The AMD Execution Thread [2007 - 2017]

Status
Not open for further replies.
And if you want to get *even* more totally and utterly crazy... Wouldn't this theoretically allow *any* company to manufacture x86 CPUs @ FAT? :) If so, that might make the name of the companies financing the spin-off a much more interesting piece of information!

If things got that crazy, possibly.
The catch is that those outside parties might have to transfer ownership of their designs in some way to AMD(FAT). The agreement seems to be restricted to the products belonging to either Intel or AMD.
The complexity in setting up such a system would take some time to figure out.

If FAT were merely fabbing, those parties would still own the x86 products, and that would imply that AMD is transfering rights to other entities. It may or may not be a problem, I don't know.

It's no problem for DEAD, because they're basically running alongside FAT anyway.

I also wonder if AMD could get greedy and restrict it so only DEAD's x86s are made in order to keep smaller players from killing the DEAD designs from below. It's far more likely AMD would get hurt before Intel by a sudden flowering of x86 designs.

On the other hand, getting a lot of x86 designs in the fab means FAT has more of a reason to maintain a high-end digital CPU process. Inviting in just any fab customer is going to bring in clients the x86 process wouldn't appeal to.

If FAT can get away with fabbing the world's x86 fringe, it would actually be more valuable.
DEAD would take a hit, though.

edit:
There is one other possible catch.
The license agreement would mean patents that go into AMD licensed products would be cross-licensed to Intel. If other parties produce x86 through AMD, they better not have any patents in their designs that they want to keep to themselves.
Once again, this is no skin off the back of the DEAD designers.
 
In 2009, the IGP segment will see the demise of ATI, and the Rise of IMG.

Since that post on 14-Dec-07:-

ATI/AMD heading out of mobile graphics (ok still no move on IGP)

IMG in bed with Intel with UMPC and MID
in bed with apple via iphone and itouch

Sometimes I surprise even myself :)
 
Since that post on 14-Dec-07:-

ATI/AMD heading out of mobile graphics (ok still no move on IGP)

wow.. that's a bold statement to make since amd show the biggest market gains in mobile in their (ati's) entire history..
 
wow.. that's a bold statement to make since amd show the biggest market gains in mobile in their (ati's) entire history..


By mobile, I meant mobile phone, i.e. handheld. Clearly AMD have said nothing about laptop/notebook segment.

"The company announced along with the release of its second-quarter earnings results Thursday that it is getting out of the handheld and digital television businesses"
 
By mobile, I meant mobile phone, i.e. handheld. Clearly AMD have said nothing about laptop/notebook segment.

"The company announced along with the release of its second-quarter earnings results Thursday that it is getting out of the handheld and digital television businesses"

ah. .that clears up a lot of stuff, by mobile we mean notebooks/laptops by handheld we mean handhelds..
 
Is it true that AMD's process is the best of all, except for Intel's? Some guys at amdzone, specialy Scientia, say so. I'd like to know why, because I can't understand the reason.
 
Is it true that AMD's process is the best of all, except for Intel's? Some guys at amdzone, specialy Scientia, say so. I'd like to know why, because I can't understand the reason.

Well, we know Scientia is not neutral when it comes to comparing AMD with anybody. He knows his stuff but then again compare what he has said about Conroe before it came out and then what he said after, downplayed it a lot even though it was plainly obvious that at least for desktop use it was more than competitive. Now they are doing the same again for Nehalem but this time they are picking single threaded apps as being no better than Penryn. We shall see.

Having said that AMD's 45nm process on SOI seems to have not hit any big problems yet and may well be released at same of bigger clockspeeds than the previous process size, which is bucking the trend for AMD. They just need to get it out quick now.
 
Notice that I said AMD's process. I didn't spacify any specificaly. That's the problem... How the hell can anyone say that the process of TSMC's 32nm for high performance application is worse than AMD's?
 
Notice that I said AMD's process. I didn't spacify any specificaly. That's the problem... How the hell can anyone say that the process of TSMC's 32nm for high performance application is worse than AMD's?

