The Official IGP Rumours & Speculation Thread

I think AMD is currently doing its best in accelerating Fusion and increasing its performance while decreasing power consumption and costs, AMD knows that 'IF' Nehalem really beats K10, their future will depend on Fusion.

Considering how AMD survived after Intel released Pentium and AMD couldn't do anymore 1:1 photocopies of Intel chips, I don't think their fate could be sealed with 1 or 2 architectures missing the boat a bit
 
Times aren't the same as back then. AMD used to have other business units (remember Spansion?).

It's sold off pretty much anything it can sell. It can't sell ATI right now because the debt is just too much. If things don't go well enough in the future, I'd expect AMD to consider selling whatever it can from there as well.

The cost of the lithography race is way beyond what it was back in the Pentium eras. Everyone is feeling the heat. Texas Instruments may be backing off from new process nodes, IBM and everyone else is collaborating.

Intel is kind of on its own, and it's not because nobody wants their fab tech. Intel really is so successful that it would lose out in sharing anything with its competitors.

If AMD can't get a broader revenue stream by getting broader market share and better margins, the rise in fab costs is going to force it into being an also-ran.

It's uncomfortably close as-is. By the time its 65nm production for the desktop hits its stride in wafer starts and performance, Intel's very good 45nm will likely be one quarter from production.
 
Times aren't the same as back then. AMD used to have other business units (remember Spansion?).

It's sold off pretty much anything it can sell. It can't sell ATI right now because the debt is just too much. If things don't go well enough in the future, I'd expect AMD to consider selling whatever it can from there as well.

The cost of the lithography race is way beyond what it was back in the Pentium eras. Everyone is feeling the heat. Texas Instruments may be backing off from new process nodes, IBM and everyone else is collaborating.

Intel is kind of on its own, and it's not because nobody wants their fab tech. Intel really is so successful that it would lose out in sharing anything with its competitors.

If AMD can't get a broader revenue stream by getting broader market share and better margins, the rise in fab costs is going to force it into being an also-ran.

It's uncomfortably close as-is. By the time its 65nm production for the desktop hits its stride in wafer starts and performance, Intel's very good 45nm will likely be one quarter from production.

Your proposition that AMD wishes to (or would be wise to) sell off ATI is ludicrous. They only bought them last year for cryin out loud! AMD has yet to see any fruits of their labor. This would be the worst business decision in the history of the industry.
 
Your proposition that AMD wishes to (or would be wise to) sell off ATI is ludicrous. They only bought them last year for cryin out loud! AMD has yet to see any fruits of their labor. This would be the worst business decision in the history of the industry.

he means that they would sell IP if they have to, not sell the whole company.
 
Your proposition that AMD wishes to (or would be wise to) sell off ATI is ludicrous. They only bought them last year for cryin out loud! AMD has yet to see any fruits of their labor. This would be the worst business decision in the history of the industry.

he means that they would sell some IP if they have to, not sell the whole company.

