ART of War, nV vs. ATi your thoughts

[_xxx_]

If the r520 came out and the g70 and r520 got the reviews both did, would that have changed the course of the r580 being released as it was? It wouldn't have thats the point.

Well, in the first place it would have made lots of money. The reviews would have seen it as a clear leader over simultaneously released competitor's product. Much more hype & glory, good sales, ATI happy. Also, it would have set a much better stage for the following products, then R580 would have been announced as a "super-shader-power" refresh with the promise of the following product wiping the floor with the competition even thorougher in "good tradition of our past products".

All these marketing advantages/arguments went down the drain thanks to the delay ONLY.

 

[Razor1]

the r580 came out January 24th, hard launched, 2 months to build up supply, puts it in Novemeber, December, There might have been 1 month delay in there, but thats about it. Its pretty much on schedule.

 

[_xxx_]

No, R580 was scheduled for the fall refresh.

Was it pushed because of not being finished on time or in order not to kill R520 sales/margins?

 

[Razor1]

Well, in the first place it would have made lots of money. The reviews would have seen it as a clear leader over simultaneously released competitor's product. Much more hype & glory, good sales, ATI happy. Also, it would have set a much better stage for the following products, then R580 would have been announced as a "super-shader-power" refresh with the promise of the following product wiping the floor with the competition even thorougher in "good tradition of our past products".

All these marketing advantages/arguments went down the drain thanks to the delay ONLY.

They could still do the hype , I don't see much of ATi's marketing machine anywhere. Not that nV is doing much either but they don't need to. Specially the Fear Benchmarks, I'm sure ATi could do some favorable things for websites, to do individual benchmarks to show thier cards as they should be.

 

[_xxx_]

They could still do the hype

Sure they can try, but the image is severely damaged and that's a huge obstacle for the wide audience to believe it as well

 

[Razor1]

Sure they can try, but the image is severely damaged and that's a huge obstacle for the wide audience to believe it as well

True

 

[Dave Baumann]

the r580 came out January 24th, hard launched, 2 months to build up supply, puts it in Novemeber, December, There might have been 1 month delay in there, but thats about it. Its pretty much on schedule.

So, you know the scheule do you?

Was it pushed because of not being finished on time or in order not to kill R520 sales/margins?

I know that by June R580 has already taped out, even though they were still working through R520 issue. It did, however, come back with the same issue as R520. I've heard they did have th option of just scrapping R520 and continuing to chase down the issue on R580, but I guess continued with R520 because a lot of work had already gone into it and any other, more normal, bringup issues had probably been resolved by that point (something that may not have been the case with R580). Once they had resolved the issue with R520 they obviously then went back subseqently changing each other part that had the issue, hence the staggered release (R520 XT, RV530 a month after that, then the hard launch of R580 in the new year) - this is why you see R580's release silicon was A22, with a base layer change (normally you aim for A12 on refresh stuff).

 

[Razor1]

So, you know the scheule do you?

Ya said fall right? Well that put it right at the end of Fall , as you explained the r580 could have came out instead of the r520, which would have given thier marketing much more appeal. It was a choice that ATi took, looking back at it it was not that great of a choice. It would have been better to have the r580 out first, then use the r520 derivitives as lesser parts, and continue along those lines.

 

[_xxx_]

Well the decision might have been a good one as well. I'm one of the people who bought X1800XT just because it offers good performance for the buck now and has some (however minor) IQ advantages over the 7800GTX and I guess I'm not alone. I would never have bought X1900XT(X) just because of the price, I'd have bought the 7900GT instead.

In the end, it still pays off in a way - better to recoup some R&D money than to throw away your R&D investments and thus lose much more. And the manufacturing probably got cheaper by now as well.

 

[Razor1]

Well the decision might have been a good one as well. I'm one of the people who bought X1800XT just because it offers good performance for the buck now and has some (however minor) IQ advantages over the 7800GTX and I guess I'm not alone. I would never have bought X1900XT(X) just because of the price, I'd have bought the 7900GT instead.

In the end, it still pays off in a way - better to recoup some R&D money than to throw away your R&D investments and thus lose much more. And the manufacturing probably got cheaper by now as well.

