If you look a little deeper, you'd see the equity increased because of the $3.2Billion "goodwill" from the ATI purchase. It means ATI worths $x and AMD paid x+$3.2B because they think ATI worth x+3.2B to them. If AMD cannot realize the $3.2B they overpaid for ATI asset, those goodwill don't mean anything and you can count on the shareholder's equity getting reduced.
AMD borrowed bunched of money to buy ATI. It is a risky move for them, especially at a time when their main product is at the end of the life cycle and they don't have a new product to replace it. They borrowed bunch of money, increase their interest payment and decreased their flexibility to borrow more money while their cash flow is squeezed because they can only compete with price cuts with the current product. Their new product is months away and they don't know what the yield is gonna be, how competitive it is gonna be, and if Intel will react with equally competitive product.
I like AMD product, don't get me wrong. But AMD is just in a bad financial position right now. They burned bunch of cash, borrowed bunch of money, and they don't have sure cash flow. Intel on the other hand have better product for now, better manufacturing capability, and deeper pocket to squeeze AMD on price. I do hope AMD can come up with Barcelona quickly, get a good yield, and make it a mass market product, not just a niche server product. They gotta be able to get some cash making product quickly, and not just beating their chest and saying how fast their processor or graphic card is. But after P4 mess, Intel seems to have their act together and I expect them, together with nVidia, to continue coming up with good products and squeeze AMD on cash flow.
- Mar 13, 2006
- 10,140
- 819
- 126
Oops, try again.
CRN: When will Barcelona ship?
RICHARD: It's slated for introduction at the end of the second quarter and will be in the market in the third quarter.
CRN: You mentioned the importance of Barcelona. What will be its impact on AMD this year?
RUIZ: This is an incredibly important product transition. We don't expect the ramp [this year] to be dramatic because it's a new core, new micro architecture and platform.
No Barcelona until end of summer, and then not many.
I wonder just how many Barcelonas they plan on shipping. Doesn't do much good to have a hot new architecture when you aren't shipping enough to overcome your operating losses. You'd think if they were really confident in the product and their process, they'd ramp immediately considering how anticipated this chip has been.
As an enthusiast, I hope that the limited shipments don't go directly to Dell (or other OEMs), but they most likely will. Here's to AMD beating production expectations (I hope).
A fair call and a good point (thanks for catching that!). But you can look at it as they spent $2.4 Billion on the direct income generation of ATI, and $3.2 Billion on the R&D generating results of the acquisition. Unfortunately, we won't really know whether they paid too much until we see the fruits of that investment (2008/9).
I agree that it's not all rosy for AMD...
That said, I don't think it's as dark as you make it seem either.
1. Yields - They have only been in volume production on Barcelona for a month or 2 now, so they probably don't know what yields will be (first wafer outs should be near the end of April or early May). But their APM 3.0 system gives them a good indication of those yields...
2. Capacity - As Hector said in the CC, they will have the capacity to produce >100 Million units this year, which is about double their current marketshare...I don't see that as an issue currently.
3. Competitiveness - This is an area where they are almost assuredly on very solid ground. They knew how competitive they were going to be as soon as C2D was released last year (which is probably why they delayed the acquisition until after that release).
a) Opteron (K8) is already more competitive in performance at >4 cores, and this should become significantly greater with K10.
b) Penryn doesn't add much to the server sector except a larger cache and better power/performance and they can pretty much guage how effective they will be vs Penryn already (they have their own HK+MG models to tell them what level that will be).
c) Unless Intel is able to bring Nehalem well forward in it's release, most of the next 2 years is mapped out.
As to Intel having their act together, that was quite apparent to me the day that they cancelled Whitefield and bit the bullet...
Unlike many of the people here, I think that Penryn is much more hype than actual substance (Intel marketing sure has done a great job though!), but Nehalem on the other hand could be as significant an improvement as C2D was. The problem is that we don't have even a glimmer of an idea as to how well the project is going!
As an aside, I am constantly amazed at some of the posters here...
They advise caution on the desktop version of K10 (fair enough, wait for the benches), but at the same time they spout absolute confidence in the timely release and success of Nehalem...Nehalem is the single largest change in architecture for both the processer and the platform that I can remember Intel ever making (aside from Itanium)!
It's an even greater change than going from P3 to C2D, and many people think it's a given that it will appear on-time and working perfectly...
But I digress...
At the end of the day, I really like AMD's chances with K10...though your cautionary points are not falling on deaf ears!
Ummm...Q3 is the day after the end of Q2. If it ships at the end of Q2, it should take at least a month to make it on to the shelves.
And, Barcelona is a quad core server processer...of COURSE there won't be many, there never are (either for Intel or AMD).
Also, remember that the Barcelona is a transitional chip...the K10 with HT 3.0 is due later in the year.
At a guess, I'd say only about 10,000 in the first month...
But you have to keep in mind what a Barcelona actually is. It's a quad core server chip (not for enthusiasts), and it will take several months for it to become "qualified" on server platforms. This is not unusual, in fact it's the norm...Woodcrest went through the same thing, as did Opteron (though AMD had presold several hundred thousand of the original Opterons before it's launch).
The chips you need to keep an eye on (for enthusiasts) will be the "Stars" processers in Q3 sometime...
I've been reading most of your post Viditor and I'm almost certain that you tend to side more or bias to AMD than with Intel although making PD of owning Intel stocks doesn't change with what our views here.
Anyway, I think eventhough Penryn doesn't have any architectural improvement vs C2D which will translate to dramatic performance improvement, I guess having this at 45nm will still make Intel competitive since it can produce more die per wafer. They can lower down cost ithout hurting more on their bottomline. This we still have to wait how far Intel can push its frequency up. Lets not forget Intel have a 1600MHz FSB chipset in their roadmap which will really make it competitive.
