ATI to the rescue?

[Cogman]

Well, It looks like AMD might not be in as bad of shape as we CPU people seem to think. I just read Anand's review of their new graphics card (4850) and I have to say, they are getting much better in the Video department.

I think that most here felt after the initial buying of ATI and a few months past it something like "Crap, AMD is going to take 2 good companies down with it" in fact they had a very rocky 2 years in both market segments. Now, their ATI purchase is starting to look more and more beneficial to them. Indeed, this may be the thing that saves their company from ruin.

With any luck, future success in ATI will give AMD the resources needed to improve their CPU segment to be competable once again with Intel.

Any thoughts? these are mine, because I am finally starting to feel something besides despair for AMD. (It might even be hope!)

 

[Andrew1990]

I dont like predicting anything with the future CPUs because I remember the says where A64 was king and everyone thought Intel was going down. I learned my lesson from that time.

I am sure AMD will take the performance crown again though.

 

[Cogman]

Originally posted by: Andrew1990
I dont like predicting anything with the future CPUs because I remember the says where A64 was king and everyone thought Intel was going down. I learned my lesson from that time.

I am sure AMD will take the performance crown again though.

Aye, hopefully AMD learned its lessons from those days, they sat for too long on the X2 with little R&D (anyways, thats what it looked like on the outside). It hurt them, A lot, and they are still suffering from not dumping more money into R&D. Hopefully though, they wont forever be the lower-level CPU of choice (though they aren't even that so much now)

 

[Idontcare]

Nothing has changed with AMD to negate the truth regarding the necessity of doing four things while restricted to the resources to really only do one per the comments of David Kirk (Nvidia chief scientist) in his interview back in May:

?AMD has been declining because it hasn?t built a competitive graphics architecture for almost two years now?ever since the AMD/ATI merger. They?ve been pulling engineers [from the GPU teams] to Fusion, which integrates GPU technology onto the CPU. They have to do four things to survive, but I don?t think they have enough money to do one thing.

?The first thing they have to do to compete with Intel is the process technology ? they have to build the new fabs. The second thing is the next-generation CPU technology. The third one is the next generation GPU technology?we?re going to invest one billion dollars in here this year and they need to invest on the same level to keep up with us. And then the fourth thing is they say the future is going to be this integrated CPU/GPU thing called Fusion, which there?s no evidence to suggest this is true but they just said it. They believe it and they?re now doing it.

?So they have to do these four multi-billion dollar projects, they?re currently losing half a billion dollars per quarter and they owe eight billion dollars. Their market cap is about three billion, so it?s hard to see where the future is in that picture. Really speaking, they?re going to have to pull not one, but several rabbits out of the hat.?

http://www.bit-tech.net/hardwa...david-kirk-interview/6

 

[OCGuy]

Originally posted by: Andrew1990
I am sure AMD will take the performance crown again though.

Unfortunately gut feelings dont mean anything in the real world. This isnt a sports competition, where the underdog can pull out a last second shot to win the game. Intel is a R&D behemoth, with the $$$ and the fab to back it up. It would take a monumental screw-up for anything to change the way things are right now. AMD is still trying to catch Conroe, let alone Wolfdale/Yorkie. And with Nehalem already benched, that sure isnt going to be a flop of any sort.

The way I see Intel losing my business in the next 2 years, is screwing with the enthusiasts. I dont mind that they charge $1000-$1500 for an unlocked multiplier, because I can beat a stock EE with a $209 Q6600. But if they start forcing MB manufactures to help them lock the speeds, you can bet my ass will be running a AMD chip, and Intel wouldnt see another dime of mine.

 

[thilanliyan]

Originally posted by: Idontcare
Nothing has changed with AMD to negate the truth regarding the necessity of doing four things while restricted to the resources to really only do one per the comments of David Kirk (Nvidia chief scientist) in his interview back in May:

If they can hold out long enough to reap some of their new GPU release success (which it should be considering the price/performance that they have), I'm sure they'll be in better financial standing by 4th quarter of this year.