That is true, but this happened more then once with ATi in the previous generation chips.

the x800 gto's and gto2's were sent out the same way, at least its not a complete right off but still, having excessive inventory, is also one of the key basic principles in business. Over extending resources where its less fruitful.

 

[_xxx_]

Oh well, call it damage control

 

[maaoouud]

Well I doubt they make a loss selling X800GTO/GTO^2. If they could make money on 9500 then they'll make money on the GTO. Sure manufacturing those cores took part of their production but what else were they going to make ? - It would not have been wise to massproduce the R520 when they did not know the solution to the problem in it.

 

[dizietsma]

It's probably difficult to judge whether ATi should have gone straight for r580 or not unless you know how many r520's they sold and how much money that generated compared to r580 straight away.

I agree with the above point though that the r520 would have looked great if it had come out before the G70 and then the G70 would have looked if it had only just managed parity when it was released, rather than it being the other way around of course. Some of the gloss was lost, maybe r580 straight away would have got that gloss back quicker though.

I'm not sure whether the 7900GT at reduced price will be a better bet than the rv570 but the rv570 looks like it will be an interesting gpu at least so I am looking forward to it's release.

There's too many sectors to the market now for my liking but I can see why that is happening. I suppose the variety of choice outweighs the confusion of options.

 

[_xxx_]

I think normal users are totally confused with all the names out there now, from both companies.

 

[maaoouud]

Not just normal users. I spend a good amount of time on keeping up to date with what is released and I find myself forced to look up the specs of cards all the time. I mean it's not just that there's a lot of different gpus out there, now there's brands that release different models of the same chip/card. XFX has 5 7900GTX and 5 7900GT models, that is just mindboggling - How are they going to price them ? I think they must be trying to trick distributors to stock more models so that they get to ship more cards than they normally would have. One time boost for sales but at the price of angered distributors and confused customers.

Sorry about going slightly off topic but I think that since ATI is now allowing companies to overclock the cards it'll become the same mess with their offerings.

 

[Razor1]

Well I doubt they make a loss selling X800GTO/GTO^2. If they could make money on 9500 then they'll make money on the GTO. Sure manufacturing those cores took part of their production but what else were they going to make ? - It would not have been wise to massproduce the R520 when they did not know the solution to the problem in it.

Its not a loss in the sense they will still be making some money, but in the sense they will not be making as much money as they can it is a loss. Also this leaves very little margins also, if nV decides to cut prices ATi will have a tough time to keep up thus lossing market share. nV is expecting to gain up to 60% of discrete card sales this year. That is quite high for have competitive products.
http://us.rd.yahoo.com/finance/external/reuters/SIG=11vg30000/*http://yahoo.reuters.com/financeQuoteCompanyNewsArticle.jhtml?duid=mtfh14958_2006-03-16_02-59-54_n15213119_newsml

Interesting they are talking about 65 micron process for this year also.

 

[maaoouud]

Its not a loss in the sense they will still be making some money, but in the sense they will not be making as much money as they can it is a loss. Also this leaves very little margins also, if nV decides to cut prices ATi will have a tough time to keep up thus lossing market share. nV is expecting to gain up to 60% of discrete card sales this year. That is quite high for have competitive products.

First, they might not have the optimal design for generating cash with the excess X800GTO/GTO^2 dies because of the diesize, but this isn't as bad as it seems at first. They will probably be able to offload them easy enough given their performance and they gain goodwill from all the cheap tweakers that buy them and unlock/overclock them.
A more efficient design (smaller die) could probably have generated more profit for them. I suspect that they were simply too busy to be able to get a completely new mainstream performance design out earlier last year, they did score the new xbox design etc. Then they might not have thought the manufacturingprocess available would have been optimal for such a design - Why create a new design for a process and then a couple of months later (or whatever) there's a better process available that will lower the manufacturing costs significantly?
It costs alot of money to get a design ready for manufacturing on the specific process, would they have been able to sell enough cores from the old process to pay for the cost of devellopment ? Would they have been able to spare the engineering resources (do they even have the manpower to do that at that particular time) without having it interfere with other designs timetables?
- There's a ton of questions that could be asked and discussed and alot of them may not even have answers. The companies do not always sit back and consider all the aspects of what they do. They've got some pretty hard planning to do, you don't want to miss the window of a particular process as that can be quite costly (as you argue). But there are no guarantees that everything will run smooth either, the process may not live up to what was expected and so on. And in this particular case something that should be safe to assume is bugtested and flawless (or atleast the parameters are known well enough), a 3rd party library, apparently had some defects (or maybe unknown characteristic is a better way to put it) and the engineers ran into trouble. These eventualities is something that I don't think it would be economical to allocate resources into preventing. ATI/Nvidia buy these libaries just so that they do not have to spend the resources it takes to come up with them, the results of such research may not even be favourable for their particular applications (they want to make chips, not devellop process technologies or optimized lowlevel circuits that may only be valid for one process).


