AMD eating ATI up, ditching high-end GPU's?!?

Bo_Fox

[H]ard|Gawd
Joined
Aug 23, 2006
Messages
1,544
http://www.xbitlabs.com/news/video/display/20061029130237.html

Oh no!!!!!!!!!!!!!!!!!!!!!! NO!!!! NO!!!!!!!!!!!!!!!!

I really hate that analyst! Can somebody kill him please?!?

Let's pray that AMD/ATI will continue making higher-end cards after the R600. If not, then we'll all be left to Nvidia's slow progress, a repeat of the GF2 --> GF3 ---> GF4 until some other company surprises Nvidia all over again.

Except this time, it might take much longer (we're looking at 5+ years of "slow" progress here). Nvidia monopoly??

Why did AMD have to buy ATI? I think AMD was being lazy in the chipset department. AMD could have bought some small integrated chipset company or just made a contract with both Nvidia and ATI to keep on designing some really nice integrated graphics for them.

Seeing the direction that AMD is headed after buying ATI with more than half of the money that it has, I do not expect AMD to go anywhere for at least 3 years. Intel and Nvidia are grinning with a big, stupid preposterous smile--so big that they cannot help it right now.

*Maybe* it is indeed a good move for AMD, to prepare for the inevitable future of combined CPU/GPU chips??? That wont be for another 3 years for sure, though... So, we'll all be experiencing slow stuff in the meanwhile.
 
And as I look into my crystal ball.. i can see NOTHING.

It is all prediction and speculation. There is no hard proof that this is going to happen and it is stories like that that keep us on our toes.
 
I don't believe anything in that article for a second.

Especially if the discrete graphics card business actually account for 50% of ATi's annual revenue. If that part of the article is even accurate.

I think AMD bought ATi, to access their chipset designs, and to access ATi's various technologies.

Even if they want to start putting graphics capabilities into their chips, there will always be more powerful discrete graphics solutions. I think they understand this.
 
I really hope AMD and ATI come up with something really innovative and cool in terms of GPU/CPU capabilities. Maybe a dual core GPU chip or something hehe :D.

All I know is if they come up with something awsome, I'll be the first in line for an AMD/ATI solution. Sorry Nvidia :cool:
 
Makes no sense, the only thing that will be changing with ATI that will be major is their management, hopefully to help them get products out faster and with better reliability and not so many disasters (IE the R520)
 
Brent_Justice said:

Nice to hear from you, but I already read those slides a while ago.

Money is name of the game, and now AMD has to figure out how to maximize its profits for the following year, and the next, and so on...

There's no saying that AMD cannot change its mind anytime during 2007.

The cold-and-hard fact is that the board of directors for AMD do not give a hamster's stick about providing top-end GPU's that will seriously compete with Nvidia. Neither do 90+% of the stock-holders. All they care about is cash-flow on a grand-scale (and the high-end GPU's now only constitutes 1% of AMD's net income).

The main reason why AMD acquired ATI was for the inevitable future of CPU-and-GPU merging. A minor reason is for better integrated chipsets, an area in which AMD has slacked off on a while ago. Manufacturing high-end GPU's is definitely not one of them.

So, let's see how the business world turns out to be. There's nothing better than using the cold-and-hard logic.
 
I doubt it would happen, but ummm seriously from a business pov, ati makes more money I imagine on midrange and low end gfx and so does nvidia. I doubt the number of people buying the 800 and 900 series gfx from nvidia and ati make a large percentage of their sales, I bet its in the 200, 300 and 600 gfx cards that make the bulk of their money. Also mobile gfx solutions is another area where they make more money then high end gfx. I think if I am not mistake that Intel gfx solutions actually out number both ati and nvida, and well they don't have the intel 449000 RBGX Dual Quad Extreme Graphics Interlink

Discrete gfx for workstations, and business solutions would probably make more sense for amd and ati, and would make more money for the both, this includes servers and workstations. When is the last time any business has ordered 2000 workstations with 7900GT's in SLI, prolly not maybe a workstation with Quadro/FireGL cards yes but again few.