They can't; the problem is that AMD is being given an honorific here without any specific criteria associated with it. What does "the best" mean when there are a lot of different goals a fab can have in mind when pursuing a process? The best for wattage, transistor density, yields, clock speeds... what? Not for all of those, to be sure.
 
Is it true that AMD's process is the best of all, except for Intel's? Some guys at amdzone, specialy Scientia, say so. I'd like to know why, because I can't understand the reason.

Without agreeing on the criteria used, no agreement can be reached.

For circuit performance for high-performance x86, this is pretty much true, but then again there are only two high-performance x86 manufacturers.

This has been referenced before:

http://www.realworldtech.com/page.cfm?ArticleID=RWT011608222300&p=6

IBM's various processes and by extension AMD tend to be better on most things compared to most process players and the foundries.
Intel's on most measures are at the top of the heap.

Historically, TSMC's circuits have lagged behind in circuit timings compared to Intel on roughly the same geometries.
 
Any news on the launch for the 45nm version of Phenom yet? I know there have been a few pics of engineering samples on the web, but any news as to if AMD are managing to execute properly this time around?
 
I'm guessing this talk of AMD's 45nm prowess was precipitated by someone seeing #s for 45nm Phenoms, because they seem to have made huge gains (drops?) in power draw. The 2.3GHz Deneb almost halved load draw and made nice improvements in idle compared to the equally clocked 9650. Pretty sweet.

I didn't pay attention to performance, but I'm sure the extra cache doesn't hurt, even if it probably won't bring 'em even with Nehalem. I think the HardSpell review (which tested on a 790FX: AM2+) says Deneb should work with AM3 and DDR3, so it might test better in the future. (It sure makes future upgrades from a cheap X2+AM2 setup more interesting, something I'm looking at.)

It's a step in the right direction, anyway.
 
http://translate.google.com/transla...-online.de/nachrichten/artikel.asp?id=1912570

AMD is selling the older hardware from Fab30 in Dresden to Russia.
The hardware was used with 200mm wafers is being bought by russian electronics company called Angstrem, and they have reserved apparently 132 million euros to buy the tech & hardware, which will the be transferred to their own plant.

The company has some history with AMD already, as they bought earlier 130nm process info/knowledge from them
 
I wonder what the "portion" going into Escrow is about? They have some targets they have to hit on DTV employees accepting Broadcom employment to get that money? I can't think what else it would be.

Presumably they kept rights to keep using current Xilleon tech. What's more interesting is what happens down the road if they need an update to that tech if they've lost most of the engineers. But then this isn't a terribly bad time in the industry to be doing this. The basic TV tech should stay relatively stable for some years, and most of the post-processing goodies happen in the shader core not the Xilleon tech.
 
Dirk Meyer confirmed the spin off strategy today, at this FORTUNE interview:

"Since it can’t afford to do this anymore, AMD plans to spin off its chip manufacturing operations by year’s end by either by hawking them outright or by inking a partnership with a larger chipmaker – a maneuver akin to selling a house and leasing it back. Meyer is vague on the exact timing of a deal, but he knows it’s probably the best thing the company can do quickly to improve its financial position, and its reputation with investors."

[...]

"“We’re going to go away from a captive fab model to more of a fables model for the CPU part of the business,” Meyer says. “Longer-term, it relieves us of the burden of having to shell out cash for these gigantic factories. So it will be more of a pay-as-you-go model like a traditional fables semiconductor company.”"


http://bigtech.blogs.fortune.cnn.com/2008/09/08/re-engineering-amd/

Some people at amdzone doesn't believe that is a confirmation:

http://www.amdzone.com/phpbb3/viewtopic.php?f=52&t=135518
 
I don't think it will be that simple given the issues related to the X86 licence. I can see them spinning the fabs into a separate company and then selling 49 percent of it to someone else (most likely investors). AMD get a load of cash and become fabless, though the new company of FAB Inc is still effectively operating a fab for AMD and is owned primarily by AMD or a holding company.
 
Status
Not open for further replies.
Back
Top