ATI have already stated that their GPU design is going in a different direction to nVidia's. While nVidia will develop huge monolithic GPUs (a la GTX 280 with 1.4billion transistors), ATI will develop smaller GPUs and put several of them on a card to get better performance (obviously the drivers still play a huge part) and so they don't necessarily have to invest as much money as nVidia to get the performance to compete (which is contrary to what David Kirk was suggesting). Their recent 4850 release is a good indication that their strategy could work as 2x4850 can cost less than half as much (Bestbuy had some for $150 each while 1 GTX 280 costs $650) and give the same performance.

Of course this is only half the story as AMD still has to compete in the CPU sector. I really don't know how that's gonna pan out, but hopefully they will eventually do a decent job.

 

[Idontcare]

Originally posted by: thilan29
If they can hold out long enough to reap some of their new GPU release success (which it should be considering the price/performance that they have), I'm sure they'll be in better financial standing by 4th quarter of this year.

Which MUST be reinvested into the GPU business unit IF they intend to stay in that race.

Whatever resources are robbed from the ATI side of things to subsidize the CPU side of things is going to further undermine the ATI side of things. This was one of David Kirk's points in his interview.

Originally posted by: thilan29
ATI have already stated that their GPU design is going in a different direction to nVidia's. While nVidia will develop huge monolithic GPUs (a la GTX 280 with 1.4billion transistors), ATI will develop smaller GPUs and put several of them on a card to get better performance (obviously the drivers still play a huge part) and so they don't necessarily have to invest as much money as nVidia to get the performance to compete (which is contrary to what David Kirk was suggesting). Their recent 4850 release is a good indication that their strategy could work as 2x4850 can cost less than half as much (Bestbuy had some for $150 each while 1 GTX 280 costs $650) and give the same performance.

Seriously how does anyone read about the ATI stated strategy and not realize this is exactly what you MUST say is the plan when you are otherwise losing?

All AMD is doing is saying they aren't playing to win, they intend to lose by intelligent design. For the longest time all we heard from AMD was "monolithic native quadcore FTW"...now they are losing to NVidia in that footrace and suddenly its "hooray MCM".

It's all marketing spin at its best. Just as I didn't care whether Intel or AMD were MCM'ing their CPU's I also could not care less whether ATI or NV are MCM'ing their GPUs provided the performance is there.

As for either side claiming their failures are actually part of a larger strategy...um yeah, and I got B's in college even though I could have got A's because I didn't want to be one of those nerds who studied all the time, I'm a B student because I chose to have a life. (right )

 

[Accord99]

Originally posted by: Cogman
I think that most here felt after the initial buying of ATI and a few months past it something like "Crap, AMD is going to take 2 good companies down with it" in fact they had a very rocky 2 years in both market segments. Now, their ATI purchase is starting to look more and more beneficial to them. Indeed, this may be the thing that saves their company from ruin.

It's going to take a lot of video cards and chipsets to pay off $5 billion dollars + interest.

 

[Viditor]

At the end of the day, the "proof is in the pudding"...
The 4850 (and most likely the 4870 much more so) is quite superior to the majority of Nvidia's current line and is priced to move.
Even many hardcore Nvidia fans have been indicating a change...it is somewhat reminiscent of the C2D release in 2006.

IDC...I disagree with David Kirk's spin on the subject.
1. He conveniently ignores the marketshare swing that began with the 38xx last year.
2. He states that engineers work either on Fusion or new GPUs, but the same engineers have been working on both since before the merger (it was one of the reasons for the merger).
3. He ignores that AMD is able to rely on partners like IBM to help develop new processes (greatly reducing the load).
4. He ignores (vehemently I would imagine) the other partnerships that allow for new processes on GPUs...for example, the reason AMD was able to receive such a headstart on TSMC's 55nm process is that they had invested in it with a significant amount of money well beforehand.
5. He ignores the fact that they are no longer losing so much money (they were within a hair's width of break even at Q4).

Accord
"It's going to take a lot of video cards and chipsets to pay off $5 billion dollars + interest"

Or a single settlement with Intel...Wells Fargo has been predicting a settlement of ~$4 billion from Intel since day one. They state that the longer it takes, the higher the settlement will be.