But that said there is something about ATI having a performance crown and AMD knows that all too well, sometimes having the performance crown people automagically assume that every offering is better then their competitors in every category. You know the 1900XTX > 7900GTX therefore ati 9200 > 7300 kind of mentality without even looking at a review.

/end rant.
 
Good point there.

Yep, it's the reputation, but Intel never needed that in the video card arena. Neither does AMD.

Unless the board of directors is strongly balanced between those of AMD and those from ATI, AMD will eventually start acting just like Intel. Heck, AMD can still beat Nvidia by having a greater market share of total GPU sales. Intel always had nearly 50% of the entire GPU market for the past several years, because of dominating the integrated chipset and notebook graphics market.

AMD's board of directors will certainly be tempted to follow Intel's model.

I just hope that ATI really stays alive within AMD and keeps on fighting hard! Don't give up, ATI, please!!!
 
Bo_Fox said:
Nice to hear from you, but I already read those slides a while ago.

Money is name of the game, and now AMD has to figure out how to maximize its profits for the following year, and the next, and so on...

There's no saying that AMD cannot change its mind anytime during 2007.

The cold-and-hard fact is that the board of directors for AMD do not give a hamster's stick about providing top-end GPU's that will seriously compete with Nvidia. Neither do 90+% of the stock-holders. All they care about is cash-flow on a grand-scale (and the high-end GPU's now only constitutes 1% of AMD's net income).

The main reason why AMD acquired ATI was for the inevitable future of CPU-and-GPU merging. A minor reason is for better integrated chipsets, an area in which AMD has slacked off on a while ago. Manufacturing high-end GPU's is definitely not one of them.

So, let's see how the business world turns out to be. There's nothing better than using the cold-and-hard logic.
Logic can only tell you whether a conclusion follows from a premise. It doesn't tell you whether the premise itself is sound, and yours isn't.

There's a certain amount of capacity in the high-end GPU market, enough for two major companies. If a third one entered, there wouldn't be enough profit to go around and it would either exit the market or cause one of the existing players to do so (like 3dfx was forced to). As long as that profit potential exists, ATI will stay in the game. They already spent a lot of money assembling a network of hardware and software specialists and a distribution network, which is a sunk cost.

Now here's the thing: If there's a lot of money to be made by expanding the notebook and embedded GPU divisions, they can do that without abandoning the high end. They wouldn't decide not to continue reaping the rewards from a market they've sunk billions into in order to get to their current position. You act as if it's impossible for them to simply raise the money to expand into the new market, and they somehow need to cannibalize a profitable venture. Once they reach the optimum supply point in low power and embedded GPU's, there's no sense in throwing away the high-end market to focus even more into the low-end, when they're already at the maximum profit making point. That would mean not only giving up the profits from the high-end for no reason, but also giving up some profit from the low end.

The fact that the low end might be more profitable is completely irrelevant. iPods are more profitable than Macs, but I didn't see Apple getting out of the computer hardware market.
 
Keep in mind that having the 'best' available graphics solution would likely help them sell CPUs as well. For a less informed buyer they'd be more inclined to buy the same brand of CPU as GPU so they could pair them up. Having the fastest solution is more a form of advertising than just a financial decision.

Also keep in mind with the cards going unified I'd imagine the R&D that is required would be significantly less since every performance range essentially uses different numbers of the same parts. So 4 units low end, 12 mid range, 36 high for example. I'm guessing the memory controller would be almost identical for each SKU.
 
Endurancevm said:
Well Apple wont be using anymore ATI products...

why is that? sorry but umm most of the ati gfx solutions in apples are 600 series cards not, 800 and 900 series cards. The MacPro comes with a 7300GT (up to four) a x1900xt or Quadro 4500. Again by default it comes with a 7300GT. The iMacs use, either Intel integrated gfx, x1600 or 7300GT or 7600GT. Mac mini Intel gfx same with the macbooks, and macbook pros use x1600 mobile chips.

