another Dave Orton interview

[Megadrive1988]

http://www.tomshardware.com/hardnews/20040507_060001.html

THG: Since you joined ATI in 2000, the company has made a remarkable transition. From a position clearly behind Nvidia in terms of technology and market shares, ATI has come back strong. From the view of your product line and your balance sheet, ATI today looks a lot like Nvidia - a reason for your success?

DO: We lost our technology leadership position in 1999 and 2000. That is a fact. It is clear that Nvidia got ahead of us with the TNT2, GeForce 1 and GeForce 2. One of the big changes we made was to focus on driving performance and technology leadership at all cost. We believe that everything starts with product and technology leadership. We opened up all constraints for the design teams and went to win in the high end PC market. Other things then fell into place. I do not think that we have become a lot like Nvidia. I believe that Nvidia has become a lot like us.

quote:
"But if we look right after the X800, the next 18 months probably will show more innovation than the past 18 months. That is a fair statement." <-------- most likely meaning R500...er, or R520

 

[Joe DeFuria]

I find the bit about memory technology the most interesting.

It certainly sounds like ATI is looking at doing something more than just using faster DDR memory for the future.

Even though ATI is going with EDRAM on the X-Box2 VPU, I have doubts about Embedded DRAM for PC chips at this point in time, but that's thinking of it from a pure "enough EDRAM for a decen framebuffer on-chip" perspective.

If ATI have devised a clever cacheing / segmentation scheme...could be interesting.

 

[AlphaWolf]

Well I would hope for more innovation in the next 18 months.

R300 was released 20 months ago.

 

[AlNom]

THG: Nvidia's 6800 has impressed large parts of the industry as well as media. Were you surprised by the performance leap of the product?
DO: We were expecting Nvidia to push ahead. But I would say that there were surprises in some areas. The die size is bigger than we thought and the power consumption is higher than we expected. They also followed up architecturally to 16 pipes. We expected them to end up expanding internal processing instead. I am also surprised that they have not been able to reach a higher core frequency.

"Where's the rest of it?!"

that's a slap in the face....

 

[martrox]

Well I would hope for more innovation in the next 18 months.

R300 was released 20 months ago.

And it's (R300) derivitives are still, arguably, the best there is......

 

[Joe DeFuria]

"But if we look right after the X800, the next 18 months probably will show more innovation than the past 18 months. That is a fair statement." <-------- most likely meaning R500...er, or R520

Well, that quote is unfortunately completely open ended. I mean, does it infer that "after 18 months have elapsed, you'll see a big innovative change", or does it infer "over the course of the following 18 months, you'll see innovative enhancements occur to the 420 core unlike what you saw from R300-->R350-->R360...which were for the most part speed bumps and minor tweaks."

 

[Megadrive1988]

Well, that quote is unfortunately completely open ended.

yeah, that's why I said 'most likely' rather than 'for certain' 8)

 

[Bjorn]

..unlike what you saw from R300-->R350-->R360...which were for the most part speed bumps and minor tweaks."

I think you could add the R420 on that list. Although with a very large speed bump.

 

[Joe DeFuria]

..unlike what you saw from R300-->R350-->R360...which were for the most part speed bumps and minor tweaks."

I think you could add the R420 on that list. Although with a very large speed bump.

I know what you're saying, but I won't add R420 to that list.

R420 is still a completley new chip, including different process. It's based on the same architecture, but still completely new. On the other hand, R300, 350, 360 --> all more or less the same chip. Just minor tweaks.

 

[Sxotty]

Joe if the 420 doesn't go on there then all the people who said the entire gforce 1-4 series were just tweaks should change their tune

 

[Joe DeFuria]

Joe if the 420 doesn't go on there then all the people who said the entire gforce 1-4 series were just tweaks should change their tune

I'm not sure what your point is?

There is no doubt that R420 is based on the same architecture as the R300. The R420 is on a different process and has boatloads of new transistors, double the pipes, etc. It's a completley new chip. It's not taking the actual R300 chip and "tweaking it a bit."

 

[WaltC]

I very much enjoyed the THG interview for the way in which Orton directly addressed the loaded questions and the biased assertions made by the interviewer. Here are the questions/assertions along with the responses Orton made that I appreciated:

THG:...From the view of your product line and your balance sheet, ATI today looks a lot like Nvidia - a reason for your success?