 

[Nathelion]

IDC is spot on. AMDs strategy right now is one of weakness, and the only reason they're doing it is because they have no choice. I think I can confidently say this: Unless nVidia headquarters are suddenly and inexplicably struck by a tornado or all Intel's fabs mysteriously sink into the sea, AMD will never take back the performance crown in either segment, at least not in a remotely foreseeable future. They are in better straits in the GPU part of the business than the CPU part, but they are at best going to equal nVidia's offerings over the next 2-3 years. In the CPU market, they'd have to do darn well to even stay in shouting distance. I think AMD will survive as a company and they will continue to develop mainstream CPUs... but they will never again be anything but a token competitor, kept alive for the benefit of antitrust legislation. If they manage to pull their GPU business out of the morass, it will be either because Intel is putting pressure on nVidia, which will only put them in yet another precarious position in the long run: If Intel trounces nVidida, they're going to trounce AMD too. The other possibility is that they somehow manage to make "transparent crossfire" (which would make one unexpected mega-rabbit indeed) and nVidia fails to come up with anything similar (by transparent crossfire I mean crossfire that is "invisible" to the run-time environment or in some other way avoids significant scaling penalties). I don't see that happening, either.

 

[Idontcare]

Originally posted by: Viditor
IDC...I disagree with David Kirk's spin on the subject.
1. He conveniently ignores the marketshare swing that began with the 38xx last year.
2. He states that engineers work either on Fusion or new GPUs, but the same engineers have been working on both since before the merger (it was one of the reasons for the merger).
3. He ignores that AMD is able to rely on partners like IBM to help develop new processes (greatly reducing the load).
4. He ignores (vehemently I would imagine) the other partnerships that allow for new processes on GPUs...for example, the reason AMD was able to receive such a headstart on TSMC's 55nm process is that they had invested in it with a significant amount of money well beforehand.
5. He ignores the fact that they are no longer losing so much money (they were within a hair's width of break even at Q4).

To be sure what David Kirk was arguing was based on a static snapshot of AMD's resource outlook of the time. As time passes and the resource outlook changes it always has the possibility of invalidating the original assertions. (this goes to Items 1 and 5 in your list)

I'm going to guess what he was thinking in stating Item #2...and that is of course you can task your engineers with >1 project and >1 job description but you do so at the peril of increasing the inefficiency of the engineer.

There was a great Dilbert cartoon strip where the pointy hair boss makes the statement "we are going to focus across the board". The joke being of course that you can't effectively focus on everything, that is the very definition of being not focused.

I'll tell you how I'd shoot holes in David Kirk's assertions (and by association, my own) - his arguments regarding AMD/ATI are directly applicable to Nvidia's own "diversification of resources" from their GPU line-up into their Chipset and low-power CPU productization efforts.

In effect whatever is good for the goose is good for the gander, so if its a bad strategy for AMD then it should likewise be an equally hopeless strategy for Nvidia...and yet they do not seem deterred.

 

[thilanliyan]

Originally posted by: Idontcare
Seriously how does anyone read about the ATI stated strategy and not realize this is exactly what you MUST say is the plan when you are otherwise losing?

All AMD is doing is saying they aren't playing to win, they intend to lose by intelligent design. For the longest time all we heard from AMD was "monolithic native quadcore FTW"...now they are losing to NVidia in that footrace and suddenly its "hooray MCM".

I'm sure AMD would love to completely own the GPU market as NVidia did for about a year and a half while making monolithic GPUs. I agree that they've got to spin everything in a positive light but from current reviews, it seems they've done a decent job of the multiple small GPUs bit.

As I said a pair of 4850 cards can cost less than half as much as a GTX 280 yet perform the same. That's pretty damned impressive. And that's not even the fastest single GPU card they're releasing. If they were losing very badly in performance (ie. Phenom vs. Core) then obviously it's complete marketing spin but they're actually not so it's easier to buy into it.

 

[Nemesis 1]

Originally posted by: Viditor

At the end of the day, the "proof is in the pudding"...
The 4850 (and most likely the 4870 much more so) is quite superior to the majority of Nvidia's current line and is priced to move.
Even many hardcore Nvidia fans have been indicating a change...it is somewhat reminiscent of the C2D release in 2006.