Apple offers only one high end card, and thats an extra on a macpro.

Apple has gone on record that it wont discriminate on AMD and Intel chips, both are x86 and use AMDs 64 bit extensions. Apple has used AMD chipsets (according to the inq) before on the G5's so yeah, I seriously doubt ATI has to worry about sales to Apple. If AMD could meet demand, and build enough chips I wouldn't be surprised if Apple did offer AMD chips, in either, the imacs or mac minis maybe even opterons in Macpros but really core2duos give better performance and well Intel can ship them cpus.
 
arthur_tuxedo said:
Logic can only tell you whether a conclusion follows from a premise. It doesn't tell you whether the premise itself is sound, and yours isn't.

There's a certain amount of capacity in the high-end GPU market, enough for two major companies. If a third one entered, there wouldn't be enough profit to go around and it would either exit the market or cause one of the existing players to do so (like 3dfx was forced to). As long as that profit potential exists, ATI will stay in the game. They already spent a lot of money assembling a network of hardware and software specialists and a distribution network, which is a sunk cost.

Now here's the thing: If there's a lot of money to be made by expanding the notebook and embedded GPU divisions, they can do that without abandoning the high end. They wouldn't decide not to continue reaping the rewards from a market they've sunk billions into in order to get to their current position. You act as if it's impossible for them to simply raise the money to expand into the new market, and they somehow need to cannibalize a profitable venture. Once they reach the optimum supply point in low power and embedded GPU's, there's no sense in throwing away the high-end market to focus even more into the low-end, when they're already at the maximum profit making point. That would mean not only giving up the profits from the high-end for no reason, but also giving up some profit from the low end.

The fact that the low end might be more profitable is completely irrelevant. iPods are more profitable than Macs, but I didn't see Apple getting out of the computer hardware market.

Agreed 100%.
 
Anarchist4000 said:
Keep in mind that having the 'best' available graphics solution would likely help them sell CPUs as well. For a less informed buyer they'd be more inclined to buy the same brand of CPU as GPU so they could pair them up. Having the fastest solution is more a form of advertising than just a financial decision.

Also keep in mind with the cards going unified I'd imagine the R&D that is required would be significantly less since every performance range essentially uses different numbers of the same parts. So 4 units low end, 12 mid range, 36 high for example. I'm guessing the memory controller would be almost identical for each SKU.

One reason why AMD might not axe the high end gfx solution is this, ATI makes both Nintendos and Xbox 360 GPUs, this means they need to develop high end gfx if they still want the console market.

Also we know R600 is gonna be a unified architecture, this means that with vista rearing its head, AMD boards can have integrated gfx that are DX10 compliant and vista ready, meaning AMD has a leg up on intel as they can offer vista ready workstation boards, where as intel has to either develop thier own unified shader gpu or get an nvidia gpu or have a dx 10 compliant gpu. Granted that unified shader != direct x, ala G80. But clearly I think the AMD ATI merger was nothing more then AMDs grab into the workstation and server market.

I really wouldn't be surprised if AMD and ATI make a third company or brand that offers AMD chipsets with integrated gfx. If it sounds like I am flip-flopping, or my argument isn't clear its I doubt they will get rid of their high end gfx solutions, but I wouldn't be surprised if it happened.
 
The Inq maintains the position that AMD will make a CPU with a GPU on it.

I doubt seeing ATi give ground in the high-performance graphics market. Why should they just roll over to them instead of trying to crush them or buy them? Nvidia was left out in the cold with this DAAMIT acquisition, no CPU partner to run to.

 
arthur_tuxedo said:
Logic can only tell you whether a conclusion follows from a premise. It doesn't tell you whether the premise itself is sound, and yours isn't.

