generations of yield graphs that say the rate of yield improvement with each generation increases through to 65nm.
Those graphs, if you noticed, don't label the Y axis, with anything concrete, but instead, with something called Mature yield, which is open to definition (and redefinition) for each successive curve, because the author of the graph (Intel or AMD) can have a different standard of what constitutes a mature yield at each node.
Another thing is the whole idea of yield. The general definition is #of good die per wafer. But a "good die" that can perform the work without errors can:
a) not be able to achieve the desired clock speed
b) leaks too much power.
Think of the challenge Prescott generation faces: Power consumption is too high, and bin splits are below target.
So Intel can have a lot of "good die" that unfortunately fall into the bad quadrant of bin splits and power consumption, and just cannot be sold, ot they can be sold as Celeron, but Intel already has too many of those.
I am not saying that this is the case, but it is just a possible explanation of good yield of working parts, but not good enough yield of the desirable parts.
Now, with DC Smithfield, things are so much tougher. The bar of how low the power consumption needs to be for the 2 die to add up to less than the limit is much tougher.
My WAG is that yield of saleable parts of Smithfield is atrocious, limiting most to the lowest bin or below (unseleable). Then, on the other hand, the performance of Smithfield is very bad compared to competition, not justifying anything other than a deeply discounted price.
I think this could potentially turn around somewhat with 65nm Pressler, with improvements in average power consumption. Further physically detaching the die will improve the yield, and thirdly, rumored introduction of even lower bin than the current bottom one will turn the former dumpster material into a seleable part.
Given that Intel has maintained their lead throughout these generations and that the 45nm lead seems ensured as demonstrated today, what, except for a good quarter from AMD has *changed* that AMD is going to hold more than a historical 20% of market share in light of Intel producing an entirely new core in the beginning of H2 that threatens to take the performance crown, take back any market share advantage AMD has and leaving the only advantage that AMD has is an "elegant" architecture that the 'droids have fallen in love with?
That is certainly a possibility, but the big hurdle to overcome, for Intel will be Q2. AMD will be capacity unconstrained. Q2 is the slowest quarter, and AMD hs a ton of capacity coming online. Anybody with any dissatisfaction with Intel will be free to jump the ship (Intel will be too afraid to retaliate because of the law suits). That can translate into a "bad" quarter for Intel (earning some fraction of a billion less) but most importantly, bad PC from lost market share.
After that, with NGA, things can turn around for Intel. Who knows.
Joe
smooth2o,
The next news is Q1. What's your best guess? What do you think the Intel sales people are doing right now.... vacationing in HI?
They may just as well be. Apparently, the entire Q1 production of AMD is already spoken for. So the only upside for Intel is to somehow enlarge the market, rather than take something from AMD.
Joe
re: toying with ides of buying Intel
I think Intel is becoming a "value" play. The problem is that Intel has to cross a very turbulent ocean over next 2 to 3 quarters, with potential of very bad PR. So you have to wonder if Intel stock out there is in strong hands, that will not sell at face of adversity, or will they sell?
Joe
Keith,
Sun shipped about 8,000 Galaxy servers
Sun Total: 20,000 servers
Opteron market total: 128,000 servers
It means that sun has less than 1/6 of Opteron servers.
With low Galaxy sales (probably because of supply issues), and talk of a backlog, I think there is a lot of additional Galaxy sales this Q.
Also, I hope that of the 12,000 non-Galaxy servers, a good chunk were 4 way servers.
Joe
wbmw,
Here's the list I've seen:
Looks like you are using discontinued parts on your list to make a point. The $21 part has a grand total of 3 motherboards listed on newegg, one of them out of stock.
The low end parts are in the range of $21-28, the mainstream are $29-40, the high end are $50. Given a volume mix of 30/60/10, that's an ASP just north of $32. At this point, you have to figure in an extra 10-15% standard discount due to volume or bundle purchases, which brings the ASP south of $30.
I guess the problem is that the low end, which should be skewing things to the low priced chipsets is much more likely supplied by 3rd party chipset vendors, so Intel chipset ASPs are skewed to the higher end.
Conveniently, there is an article out today proving both of my points:
Intel resuming 865-series chipset production may negatively affect third-party suppliers
http://digitimes.com/mobos/a20060124A2006.html
Meaning that 865 line was discontinued, and that resumption of the low end chipset will affect thrid party vendors.
Curiously, the link shows that SiS and Via are able to make gross margins of 34% and 22% on the el-cheepo stuff, below Intel's lowest end, but Intel can only make 15 to 20% on the mainstream and premium stuff. Sis and Via are using foundries (at full prices including profit for foundry), Intel is using depreciated fabs, keeping the profit.
May I suggest that when this deep in a hole, that you stop digging?
Another suggestion: step back to review what you are arguing. Your argument that Intel CPU cost is (substantially) lower than AMD's depends on you proving that Intel cost of producing a chipset is roughly the same as cost of producing a CPU. This argument obviously does not pass the smell test.
Well sure they can. They have the ASPs of CPUs in front of them, and a good, educated guess as to the ASPs of chipsets (MCH+ICH combos) that I provided above. They can also judge for themselves if chipsets have a high GM of 35% like you suggest, or if they are more familiar with this business they can use the value of 15-20%, which I believe is more standard, and does not constitute "losing money" or "predatory pricing", like you shamelessly suggested in a previous post.