IDC...I disagree with David Kirk's spin on the subject.
1. He conveniently ignores the marketshare swing that began with the 38xx last year.
2. He states that engineers work either on Fusion or new GPUs, but the same engineers have been working on both since before the merger (it was one of the reasons for the merger).
3. He ignores that AMD is able to rely on partners like IBM to help develop new processes (greatly reducing the load).
4. He ignores (vehemently I would imagine) the other partnerships that allow for new processes on GPUs...for example, the reason AMD was able to receive such a headstart on TSMC's 55nm process is that they had invested in it with a significant amount of money well beforehand.
5. He ignores the fact that they are no longer losing so much money (they were within a hair's width of break even at Q4).

Accord
"It's going to take a lot of video cards and chipsets to pay off $5 billion dollars + interest"

Or a single settlement with Intel...Wells Fargo has been predicting a settlement of ~$4 billion from Intel since day one. They state that the longer it takes, the higher the settlement will be.

I hope your sitting down . I kinda agree with ya! But If you really look at this viditor with the future in mind. If you don't think Intel is worried about cuda your way wrong.

Now taking this into consideration. If I was AMD I would turn my complete attention to developing a GPCPU. And thats exactly what there doing . But AMD should fab their own GPUs . I was avtually excited that they are fabing fusion. THese little SPs that ATI developed are alot more important than many realize for both Intels and AMDs future.

Now on the lawsuite . I agree their should be closure soon . But not the way you think .

I believe maybe 1billion and another crosslicense agreement concerning compilers and SPs.

 

[Viditor]

Originally posted by: Nemesis 1


Now on the lawsuite . I agree their should be closure soon . But not the way you think .

I believe maybe 1billion and another crosslicense agreement concerning compilers and SPs.

I hope you're sitting down as well...because I believe that you may be partially correct here as well.
While I don't think AMD actually needs or wants another cross-license agreement, there is something they could use and would be beneficial to both companies for a transfer.
Instead of all cash, I can see Intel negotiating a deal (keep in mind that this is absolute speculation) whereby AMD receives equal ownership of all x86 patents.
1. This would drastically reduce the amount of cash Intel would have to pay
2. AMD would no longer have to pay any licensing fees at all
3. The 2 companies could collect and enforce all x86 fees from 3rd parties (sharing the cost of enforcement) providing a new revenue stream for AMD
4. It would reduce bureaucratic requirements for both companies, thereby increasing efficiency (there are currently a large number of people on both sides keeping track of licensing numbers).

 

[nerp]

An NVIDIA guy commenting about the competition proclaims that the competition is floundering. Surprise!

 

[YANIV]

and I got B's in college even though I could have got A's because I didn't want to be one of those nerds who studied all the time, I'm a B student because I chose to have a life. (right )

if u look at it. that statement could be true... its a question of choice.. (neo) hahaha

 

[pmv]

I'm sure its merely a superficial similarity and isn't technically the same thing at all, but I can't help finding the ATI line about putting multiple GPU's on a card instead of going with one big monster GPU to be worryingly reminiscent of 3dfx's strategy in its dying days ('we don't need monster gpus with t&l, SLI on a single card is the best way to increase performance'). That didn't turn out too well, did it?

 

[thilanliyan]

3DFx lost out more because they decided to use Glide instead of D3D didn't they?

 

[bryanW1995]

Originally posted by: Idontcare

Originally posted by: thilan29
If they can hold out long enough to reap some of their new GPU release success (which it should be considering the price/performance that they have), I'm sure they'll be in better financial standing by 4th quarter of this year.

Which MUST be reinvested into the GPU business unit IF they intend to stay in that race.

Whatever resources are robbed from the ATI side of things to subsidize the CPU side of things is going to further undermine the ATI side of things. This was one of David Kirk's points in his interview.

Originally posted by: thilan29
ATI have already stated that their GPU design is going in a different direction to nVidia's. While nVidia will develop huge monolithic GPUs (a la GTX 280 with 1.4billion transistors), ATI will develop smaller GPUs and put several of them on a card to get better performance (obviously the drivers still play a huge part) and so they don't necessarily have to invest as much money as nVidia to get the performance to compete (which is contrary to what David Kirk was suggesting). Their recent 4850 release is a good indication that their strategy could work as 2x4850 can cost less than half as much (Bestbuy had some for $150 each while 1 GTX 280 costs $650) and give the same performance.

Seriously how does anyone read about the ATI stated strategy and not realize this is exactly what you MUST say is the plan when you are otherwise losing?