There's a certain amount of capacity in the high-end GPU market, enough for two major companies. If a third one entered, there wouldn't be enough profit to go around and it would either exit the market or cause one of the existing players to do so (like 3dfx was forced to). As long as that profit potential exists, ATI will stay in the game. They already spent a lot of money assembling a network of hardware and software specialists and a distribution network, which is a sunk cost.

Now here's the thing: If there's a lot of money to be made by expanding the notebook and embedded GPU divisions, they can do that without abandoning the high end. They wouldn't decide not to continue reaping the rewards from a market they've sunk billions into in order to get to their current position. You act as if it's impossible for them to simply raise the money to expand into the new market, and they somehow need to cannibalize a profitable venture. Once they reach the optimum supply point in low power and embedded GPU's, there's no sense in throwing away the high-end market to focus even more into the low-end, when they're already at the maximum profit making point. That would mean not only giving up the profits from the high-end for no reason, but also giving up some profit from the low end.

The fact that the low end might be more profitable is completely irrelevant. iPods are more profitable than Macs, but I didn't see Apple getting out of the computer hardware market.

Your premise does sound quite logical.. (playing with words here.. ), but anyway, it is not completely viewed from all angles. You seem to be absolutely certain with how things will turn out with AMD.

See, Nvidia is now the only independent GPU maker. AMD slacked off with making its own chipsets and left it up to third-party manufacturers.. which was a bad move that made AMD have to rely upon Nvidia, VIA, SIS for motherboard releases in spite of delays and issues. If the pattern holds, AMD could also slack off with making high-end GPU's.

I am not saying that AMD will stop at R600. What seems most likely to me is that AMD will significantly "slow-down" after the R600, and let Nvidia become the king. Heck, AMD will let Nvidia beat it with GPU's that are twice as fast by the beginning of 2008 when AMD has an R640 while Nvidia has a G90+ or something like that.

Why? It is because Nvidia still makes chipsets for BOTH Intel and AMD. Intel makes it own integrated graphics, and AMD wants to do the same. Also, AMD sees a need to prepare for the inevitable future of combined CPU/GPU technology that could almost be "standard" beginning with the 45nm process technology.

Nvidia will *continue* to provide a need for faster CPU's, which benefits AMD. So, it allows AMD to slack off in the high-end gfx department. I think that the R600 will be much faster than the G80, but I strongly believe that it will be the last time ATI is king.

Re: the console market... oh yeah, AMD can continue making gfx solutions for future consoles with little tweaks here and there. The gfx for Nintendo Wii is hardly any better than the previous one. Guys, do you think that the Playstation 4 or Xbox III would be a priority for AMD right now?

So, the more I think about it, I do not see AMD abandoning the high-end GPU market. Instead, they will just eventually slack off with the high-end stuff by hardly improving their stuff after the R600, to the point where Nvidia just dominates. As long as Intel does not buy Nvidia or make an exclusive contract with Nvidia, AMD has Nvidia also.
 
This thread is full of a bunch of idiotic armchair CEO's...

getting the volume to make money in discrete graphics (50% of ATI) is about keeping up/exceeding your competitors in performance.

ATI as part of AMD will continue to do so because even after the R&D money it's profitable. That's not to say they may back down a bit from the 6 month release cycles or always offer the best performing most expensive highest end card in the market but they aren't going to give up all that marketshare to nvidia.
 
Ok, the main motive for AMD to want to keep on competing with Nvidia in the high-end is if AMD/ATI creates its own special gfx solutions like super-quad Crossfire with R700's that will not be available on Intel motherboards. To support that, there are already well-sourced rumors that the Crossfire license is going to expire for good with Intel. A less risky move would be to make sure that Crossfire/Intel solutions just cannot match the gfx power offered on AMD's platforms by a significant margin.

However, this margin would be quite pointless if Nvidia competes fiercely enough with ATI in the first place while still offering at least equal gfx performance on Intel platforms. So, AMD/ATI would have to completely pwn Nvidia for this to really work... (which is unlikely seeing how fierce Nvidia is)!