edit : sorry for the double post
 
Last edited by a moderator:
Your proposition that AMD wishes to (or would be wise to) sell off ATI is ludicrous. They only bought them last year for cryin out loud! AMD has yet to see any fruits of their labor. This would be the worst business decision in the history of the industry.
What would happen if AMD was badly cash-starved, I believe, is that they'd try to sell specific divisions or even specific design centers, not just selling all of ATI for example. A couple of the things in the balance that they could sell if they needed short-term money but were confident in the long-term:
  • Fab 30, or Fab 36, depending on timeframes; basically relying on half the fabs + Chartered. That only works if they hope to get New York online soon enough though, or if Chartered's ramping is really, really good (and I've seen some numbers indicating they'll be very aggressive in the next 2 years, actually)
  • The handheld graphics division; this doesn't make so much sense because it's a high-growth division and a profitable one, but that also means it could sell for a nice load of money. And its synergies with the rest of the company are fairly minor; although it does benefit from ATI's graphics divisions (see: mini-Xenos), but Bitboys know their shit so who cares?
  • A specific design center, but that's more difficult in terms of finding a good buyer. For the Austin CPU design center (AMD has more than one), for example, I can imagine NVIDIA and IBM being interested; the former is not so desirable for obvious competitive long-term reasons, while IBM might be interesting. So it depends on who's interested, really. I'm fairly confident Fusion is NOT being worked on at Austin, before anyone asks.
AMD selling off specific assets like that could, IMO, happen on its own, or in the rather problematic event that a private equity buyout happens. This is all speculative, of course, but I think it's worth pointing out that there's plenty of smaller things AMD can do to get some cash if needed. On the negative side of things, some of these choices would significantly increase their 'focus'; read that to mean that they'd be a pure PC (+console?) company, and even more dependent on Intel and NVIDIA's ferocity and market trends. Of course, this all depends on whether AMD can or cannot find cash/reduce expenses another way.

Unless AMD fuckups completely, which is very unlikely, I don't think there is any chance whatsoever that Fusion would die before completion.
 
I think it would be wise to withhold judgment on the prospect of AMD having a firesale at least until R6xx and Barcelona launch and start to generate revenue.

I imagine this is AMD's philosophy ATM, anyway.
 
I think until we hear that Hector and Dirk are teetering with their Board that there's no particular cause for alarm. Will they raise money in the capital markets? Yeah, looks likely. Will they completely change direction? Not under this executive team they won't. Not a chance. So until you hear that it might not be this executive team running things, no huge changes.
 
Your proposition that AMD wishes to (or would be wise to) sell off ATI is ludicrous. They only bought them last year for cryin out loud! AMD has yet to see any fruits of their labor. This would be the worst business decision in the history of the industry.

I didn't say immediately, I said in the future. I also didn't mean to say that all of ATI would be sold/spun off, that probably would be too much.

AMD, like many other companies, has sold or spun off divisions when times were lean.
AMD used to be a major flash supplier, and flash used to be right up there with the CPU lines as one of AMD's core businesses. (Interesting note: now that AMD is no longer directly tied to the flash market, Spansion is doing better. One possible reason is that another competitor in the flash market, Intel, eased up once AMD was no longer a direct player there.)

The awkard situation was that prior to the ATI purchase, AMD was completely out of things to spin off.

After a few years, I think AMD will have a better idea of what works in their product lineup and what doesn't. After a few more years, I'd expect to see some movement to paring off sections of the combined company.

The more lean things get in two to three years, the more I'd expect them to sell to keep funding their CPU business.
I expect this to happen because it's not like things are going to get any easier for AMD.
 
Considering the transistors & heat, are they really that crappy, or is it just the impression of 'em?
I remember reading that the IGP portion of the northbridges is extremely small, and the chip does quite well compared to that.

I think they are really that crappy, personally. Though I do see what you are saying, they do okay for their size...they are still pretty bad. Granted, IGP's aren't supposed to be exciting, but wow, do they have to automatically be awful :D ?
 
I didn't say immediately, I said in the future. I also didn't mean to say that all of ATI would be sold/spun off, that probably would be too much.

AMD, like many other companies, has sold or spun off divisions when times were lean.
AMD used to be a major flash supplier, and flash used to be right up there with the CPU lines as one of AMD's core businesses. (Interesting note: now that AMD is no longer directly tied to the flash market, Spansion is doing better. One possible reason is that another competitor in the flash market, Intel, eased up once AMD was no longer a direct player there.)

The awkard situation was that prior to the ATI purchase, AMD was completely out of things to spin off.

After a few years, I think AMD will have a better idea of what works in their product lineup and what doesn't. After a few more years, I'd expect to see some movement to paring off sections of the combined company.

The more lean things get in two to three years, the more I'd expect them to sell to keep funding their CPU business.
I expect this to happen because it's not like things are going to get any easier for AMD.