All AMD is doing is saying they aren't playing to win, they intend to lose by intelligent design. For the longest time all we heard from AMD was "monolithic native quadcore FTW"...now they are losing to NVidia in that footrace and suddenly its "hooray MCM".

It's all marketing spin at its best. Just as I didn't care whether Intel or AMD were MCM'ing their CPU's I also could not care less whether ATI or NV are MCM'ing their GPUs provided the performance is there.

As for either side claiming their failures are actually part of a larger strategy...um yeah, and I got B's in college even though I could have got A's because I didn't want to be one of those nerds who studied all the time, I'm a B student because I chose to have a life. (right )

well, if the 4850 is any indication, the end will be put off a while longer at least. This is the same sort of feeding frenzy that we saw when 8800gt was released, but amd has more experience with their 55nm process technology than nvidia had at 65nm at the time, meaning that amd probably has much better yields and can make just as much $$$ at a lower asp. Also, when bfg10k starts making indications that he's leaning towards a 48xx gpu then daamit has to be doing something right.

 

[bryanW1995]

Originally posted by: Nathelion
IDC is spot on. AMDs strategy right now is one of weakness, and the only reason they're doing it is because they have no choice. I think I can confidently say this: Unless nVidia headquarters are suddenly and inexplicably struck by a tornado or all Intel's fabs mysteriously sink into the sea, AMD will never take back the performance crown in either segment, at least not in a remotely foreseeable future. They are in better straits in the GPU part of the business than the CPU part, but they are at best going to equal nVidia's offerings over the next 2-3 years. In the CPU market, they'd have to do darn well to even stay in shouting distance. I think AMD will survive as a company and they will continue to develop mainstream CPUs... but they will never again be anything but a token competitor, kept alive for the benefit of antitrust legislation. If they manage to pull their GPU business out of the morass, it will be either because Intel is putting pressure on nVidia, which will only put them in yet another precarious position in the long run: If Intel trounces nVidida, they're going to trounce AMD too. The other possibility is that they somehow manage to make "transparent crossfire" (which would make one unexpected mega-rabbit indeed) and nVidia fails to come up with anything similar (by transparent crossfire I mean crossfire that is "invisible" to the run-time environment or in some other way avoids significant scaling penalties). I don't see that happening, either.

in early august 4870x2 will almost certainly take the gpu performance crown. nvidia won't be able to even contemplate a sandwich card until they get gt200 on 55nm, and even then the thermals would most likely be a deal killer. it's not set in stone by any stretch, but it now appears that daamit has used their shoestring budget to kick the shit out of nvidia this time around. Can they do it again if they don't spend at least a reasonable % of nvidia on the investment side? Almost certainly not, which is why I said that it would just keep the end at bay a little bit longer.

The gpu side is very different than the cpus, however, so this could be interesting going forward. AMD has the process advantage in gpus, while we all know where they stand vs intel in cpus. AMD has a strong ally in gpus in intel because intel has correctly deduced that nvidia is a hell of a lot more dangerous to them right now than amd. How many people have bought p35, x38, and x48 crossfire mobos and run nvidia gpus for the past couple of years? The vast majority, right? Well, what are those people going to do now that they can spend a few hundred $$ and get gtx 280-type performance? Right. This situation will improve even more in favor of amd next year if nvidia doesn't figure out a way to get sli or their own chipsets at least onto nehalem. AMD could also leverage that potential situation to their great advantage in the cpu market by keeping sli-compatible chipsets on their future cpu platforms. They would basically be guaranteed that MOST enthusiasts would use them for something (either gpu if intel or cpu if nvidia fan) and that SOME would use them for both. In spite of hector's cluelessness there is still at least a chance that amd could come through this intact.

 

[bryanW1995]

Originally posted by: Idontcare

Originally posted by: Viditor
IDC...I disagree with David Kirk's spin on the subject.
1. He conveniently ignores the marketshare swing that began with the 38xx last year.
2. He states that engineers work either on Fusion or new GPUs, but the same engineers have been working on both since before the merger (it was one of the reasons for the merger).
3. He ignores that AMD is able to rely on partners like IBM to help develop new processes (greatly reducing the load).
4. He ignores (vehemently I would imagine) the other partnerships that allow for new processes on GPUs...for example, the reason AMD was able to receive such a headstart on TSMC's 55nm process is that they had invested in it with a significant amount of money well beforehand.
5. He ignores the fact that they are no longer losing so much money (they were within a hair's width of break even at Q4).