Refusing Crossfire on Intel would be a very very risky move for AMD if it ever did that. Who knows what Intel's payback would be... making an exclusive contract with Nvidia's SLI or even buy out Nvidia? Maybe buying ATI was a preemptive move made by AMD after sensing that Intel was planning on doing the same by buying either ATI or Nvidia? AMD needed that more than Intel anyway...

Unless AMD knows for sure that Intel cannot make such a deal with Nvidia, or that Intel absolutely cannot buy out Nvidia due to excessive monopoly practices, AMD should go ahead and refuse Xfire support for Intel. Well, I bet that AMD is not 100% certain about Intel's abilities and strategy. It's quite possible for Intel to get away with buying out Nvidia (especially after AMD did it first.. saying, "but he did it first"), so AMD probably does not want to encourage that, or else AMD would be guaranteeing itself a perpetual second-place Silver medal in the field.

As long as AMD does not try to punch Intel too hard, Intel probably will take it easy while basking in its Core2 glory. AMD will probably just gradually slack off to the point where Crossfire is no longer a great thing after Nvidia has long surpassed it. Then Intel will not care about Crossfire, and leave Nvidia alone with no need for vengeance (since Intel was never threatened in the first place). Of course, Nvidia will continue to provide the fastest gfx solutions for both and make more $$$ than ever while AMD strictly focuses on developing a magical card up its sleeve against Intel's Core2 stuff.

Maybe after AMD develops that magical card, AMD/ATI will decide to release another "9700 Pro" after its old 8500, which is the R600 respectfully. It'd be a chess move that would shock all others with a checkmate!

Only the dumb ones do not see the history being repeated (i.e., ATI/AMD slacking off, then causing Nvidia to enjoy its fame while starting to get lazy like it did with its GF3/GF4 series... and Intel not sensing any immediate threats, then AMD decides to conquer all with the R800+ and K9 while refusing R800 support on Intel platforms)! Intel & Nvidia = dumb??
 
madgun said:
Dont worry.. Intel is about to surprise nVidia in a major way.

What the heck do you mean? That Intel is going to offer $12 billion to buy out Nvidia?
 
One fact remains constant in the Technology Biz. You have to innovate to stay alive. Period.

ATI, as well as Nvidia both have to create new technologies to stay alive. Todays "High End" is tommorows Mid Range and the next days Entry Level. Break that cycle for even one generation and your Multi-billion dollar company quickly becomes worthless.

AMD and Intel both know this Very Well.
 
TheBluePill said:
One fact remains constant in the Technology Biz. You have to innovate to stay alive. Period.

ATI, as well as Nvidia both have to create new technologies to stay alive. Todays "High End" is tommorows Mid Range and the next days Entry Level. Break that cycle for even one generation and your Multi-billion dollar company quickly becomes worthless.

AMD and Intel both know this Very Well.

So does nVdia. They slipped with the FX series, and they had to fight damned hard to get back what they had lost.
 
Hey big surprise, another ATI is dead thread. Why don't people just calm down and wait and see what happens. I highly doubt ATI will stop making high end graphics, but unless anyone here is on the board of directors at AMD, its all foolish speculation anyway.
 
Dan_D said:
So does nVdia. They slipped with the FX series, and they had to fight damned hard to get back what they had lost.

The FX series came after being lazy with the GF3 and GF4 series. Then when Nvidia was ambushed with the 9700 Pro, Nvidia had to rush an FX 5800 Ultra right out the door within a matter of months. Hence, a lack of complete DX9 design... new generation designs are supposed to take up to 16 months.

That's what happens when you let Nvidia be king for too long. It's like setting up Nvidia for an ambush all over again soon in the future, but Intel will also be a victim this time.
 
Right now, I just do not see how AMD can possibly compete at all. So, it's probably time for AMD to just lay low for a while and do just that.
 