Ok fine. The same article proves my third point: Sis makes money with 34% gross margin, Via loses money with 22% margin. Therefore, would Intel make money in:
- Your scenario of 15% to 20% gross margin? Answer: No
- My scenario of 35% gross margin? Answer: Yes
http://digitimes.com/mobos/a20060124A2006.html
Joe
Tenchusatsu,
Same thing applies to AMD, only AMD's losses would cut deeper.
It might be fun to subtract $50 per CPU that Intel sells and "prove" that AMD can be more profitable if you equalize the ASPs.
I used the ASPs, their differences, and changes only as an illustration of where the big bucks are, and that silicon cost of the CPU is just pennies compared to the ASPs dollars.
But as far hypothetical scenario of equalizing the ASPs, I would ba a lot more interested in scenario of AMD ASPs going up, rather than Intel's going down. Extra $50 per CPU would earn AMD extra 650M, which would be extra $1.50 per share per quarter. Something like extra $120 per share with 20 multiple.
Joe
wbmw,
No, I didn't. For the "mentally challenged" individuals who think they can calculate chipset costs by applying the 15-20% margin figure to Intel's $50 high end enthusiast chipset that sells in micro-volumes, consider doing that instead to Intel's $25-30 mainstream chipsets that sell in >10x higher volumes.
No, I included value to mainstream 915P and 915G parts as well. The prices were $37 and $41.
you should be calculating the 15-20% margin to a <$30 average chipset price. That gives a chipset cost of ~$25 or less for an MCH + ICH combination, far below Intel's CPU costs.
First, I calculated $30 and slightly higher on the costs of mainstream parts (that is the cost based on your formula), and you also claimed CPU cost is also $30. Your claim that the cost of a chipset and a CPU are roughly the same is, frankly, laughable.
My claim is that Intel CPU costs are roughly $38 and chipsets are no higher than low $20s. Others can judge for themselves which set of numbers is believable.
Should I take this to mean "I don't have a point, I was just trolling" and I will not be hearing from you any more about how AMD CPUs are more cost effective?
Cost effective? I am not sure what you mean. To a buyer? Certainly across server and desktop segments and in low end of the notebook market. You could convince me of the cost effectiveness of DC Yonah for some specific needs.
Joe
Tenchu,
where do you get these numbers? Are your ASP figures only counting CPUs for both Intel and AMD?
The best estimate I have seen to date is here:
http://www.investorshub.com/boards/read_msg.asp?message_id=9311982
BTW, TWY came within a decimal point to what mercury research estimates the market shares.
Is this a real apples-to-apples comparison, considering that Intel is much more than CPUs?
Well, nothing else Intel sales is worth much. Networking loses a bundle, flash loses a little. Chipsets make some profits and CPUs make a bundle. If you erase CPU profits, Intel is losing money.
And why should cost reduction only be measured in pennies per unit for Intel? Has AMD figured out the magic formula for saving dollars per unit?
AMD is losing about $50 per CPU, compared to what the market price should be on merit (or Intel is earning extra $50). Pennies or a dollar on the production cost can help, but AMD is after the $50 difference per unit.
And if die size is so crippling for Intel, then should SOI be considered "free" for AMD?
I don't think there is anything that is crippling Intel. My point is that theoretically, Intel should have lower costs, but there are things that prevent Intel from having lower costs. The end result is approximate cost parity.
IMO, the things that are preventing Intel from achieving the theoretical lower costs (than AMD) are 1 or a combination of the following 3:
a) less than full utilization
b) low yields
c) slightly larger average die
While AMD has reached and is selling maximum output possible out of its fab, Intel is selling somewhere between 1/2 and 2/3 of its maximum CPU output should be. If Intel were at full utilization with yields comparable to AMD, Intel should have somewhat lower costs due things like 300mm wafers, bulk silicon. It is just not working out that way based on the numbers.
But anyway, we are back to talking about pennies. There is a purpose behind this, the fact that Intel spokesmen and the followers torturously talk about these penny cost reductions: It is a good PR. Compare the following statements by CFO:
a) we made $3 billion dollars, and cost savings due to our technological breakthrougs and capital investemnts contributed to it
b) we made $3 billion dollars because we were able to (over)charge our customers by $50 dollars more than our competitor.
Joe
windsock,
Did you decided to troll over here after you got banned from the AMD Strictly Moderated thread for out of bounds conduct ?
Surely I must have banned myself.
Keep up your record of being wrong 100% of the time. It is something to aspire to.
Joe
Tenchu,
With all due respect, I think your challenge is unfounded and an attempt to troll Beamer. Which is fine if that's exactly what your intention is, but the reality is that Intel has huge economies of scale on its side, which gives Intel a ton of leeway when it comes to competitive threats.
Intel's leavay is in ASP $50 to $55 higher than AMD's, at which point 2/3 of the profit disappears. With $68 reduction in ASP (maintaining the same units) Intel would be losing money.
Now when you hear that there will be some great cost reduction of say 25% (generally offset by increase in die size eating the reduction), they are talking about some 5% on $25 we are talking about a dollar and a quarter.
Just to put this number in a perspective, compare it to only a minor movement on ASP, which TWY estimated to have dropped by $5 last Q. So magnitude of a very minor drop in ASP that barely gets mentioned during the CC can be as 4x of some "cost reduction" that Intel trumpets.
The cost reduction is pennies, processor ASPs are dollars.
So the leaway you are speaking of are pennies, and you are ignoring the dollars.
Joe
Keith,
It is a little unclear to me: Is AMD promising quad core on existing Socket 940, or only on upcoming Socket F (or 1207)?
Also, I wonder about the power envelope.
Joe