To be sure what David Kirk was arguing was based on a static snapshot of AMD's resource outlook of the time. As time passes and the resource outlook changes it always has the possibility of invalidating the original assertions. (this goes to Items 1 and 5 in your list)

I'm going to guess what he was thinking in stating Item #2...and that is of course you can task your engineers with >1 project and >1 job description but you do so at the peril of increasing the inefficiency of the engineer.

There was a great Dilbert cartoon strip where the pointy hair boss makes the statement "we are going to focus across the board". The joke being of course that you can't effectively focus on everything, that is the very definition of being not focused.

I'll tell you how I'd shoot holes in David Kirk's assertions (and by association, my own) - his arguments regarding AMD/ATI are directly applicable to Nvidia's own "diversification of resources" from their GPU line-up into their Chipset and low-power CPU productization efforts.

In effect whatever is good for the goose is good for the gander, so if its a bad strategy for AMD then it should likewise be an equally hopeless strategy for Nvidia...and yet they do not seem deterred.

nvidia is not in dire financial straights like amd is. in fact, they have been making record profits for quite a while now, and they're becoming SO big that intel is just about the only team left to go after. Mr Huang, unlike Mr Ruiz, is the real deal and would probably be happy to tackle that mountain even if there weren't numerous legitimate financial reasons to do so.

 

[Acanthus]

Originally posted by: thilan29
will develop smaller GPUs and put several of them on a card to get better performance

fixed

 

[scoobyx13x]

Originally posted by: Ocguy31

The way I see Intel losing my business in the next 2 years, is screwing with the enthusiasts. I dont mind that they charge $1000-$1500 for an unlocked multiplier, because I can beat a stock EE with a $209 Q6600. But if they start forcing MB manufactures to help them lock the speeds, you can bet my ass will be running a AMD chip, and Intel wouldnt see another dime of mine.

This is a very real possibility. Just three years ago, Intel was a very closed door company that only grudgingly acknowledged the enthusiast market. Now, of course people will scoff at the hardware geeks, but lets face it - who else is buying these $1k processors? It sure as hell ain't business customers or grandmas looking to check their emails. By the same token, if Intel abandons the lower end in favor of pursing high end sales they will be lost as well. It is a fine line to walk.

Intel has the lead, no doubt, and even though no current AMD chip can match them, if they start jerking around the marketplace or treating the enthusiasts like a bunch of red headed stepchildren again, they will have the fastest chip that noone wants to buy.

Now as to ATI vs Nv, you can see to very different approaches to the marketplace. Just look at the past week for proof. Nv comes out with the fastest single card, a huge $650 monster. ATI comes out with a smaller, not the fastest but still very impressive card for $200. ATI has created quite a stir because in meaningful tests the 4850 is an amazing card that is 70% of the cost of the R280.

What makes this such a beautiful strategy is that the two products can exist in the marketplace. Those that want the absolute fastest card in the world can buy a $650 R280. The rest of us who would come home to divorce papers if they spent that much on a video card, can buy a good enough to get the job done 48xx for $200

 

[gregt29]

I don't see amd doing as bad as everyone else, at least not on the
gpu side of things. This was set in motion with the incredible price to
preformance ratio 38xx series cards. We know those were offset by
the 8800gt, but once you rap your heads around the x2 48xx cards
your start to see that amd is really close to nvidia if priced right.
And although intel is killing them by simply dropping the price of
old better performing chips, I can see a lot more amd chipsets
being purchased to run crossfire on the new gpus.

 

[thilanliyan]

Originally posted by: Acanthus

Originally posted by: thilan29
will develop smaller GPUs and put several of them on a card to get better performance

fixed

Lol, I know it was done before. I actually have a V5 5500.

However, for the moment 4850 Crossfire seems to be doing fairly well, which is basically what a 4850X2 / 4870X2 will be so it's not that bad. In fact there's rumours that the X2 cards will communicate in a different way to how the 3870X2 did making them faster than a regular crossfire setup. If there's truth to that then I'm all for it and would rather pay $400 as opposed to $650 for the same performance.