It will be interesting to see where AMD gets the money to pay off the loan they needed to buy ATI.

While us computer geeks can argue about high end GPUs, the shareholders don't give a sh** what we think as we only represent a small amount of sales. It's what management thinks is right and that could mean just about anything.
 
Bo_Fox said:
Nice to hear from you, but I already read those slides a while ago.

Money is name of the game, and now AMD has to figure out how to maximize its profits for the following year, and the next, and so on...

There's no saying that AMD cannot change its mind anytime during 2007.

The cold-and-hard fact is that the board of directors for AMD do not give a hamster's stick about providing top-end GPU's that will seriously compete with Nvidia. Neither do 90+% of the stock-holders. All they care about is cash-flow on a grand-scale (and the high-end GPU's now only constitutes 1% of AMD's net income).

The main reason why AMD acquired ATI was for the inevitable future of CPU-and-GPU merging. A minor reason is for better integrated chipsets, an area in which AMD has slacked off on a while ago. Manufacturing high-end GPU's is definitely not one of them.

So, let's see how the business world turns out to be. There's nothing better than using the cold-and-hard logic.


Perhaps your correct however the high end is what drives the technology foward and so it has its place. RD is RD and while a small percentage of actual sales , I suspect the revenue is greater than 1%
 
jacuzz1 said:
Perhaps your correct however the high end is what drives the technology foward and so it has its place. RD is RD and while a small percentage of actual sales , I suspect the revenue is greater than 1%

Yeah, for heck's sakes, it might be 5%. Let me tell you.. I really do hope that ATI/AMD will continue making high-end GPU's that compete well with Nvidia. That would be something unlike AMD's laziness in the chipset department over the past couple years..
 
PRIME1 said:
It will be interesting to see where AMD gets the money to pay off the loan they needed to buy ATI.

While us computer geeks can argue about high end GPUs, the shareholders don't give a sh** what we think as we only represent a small amount of sales. It's what management thinks is right and that could mean just about anything.
That's what I've been wondering about this (and other buyouts)...if one company buys another company for 1.5 billion dollars, then they own both the company who paid the 1.5 billion, AND the company who got the 1.5billion, did they really spend anything?
Isn't it like selling yourself a box of Oreos? You pay yourself $1.50 for the Oreos, you take the Oreos and eat them, while you enjoy the $1.50 you just got paid for selling your box of Oreos.
 
LOL!!! That's a good way to look at it!!! Never thought of it myself that way... You're cool in my book!!
 
Star Fox said:
That's what I've been wondering about this (and other buyouts)...if one company buys another company for 1.5 billion dollars, then they own both the company who paid the 1.5 billion, AND the company who got the 1.5billion, did they really spend anything?
Isn't it like selling yourself a box of Oreos? You pay yourself $1.50 for the Oreos, you take the Oreos and eat them, while you enjoy the $1.50 you just got paid for selling your box of Oreos.
Not quite. The price of a company is based not only on what it's worth today, but what it is projected to be worth in the future discounted to the present value and adjusted for risk. It's just like any other investment.
 
Yeah, I'm not a businessperson myself, but now I'm thinking about it....

Doesn't that money go to the stockholders? It's like AMD paying its own stockholders $5 billion in cash. Right?
 
Bo_Fox said:
The cold-and-hard fact is that the board of directors for AMD do not give a hamster's stick about providing top-end GPU's that will seriously compete with Nvidia. Neither do 90+% of the stock-holders. All they care about is cash-flow on a grand-scale (and the high-end GPU's now only constitutes 1% of AMD's net income).

Wait.
So, you are saying that, for some reason.... amd spent 5.4 BILLION dollars so that nvidia could get a monopoly on the video card market?


Nvidia must REALLY love amd for that one.
 
CyberDeus-RagDoll said:
Wait.
So, you are saying that, for some reason.... amd spent 5.4 BILLION dollars so that nvidia could get a monopoly on the video card market?