I understand there is precedent for AMD selling off divisions, but I do not believe the reference to Spansion could be compared with a potential sell-off of any ATI division.

Flash memory is a commodity. There are many players in the market (and it is very volatile). The same cannot be said of GPUs. Also, Intel's aggressive pricing of their own flash products forced AMD into an unwanted price-war, destroying the margins of a formerly profitable division. Thankfully Intel cannot do the same in the x86 CPU market given the lack of competing companies and the ongoing anti-trust investigation.
 
When does Anti trust become Anti trust when Intel is still over 50% in gross margins on average for thier CPU's and AMD is below that and can't produce meaningful net profits? Intel isn't going to the point where they are cutting as much as AMD is, AMD can't compete in performance per $ at the moment.
 
When does Anti trust become Anti trust when Intel is still over 50% in gross margins on average for thier CPU's and AMD is below that and can't produce meaningful net profits? Intel isn't going to the point where they are cutting as much as AMD is, AMD can't compete in performance per $ at the moment.

You're making my point for me... Intel cannot price AMD out of the x86 market right now, because it would be obvious were they to attempt it. Their x86 margins should be higher than those of AMD, they have the superior products ATM.
 
Intel doesn't need to price equivelently to AMD that is what is happening right now, thats the whole point, Intel has gone low enough and performance of every single Intel CPU (core duo) is higher then AMD for the $. So AMD just did or are they planning on another round a price cuts, thinning margins is not where AMD wants to be at this stage, but they have to do it to keep the marketshare bleeding as low as possible. The company is already cash starved as it is.

There are many factors here not just the one you mentioned.

Just a general overview of how I look at company strength and investment for AMD:

AMD has 5 main areas of diversification in thier structure that bring in money (I know they have other areas just that money wise they are much smaller):

CPU
GPU
Motherboards
Handhelds
notebooks

CPU by far is AMD's bread maker
Handhelds are high profitablity
Motherboards go hand in hand with CPU
GPU are a finishing touch to the whole PC package
Notebooks is a factor of both the GPU, motherboards, and CPU

Handhelds won't be sold, high margins, means always money with no worries.

CPU's and motherboards go hand in hand, if we can sell a CPU, we can give a motherboard at almost no profit not a big deal, this is what Intel has been doing.

GPU definitly not as high margin as CPU's, developement time is very aggressive R&D money is increasing on a per year basis, unlike CPU cycles where it might be 2 years or more in some cases for a totally new architecture.

notebooks, good profitablity, great for selling the complete pakage and keeping margins high.

So now what seems to be the place that could possible be cut, I'm not saying it will be cut, just possibility if AMD stays in the rut they are in right now.


Now looking at the investment side of it, yeah this is exactly why AMD's stock is low, the areas where they should be minting money CPU's and GPU's, it ain't there right now. Competition on the CPU end and bungling up the launch of the r600 is taking its toll and will continue for some time. AMD can't magically wish these problems to be fixed against Intel and nV, they are trying to minimize the damage right now, and even that is not in the best interest of the company. We can always say barcelona and the r600 will help AMD come out of the rut, but well still have 2 dangerous competitors on the other side of the fence, do we know what they are doing?
 
Last edited by a moderator:
Intel doesn't need to price equivelently to AMD

I've said as much.

that is what is happening right now, thats the whole point, Intel has gone low enough and performance of every single Intel CPU (core duo) is higher then AMD for the $. So AMD just did or are they planning on another round a price cuts, thinning margins is not where AMD wants to be at this stage, but they have to do it to keep the marketshare bleeding as low as possible. The company is already cash starved as it is.

There are many factors here not just the one you mentioned.