Anyway, I think eventhough Penryn doesn't have any architectural improvement vs C2D which will translate to dramatic performance improvement, I guess having this at 45nm will still make Intel competitive since it can produce more die per wafer. They can lower down cost ithout hurting more on their bottomline. This we still have to wait how far Intel can push its frequency up. Lets not forget Intel have a 1600MHz FSB chipset in their roadmap which will really make it competitive.
I won't really dispute that...I do have a bias towards AMD, but I try to keep it within rational boundaries.
I wouldn't count on them being able to produce more dies per wafer...remember that the extra cache takes up a huge amount of the added room, and there are many other additional transistors on Penryn as well. Penryn has ~50% more transistors than a C2D with 4MB cache.
I do agree with you that the faster FSB will be quite effective, but I think we need to wait and see a demo of it first...
The issue is neither the Core 2 Duo or K8 let alone K10 which has even largers on die caches are likely to be memory bandwidth starved, so I don't think going to 1.6GHZ FSB will do that much, but we will see.
45nm Core 2 Duo's will allow Intel to up the clock frequencies while maintain the same thermal envelope for Dual Core and lower it for Quad Core.
Penryn derivatives will have some improvements, like increased cache size and SSE4, so we will have to see how competitive they are to AMD's offerings.
On a final note, your first paragraph is quite wise and I agree with the assessment.
That is pretty much guaranteed, you can already easily derive the die size of the 45nm Core 2 Duo's today, considering they have 410 Million Transistors, and a 41% increase in cache doesn't have that large an impact on die size.
Look at Prescott to Prescott 2M, transistor count increased 35% while the die size increased by 20% and that was an all cache addition. Remember cache on average is the higher density transistors on a given process.
A perfect 45nm shrink yields a die that is 48% as large, to incorporate real world shrinks simply increase this figure to 63% to be on the safe side.
Since 80% of the increase is coming from cache, using Anandtech's figures.
http://anandtech.com/cpuchipsets/showdoc.aspx?i=2915&p=3
A theoretical optical shrink of Conroe without cache additions is 90mm2 using the 63% baseline, so if Penryn has 41% more transistors and given that 80% of that increase is cache, +28% die size is not unreasonable. So your looking at a die size similar to what Allendale is now of about 115mm2.
But what is for absolute certain is that Penryn is not as large as Conroe, as even going by a conservative 63% estimate, this allows 58% more transistors in the same area, and Penryn is only 40% larger in transistor count as compared to Conroe.
Nehalem is too far off at this point to really discuss in detail at this point in time, what is up and coming which is exciting is 45nm based Core 2's. So Intel is choosing to focus on that at this point in time, expect Intel's marketing team to talk about Nehalem once the 45nm Core 2's are shipping for revenue.
Penryn obviously won't be as significant an improvement as Pentium D to Core 2 Duo was, but as far as semi-optical shrinks go, it should be a nice boost.
I agree that for the single sockets, the net gain will be far less than it will be for the 2P+ chips. But a faster FSB is absolutely imperative for Intel's server line (at least those above the low end) as it should help their scaling quite a bit.
But how much are you expecting them to increase power/performance (shorter way of saying the same thing)? I think there has been a great deal of "irrational exuberence" on the part of many Intel enthusiasts as to the degree of this...JMHO.
We have gone through this before, but the cache gets them very little (probably the least of the additions), and AMD is adding the extensions as well...
Cut me some slack here Cold...at least I am straight up about it!
BTW, I have yet to see you post a positive note about AMD for as long as I can recall...don't throw stones in glass houses, mate.
Originally posted by: Viditor
Cut me some slack here Cold...at least I am straight up about it!
BTW, I have yet to see you post a positive note about AMD for as long as I can recall...don't throw stones in glass houses, mate.
There is no glass houses to break, I do say positive things about AMD, when they are deserving of it of course. Don't you worry, I can see positive points about AMD, such as the better gaming performance of the Athlon 64 vs the Pentium 4 line.
Originally posted by: coldpower27
Originally posted by: Viditor
Cut me some slack here Cold...at least I am straight up about it!
BTW, I have yet to see you post a positive note about AMD for as long as I can recall...don't throw stones in glass houses, mate.
There is no glass houses to break, I do say positive things about AMD, when they are deserving of it of course. Don't you worry, I can see positive points about AMD, such as the better gaming performance of the Athlon 64 vs the Pentium 4 line.
That's sort of like me saying I can see some true benefits from the P4...it keeps the house warm. :roll:
My point is that you seem to be just as biased as I am, but towards Intel (and yes, I too say positive things about Intel...when they deserve it).
Originally posted by: Viditor
Originally posted by: coldpower27
Originally posted by: Viditor
Cut me some slack here Cold...at least I am straight up about it!
BTW, I have yet to see you post a positive note about AMD for as long as I can recall...don't throw stones in glass houses, mate.
There is no glass houses to break, I do say positive things about AMD, when they are deserving of it of course. Don't you worry, I can see positive points about AMD, such as the better gaming performance of the Athlon 64 vs the Pentium 4 line.
That's sort of like me saying I can see some true benefits from the P4...it keeps the house warm. :roll:
My point is that you seem to be just as biased as I am, but towards Intel (and yes, I too say positive things about Intel...when they deserve it).
You can believe whatever you wish, good try at poking humor at the Pentium 4 though,
considering the gaming performance of the Athlon 64 is a good point for the people of this forum, I don't see how it compares to what you said, but at any rate I don't think I will change your stance regardless of what I say, so I won't say anything. From my perspective you are seeing something which is not there.TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 22K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
Facebook
Twitter