Second, if Nvidia decides to cut prices that does not automatically mean that ATI will lose marketshare. ATI can probably cut prices and stay compeditive here (X800GTO/GTO^2), or maybe they don't even have to, they only need to sell the cores they have. Sure lowering prices too much might very well bring indirect consequences such as lower average sales prices which in turn may cause a stock analyst to lower their rating of the company, but you cannot run a company based on stock analysts oppinions anyway.


Third, Michael Hara did not say that they were expecting to gain up to 60% marketshare in discrete card sales. The quote was "Our objective is to drive towards 60 percent," "We will try to make that this year.". That basically means that they'd like to have 60% share of something. And this is just your general financial peptalk that they feed investors and analysts to make them sound aggressive and give them an impression that they are compeditive etc. etc. without actually promising anything.

Don't make the common mistake to think that company B will go out of business just because company A have some percieved advantage for a period of time.

 

[Razor1]

First, they might not have the optimal design for generating cash with the excess X800GTO/GTO^2 dies because of the diesize, but this isn't as bad as it seems at first. They will probably be able to offload them easy enough given their performance and they gain goodwill from all the cheap tweakers that buy them and unlock/overclock them.

Yes those were good performers for the price they were going at, but the x1800 gto isn't, its slated for $250 and just about equal to a 7600 gt which is for a stock version at $50 less.

A more efficient design (smaller die) could probably have generated more profit for them. I suspect that they were simply too busy to be able to get a completely new mainstream performance design out earlier last year, they did score the new xbox design etc. Then they might not have thought the manufacturingprocess available would have been optimal for such a design - Why create a new design for a process and then a couple of months later (or whatever) there's a better process available that will lower the manufacturing costs significantly? It costs alot of money to get a design ready for manufacturing on the specific process, would they have been able to sell enough cores from the old process to pay for the cost of devellopment ? Would they have been able to spare the engineering resources (do they even have the manpower to do that at that particular time) without having it interfere with other designs timetables?

This is where ATi has to focus and get on the ball, otherwise nV's 60% will become reality. The rv570 is going to be out June, Julyish, thats going to be a bit late, either ATi will immidiately cut prices or they have to have a part that is conciderably a higher performer when that chip comes out (they might be able to cut prices if they use 80nm process)

- There's a ton of questions that could be asked and discussed and alot of them may not even have answers. The companies do not always sit back and consider all the aspects of what they do. They've got some pretty hard planning to do, you don't want to miss the window of a particular process as that can be quite costly (as you argue). But there are no guarantees that everything will run smooth either, the process may not live up to what was expected and so on. And in this particular case something that should be safe to assume is bugtested and flawless (or atleast the parameters are known well enough), a 3rd party library, apparently had some defects (or maybe unknown characteristic is a better way to put it) and the engineers ran into trouble. These eventualities is something that I don't think it would be economical to allocate resources into preventing. ATI/Nvidia buy these libaries just so that they do not have to spend the resources it takes to come up with them, the results of such research may not even be favourable for their particular applications (they want to make chips, not devellop process technologies or optimized lowlevel circuits that may only be valid for one process).

That is true, but this is also taken in to concideration in planning to some degree, unfortunatly for the r520, it took much longer then probably expected to find/fix the problem.