Just a general overview of how I look at company strength and investment for AMD:

AMD has 5 main areas of diversification in thier structure that bring in money (I know they have other areas just that money wise they are much smaller):

CPU
GPU
Motherboards
Handhelds
notebooks

CPU by far is AMD's bread maker
Handhelds are high profitablity
Motherboards go hand in hand with CPU
GPU are a finishing touch to the whole PC package
Notebooks is a factor of both the GPU, motherboards, and CPU

Handhelds won't be sold, high margins, means always money with no worries.

CPU's and motherboards go hand in hand, if we can sell a CPU, we can give a motherboard at almost no profit not a big deal, this is what Intel has been doing.

GPU definitly not as high margin as CPU's, developement time is very aggressive R&D money is increasing on a per year basis, unlike CPU cycles where it might be 2 years or more in some cases for a totally new architecture.

notebooks, good profitablity, great for selling the complete pakage and keeping margins high.

So now what seems to be the place that could possible be cut, I'm not saying it will be cut, just possibility if AMD stays in the rut they are in right now.

I'm not one to judge the commercial viability of a product (about which very little is known anyway) prior to launch. I leave that up to those that actually know what the product is capable of ;)

Besides, even if R6xx is a sales flop (no pun intended), there's always the refresh lineup.
 
I understand there is precedent for AMD selling off divisions, but I do not believe the reference to Spansion could be compared with a potential sell-off of any ATI division.

Flash memory is a commodity. There are many players in the market (and it is very volatile). The same cannot be said of GPUs.
Much the same can be said for ATI's chipset division. The buyout basically killed two thirds of ATI's business in that arena. Off the bat, I can think of VIA, Intel, Sis, and Nvidia as competitors.

AMD also used to be in the chipset business. It's not anymore.
That's what I'd probably see being sold or shut down first.

Also, Intel's aggressive pricing of their own flash products forced AMD into an unwanted price-war, destroying the margins of a formerly profitable division. Thankfully Intel cannot do the same in the x86 CPU market given the lack of competing companies and the ongoing anti-trust investigation.

Sure it can, and it has. AMD's incredibly depressed ASP is evidence of this. There's no law against having competitively priced products. Intel is not selling its products at a loss, and it's already taken a beating on margin to put the hurt on AMD. Last I checked, AMD was losing money, so it's obviously having an effect.
That's also not illegal.

Whatever antitrust litigation is not related to Intel's general pricing strategy, but on less obvious exclusionary practices, some of which may or may not be illegal.
If we go by history, Intel has little to fear in the event that it is found guilty of antitrust violations.
I'd look at the penalty levied against Microsoft.
 
I think they are really that crappy, personally. Though I do see what you are saying, they do okay for their size...they are still pretty bad. Granted, IGP's aren't supposed to be exciting, but wow, do they have to automatically be awful :D ?

Well, they are usually designed to just allow the OEM to check the necessary boxes. With Vista actually using the 3D pipeline, we might see IGP's get less awful. But I don't know how much less awful, with low cost being the primary design goal.
 
Well, they are usually designed to just allow the OEM to check the necessary boxes. With Vista actually using the 3D pipeline, we might see IGP's get less awful. But I don't know how much less awful, with low cost being the primary design goal.

less awful with better drivers :cool: at least. If the rumours are true, the RS780 (or 790 cant remember) will integrate a RV610, which is also less awful.
 
Looks like RD790 has successfully completed the rigorous testing procedures of the Compliance Workshop.

Source X-bit labs:
Advanced Micro Devices has quietly validated its next-generation core-logic code-named RD790 with PCI special interest group (PCI SIG) which supervises implementations of PCI and PCI Express busses. The validation means that the company may start shipping its first high-end core-logic in years commercially to mainboard makers.
 
  • Like
Reactions: Geo
Looks like RD790 has successfully completed the rigorous testing procedures of the Compliance Workshop.

Source X-bit labs:

Yeah, this puzzled me a bit. Somehow I wasn't expecting commercial products on the 790 line until Q4. So I don't quite know what to make of this report. Is this typical lead time for such things, or should we be moving up our expectations on the timeline?
 
Back
Top