Nvidia must REALLY love amd for that one.

Yeah, that is quite possible.

AMD's priority is its own survival in the long term. I think that it was the best decision AMD ever made, by buying ATI. If AMD never slacked off in the chipset department and actually tried to develop its own graphics department, it would have never really needed ATI in the first place. I do not think that Intel needs Nvidia at all. Buying Nvidia would most likely do more harm than good to Intel, by opening a "Monopoly" can of worms like Microsoft did.

But, AMD better not threaten Intel to the point where Intel think that it's worth opening that can of worms!
 
It was very depressing to go get my latest catalysts today and seeing AMD green all over the place. In all honesty gpus, as a gamer, I mean cpus take a seat far in the rear. An integrated cpu/gpu will be good for joe blow who wants to play UT2003 at work. I couldn't give a fuck less.
 
This is so old. At least the report that started it all.

I'm sure Nvidia !!!!!!s will keep bringing it up over and over though, wishful thinking.

You can hope this comes true but sorry, that's false. R600 on the way.
 
AMD has a lot of options now and it seems there is a good way to screw up a lot of them. If they integrate the best GPU silicon ati has on to the CPU that makes changes to the core more time consuming. AMD's cpu is gonna be like the titanic, big, clumsy and takes forever to change direction. Intel popped out a ddr2 chipset like nothing when it took an entire socket change on AMD's side. Imagine AMD not only making changes to future memory technologies to compete with Intel but making drastic changes to the GPU portion of the core to compete with nVidia.

I keep coming back to the memory problem with GPU/CPU integration. DDR2-800 is fast but nothing compared with 2GHz memory on video cards. Not only that, but sharing it with the CPU?? We critisize intel for starving a quad core on the same memory bus of a single core, AMD will find similar problems.

AMD needs a uniform way to go about this. Putting the best GPU core on the CPU I dont think will work for reasons listed above. Integration is great but I think nvidia will have to die first. So how about making a new socket on the AMD motherboard to house only the GPU.. Not sure who first suggested that, but I think that would ultimately be the best thing. Imagine having a heatsink similar to one used on a cpu being used on a gpu. They can be taller and thus cool alot better. By putting them in a package similar to that of a cpu, it gives the gpu more legitimacy. No longer takes up expansion slots. It could be located geographicaly closer to the cpu thus lower trace latency. It could be connected to the cpu directly (trace wise) using hypertransport which is still better than pci-e. This would give them an easy advantage over nVidia. The remaining problem is memory again. I say (if it will fit) place a new memory slot on the motherboard just for this GPU socket. This can contain something faster like DDR4. This way people can upgrade the amount of memory for their video cards(chips rather). If the memory is sucessful AMD can use it as a guinipig (sp?) for what it could later use on its CPU.

To wrap this up (sorry guys a lot of ideas here) AMD should integrate the pci-express controller on to their CPUs. Infact I thought I heard this as a posiblity. That would give them a boost on nvidia cards instantly. We also know that pci-e is new and shouldnt be going anywhere for awhile. AMD can also make video cards for the intel platform, but would have to put the GPU socket on a standard video card (but no biggy). AMD could then license the GPU socket idea to intel/nVidia and make some cash on the side.

I dont think the integration idea should be totaly shot though. I think AMD should grab one of ATIs old GPU designs like the 9000, something with a low transistor count, and slap it on all CPUs. A radeon 9000 can get that instant boost from integration that can put it over the top of maybe some modern video cards. The power savings in the mobile market would be huge. It wouldnt beat up a 7800go but it would be cheap, and businesses and budget notebooks would eat it up.

For desktops, the old integrated radeon can be turned off via the bios and a GPU chip can be inserted in the GPU socket if you needed something with a little more vrooom. Or if nvidia does infact make something that outdoes it all, pop in a nvidia and have it roar as it is running off of an integrated pci-e controler on the CPU.

Anywho... they probably will screw it up.
 
Back
Top