Orton: I do not think that we have become a lot like Nvidia. I believe that Nvidia has become a lot like us.

THG:Nvidia's 6800 has impressed large parts of the industry as well as media. Were you surprised by the performance leap of the product?

Orton: ...I would say that there were surprises in some areas. The die size is bigger than we thought and the power consumption is higher than we expected. They also followed up architecturally to 16 pipes. We expected them to end up expanding internal processing instead. I am also surprised that they have not been able to reach a higher core frequency.

THG: Nvidia has been left with a sour taste of deals with Microsoft.

Orton: We pursue a different model than Nvidia did with the Xbox 1. We did not go into the business thinking we can sell chips to Microsoft. We are providing our technology under a royalty model. Microsoft has worked to get the lowest price solution from a chip level. TSMC will be building these chips.

Those were the main ones I saw, and I was happy to see Orton unabashedly address the false or misleading assertions buried in the THG questions as they were put to him. Such things deserve responses and I think Orton did a great job in tactfully, but forcefully, setting them straight. (Of course, whether that will ever be absorbed by THG is another matter, but now THG can't say going forward that they didn't know better...)

I will say as well that I was surprised at how abruptly THG ended the interview--no, "We'd like to thank Mr. Orton for taking time out of his busy schedule to answer our questions," or anything...! I looked for 15 seconds for "page 2" before it became clear that the interview had ended. Aside from an obvious lack of manners, I'd also like to know whether anyone at THG actually personally "sat down" with Orton, or whether by "sitting down" with Orton THG meant "sitting down with Orton's email responses to our questions" (just out of curiosity, of course...)

 

[Laa-Yosh]

We are still learning in the channel, especially how to be more effective in Eastern Europe, China and South America.

I wonder what Dave Orton means by being more effective - they already have the majority of the enthusiast market here in Hungary

 

[Quitch]

The interview was rubbish, but the way Dave Orton handled his answers were a credit to the man.

 

[bloodbob]

I know what you're saying, but I won't add R420 to that list.

R420 is still a completley new chip, including different process. It's based on the same architecture, but still completely new. On the other hand, R300, 350, 360 --> all more or less the same chip. Just minor tweaks.

Mkay how about RV350->RV360->R420?
I guess 4x number of pipes is a big speed jump but apart from that their I dunno how much of a speed bumb their is.

I dunno how Mr Orton can poke the finger at NV's clock speeds since ATI couldn't manage to build a chip with PS3.0 on .13

 

[AlphaWolf]

I know what you're saying, but I won't add R420 to that list.

R420 is still a completley new chip, including different process. It's based on the same architecture, but still completely new. On the other hand, R300, 350, 360 --> all more or less the same chip. Just minor tweaks.

Mkay how about RV350->RV360->R420?

Well then rv350 would be a different process.

Only the rv360 is .13 low-k

 

[Bjorn]

Orton: ...I would say that there were surprises in some areas. The die size is bigger than we thought and the power consumption is higher than we expected. They also followed up architecturally to 16 pipes. We expected them to end up expanding internal processing instead. I am also surprised that they have not been able to reach a higher core frequency.

You could of course turn that around and say that most people were surprised that the R420 has close to zero new features. It was clear pretty early that it wasn't going to be a SM3.0 part but that it didn't have VS3.0 or any tweaks to FSAA/AF or things like that was not what i expected.

And i don't think it's that surprising that Nvidia haven't been able to reach as high clock frequencies as they have with the R420, it's a 222 million transistor chip on a different process.

Not that i expected him to be unbiased though. It's not exactly his job

 

[bloodbob]

And i don't think it's that surprising that Nvidia haven't been able to reach as high clock frequencies as they have with the R420, it's a 222 million transistor chip on a different process.

I thought both chips where on Low-k .13 what are they on?

 

[Bjorn]

I thought both chips where on Low-k .13 what are they on?

I thought that the IBM 0.13 process is not Low-K. And there's still the transistor difference to take into consideration.

 

[PatrickL]

I dunno how Mr Orton can poke the finger at NV's clock speeds since ATI couldn't manage to build a chip with PS3.0 on .13

I think the problem in your sentence is you decide that ATI could not do it while they clearly state they did not want to, considering all the other factors like price, process, power and so on.