Second, if Nvidia decides to cut prices that does not automatically mean that ATI will lose marketshare. ATI can probably cut prices and stay compeditive here (X800GTO/GTO^2), or maybe they don't even have to, they only need to sell the cores they have. Sure lowering prices too much might very well bring indirect consequences such as lower average sales prices which in turn may cause a stock analyst to lower their rating of the company, but you cannot run a company based on stock analysts oppinions anyway.

nV is in a position and a very strong position to cut prices. There is no immidate danger form ATi in this area, unlike before. this goes to what Intel did to AMD to prevent AMD to gain marketshare. It will also work in these circumstances.

Third, Michael Hara did not say that they were expecting to gain up to 60% marketshare in discrete card sales. The quote was "Our objective is to drive towards 60 percent," "We will try to make that this year.". That basically means that they'd like to have 60% share of something. And this is just your general financial peptalk that they feed investors and analysts to make them sound aggressive and give them an impression that they are compeditive etc. etc. without actually promising anything.

It could be a pep talk, but thier goal of 60% is reverberated on how they are planning and marketing thier chips. Everything they do points to this goal, and only this goal.

Don't make the common mistake to think that company B will go out of business just because company A have some percieved advantage for a period of time

Hmm ATi won't go out of business, but time and time again, they are showing thier inability to shift according to market conditions that nV is setting. A company like ATi should also be able to set the tone in a market place not follow. This has been a thorn on thier side for along time.

 

[maaoouud]

I was under the impression that the X1800GTO was created to offload old R520 dies and maybe R580 dies that didn't perform well enough. As I saw it, it's a shortlived product to enable them to sell cores that otherwise would probably have ended up on a landfill.

You're still coming across as predicting gloom and doom for ATI simply because you think that nvidia is in a better position. Look back at what happend during the good old Geforce FX vs. R300 days. ATI had a superior product and at least to me the difference between their respective offerings at that time seemed to be far bigger than what it is now. Nvidia did not go into a financial crisis or had their marketshare collapse, but they did earn less than they were expecting. I'd say that it did spur them to redouble their efforts at creating a compeditive design for the future. Maybe this will spur ATI to pay more attention to having even more appealing products in all the pricebrackets in the future, but they are far from outclassed right now and should still be able to sell their products.

I will agree that ATI sometimes seem to make what seems to us as poor decisions, but they are far from alone at doing this - Nvidia is neither omnipotent nor make all the right choices. But since they both cannot know exactly what the future will be like and the time to market of their products is so long these situations will carry on in the future. What you have to realize is that "shifting according to market conditions" is not always a valid way of thinking about it. They have their designs and can create a limited amount of offerings for the public with them, and they need be able to make money off of those offerings aswell. Sometimes they cannot create a superior product and make money on it, but there is little they can do about it since develloping a new design takes such a long time. Nor is it likely that they have (or will have) the resources to devellop several different designs to be able to guarantee that they will have a superior product in evey bracket at all times. - Compromises must be made and they may be far for optimal at times (but that goes for both companies).

 

[Razor1]

I was under the impression that the X1800GTO was created to offload old R520 dies and maybe R580 dies that didn't perform well enough. As I saw it, it's a shortlived product to enable them to sell cores that otherwise would probably have ended up on a landfill.

Xbit labs has reported in an interview with ATi the x1800 chip is still in production and is not slated to be dumped

You're still coming across as predicting gloom and doom for ATI simply because you think that nvidia is in a better position. Look back at what happend during the good old Geforce FX vs. R300 days. ATI had a superior product and at least to me the difference between their respective offerings at that time seemed to be far bigger than what it is now. Nvidia did not go into a financial crisis or had their marketshare collapse, but they did earn less than they were expecting. I'd say that it did spur them to redouble their efforts at creating a compeditive design for the future. Maybe this will spur ATI to pay more attention to having even more appealing products in all the pricebrackets in the future, but they are far from outclassed right now and should still be able to sell their products.

This was quite a different situation, and nV's marketshare sure did collapse, ATi at the time had very little market share, nV on the other hand had around 70% of market share, ATi did take a good chuck out of nV but it took time, right before the 6800/x800 series were out, ATi had 65% + market share. It was almost a roll reversal. It took ATi 3 years to do this. Now nV is coming back strong both companies have strong products, but nV is chipping away fast.