I am sure Dave B will disagree with me on this one...

[jb]

How much do you think that was due to shortages in manufacturing at ATI. And if it was due to it, and it seems the R520 may have even worse yields than the R420.....

I think that plays a large part. The other part thats pretty big is back then the x800 was usally behind the 6800gt is most benchmarks. So you had a card that you could hardly get vrs a card tht was easier to get that had better perfromace. It made for an easy choice.

 

[ondaedg]

How much do you think that was due to shortages in manufacturing at ATI. And if it was due to it, and it seems the R520 may have even worse yields than the R420.....

While shortages did effect ATI's sales, it was primarily the 6800 GT's price and performance that made it sell well, and the fact that ATI didn't really have anything to compete until the X800 XL which came many months later. Same applies to the 6600 GT though less so since 9800 Pros are readily available for cheap and have similar performance. But still, people want newer products.

I agree that shortages to ATI's high end chips played a big part in Nvidia's recent success. If they could have produced the x800xt in quantity at all, the price would have been similar to the 6800 gt and ATI would have enjoyed two good quarters. It is also quite possible that the NV40 may have caught ATI off guard. If you think about it, Nvidia talked about the nv30 like it was going to use that architecture for quite some time. ATI probably thought they were going to be playing catch-up for at least one more generation. They very well could have planned to produce the 16 pipe cards in low quantity as an "ultra" version right from the start. Nvidia brings a 6800gt out of nowhere and suddenly ATI has to make their 16 pipe card in much higher quantities than they initially planned.

 

[ANova]

The 6800 GT definitely caught ATI off guard. I don't think anyone was expecting a $400 card that performed almost as good as the $500 high end Ultra.

 

[Randell]

yeah remember early benches and reviews x800Pro stomped all other the 6800 and the X800XT PE just edged the Ultra, then along comes the 6800GT and now I have one

 

[JoshMST]

You know, I haven't really heard much since the release of the article. All sources have pretty much clammed up, or they haven't received any new information. I think you all know about as much as I do. I haven't heard any rumors of recent TSMC 90 nm orders from ATI, so I am seriously wondering when availability of R520 will hit. Then again, I don't have any contacts at TSMC that I could get solid info from, so that is just mere speculation on my part based on other people's input.

And yes, 6800 GT was a great seller for NVIDIA. In hindsight it certainly seems that NVIDIA overcame its problems with IBM far sooner than ATI did with TSMC. Then again, was it so much a yield problem with ATI as compared to a simple supply problem with 130 nm low-k? IBM had plenty of fab space for 130 nm FSG as compared to tSMC with 130 l-k.

I am very curious to see how July pans out for ATI in terms of Crossfire shipments and the R520. Should be an exciting month!

 

[Geo]

I think the success of x800xl supports what my own informal estimation also is --enthusiasts are not real comfortable (and it is probably snobbery as much as anything) differentiating at the top end by pipes (i.e. X800 Pro). They want all their pipes guaranteed to work, and then maybe have a go at oc'ing as their "value add", particularly for the lower-clocked "high-end".

Turning on pipes and hoping for the best is more of a mainstream-y thing, and more risky, and thus when you are willing to have a go there for price/performance reasons and like the risk, well okay, but fewer people are willing to take that risk when they feel they are dealing with the top-end (i.e. X800) $350+ parts.

 

[trinibwoy]

I think the success of x800xl supports what my own informal estimation also is --enthusiasts are not real comfortable (and it is probably snobbery as much as anything) differentiating at the top end by pipes (i.e. X800 Pro). They want all their pipes guaranteed to work, and then maybe have a go at oc'ing as their "value add", particularly for the lower-clocked "high-end".

What you're saying is true to some extent but I would bet that the lowball price had a lot more to do with XL sales than the number of pipes. If the X800PRO/X850PRO was $300 it would have sold like hotcakes as well.

 

[Geo]

I think the success of x800xl supports what my own informal estimation also is --enthusiasts are not real comfortable (and it is probably snobbery as much as anything) differentiating at the top end by pipes (i.e. X800 Pro). They want all their pipes guaranteed to work, and then maybe have a go at oc'ing as their "value add", particularly for the lower-clocked "high-end".

What you're saying is true to some extent but I would bet that the lowball price had a lot more to do with XL sales than the number of pipes. If the X800PRO/X850PRO was $300 it would have sold like hotcakes as well.

I think we're saying pretty close to the same thing. . .as the price slides down people start liking more the possibility that they might get a "free" bump from turning on pipes (assuming it works out). As the price goes up they start seeing the risk more than the reward.

 

[Charmaka]

The X800XL has been very successful because it provides easily the best performance per currency unit for high-end cards, and as such is a no-brainer for the majority of people interested in X800+/6800+ cards.

 

[kemosabe]

I haven't heard any rumors of recent TSMC 90 nm orders from ATI, so I am seriously wondering when availability of R520 will hit. Then again, I don't have any contacts at TSMC that I could get solid info from, so that is just mere speculation on my part based on other people's input.

R520 production was reportedly underway at TSMC, although it's hard to tell whether they're referring to the (allegedly) poorly yielding revision or the presumed respin. Anyone who knows is welcome to elaborate. :?

 

[karlotta]

How much do you think that was due to shortages in manufacturing at ATI. And if it was due to it, and it seems the R520 may have even worse yields than the R420.....

Well low yields can only be measured from a release date, so the 520 hasnt had low yields yet. The R420 was made at the 130nmlowK fab, and there is only one .The 520 is being made 90nmLowK and they have 3 fabs. On that alone there should be many more R520 chips. OEM presure and PCIe production is more to the "shotage" of R420AGP parts than "yield" issues.

 

[ondaedg]

I honestly hope they do not have low yields. Two readily available high performing parts are necessary for the consumer to win.