NVIDIA Fermi: Architecture discussion

Not the AA effect by itself, but the denomination of zero sub-samples determining the pixel value is leading to that conclusion. A pixel with no sampling has no resolved value to display.

BTW, that's the weirdest off-topic ever!

 

http://tech.tbreak.com/2009/11/nvidia-past-present-future-an-interview-with-jen-hsun-huang/1/

Interview with Jen Hsun Huang, who dangerously compares Fermi with NV30...

 

Maybe he can also explain why there's no DX10.1 then on GT200 and they need a new architecture for it?

Bad editing in the article all-around, is this a botch translation?

 

Fermi possibly delayed til March or Apri

From Driverheaven...

We heard word a short while ago from our contacts in the Far East that Fermi may very well be delayed even longer than the initial launch estimate in January. We reported last week that the goal for nVidia was to launch the product on some level at CES with reviewer kits to go out shortly afterwards. Our sources tell us today that the Fermi Build Kits were meant to go to AIBs last week but nVidia has missed this time frame. They also told us that "Gemini" is meant to be the code name for the new dual GPU Fermi card although nVidia are not confirming this.

Going on this and other inside information passed to us at the weekend it appears that the initial launch date in January may very well be pushed back until March or possibly even early April. Giving solid dates for the launch is going to be hard as nVidia seem to be running into more issues every week.

http://www.driverheaven.net/news.php?newsid=345#ixzz0Xhun3tPA

And HardOCP

http://hardocp.com/news/2009/11/23/more_fermi_rumors_good

NVIDIA had a big get-together in Hong Kong last week where a lot of the NVIDIA upper brass was on hand. It is being rumored that the first "Fermi build kits" were supposed to be prepped to go out to AIBs but NVIDIA missed the mark on having its Fermi tech ready to pass off to builders. We also heard the "Gemini" name being batted around as well and this seems to be connected to a dual GPU Fermi card. We still feel as though NVIDIA is aiming for getting Fermi shipping by the last week of February, but still think it will miss this mark. HardOCP's best guess right now is late March with any quantity to speak and we are still giving NVIDIA about even odds of marking March. April 2010 is looking more reasonable to us at this time.

 

Ring up another one for Charlie? It will be fun seeing them try to compete with GT2xx till April next year.

 

I'll concede now, regardless of when Fermi launches there will some people here saying that
a) it's coming soon
b) it's not delayed, never has been
c) x months after Cypress does not matter

 

I told you it would be six months away, if they were lucky. And we still don't know if they can deliver what they've been promising or have to settle for something less.

 

Depends on whether you mean Cypress launch or availability

Yep, people's memories are very short. The delay in itself won't be a disaster, the delay plus underwhelming performance would be.

 

well according to some (Razor1 @ R3D) fermi based products have already launched seeing as you can "pre-order" ? /lol

 

The perfect R600 Recipeᵀᴹ .

 

Indeed

 

My my. How quickly times can change in gpu industry. Within 3 years, and 2 major generations, we have history repeating itself. AMD has the first DX11 card, with a similar kickass performance while nv is late and it's perf is as yet unknown. Then nv launched CUDA, while ocl makes it's debut for AMD this time.

 

Since when is OpenCL an AMD thing?

 

That's rather revisionist. The grandaddy of all facepalms on which the R600 is only a footnote is of course... NV30. That was noisy, underwhelming, hot, late and almost instantly EOL'ed, and hyped like crazy as the best thing ever for six months before release. Nothing has beaten that low bar... yet...

 

Soon, my young padawan, happen it will!

 

I don't share that opinion, while the R600 (HD2900XT) release was late, ran hot and was under performing it did eventually spawn the likes of the RV670, RV635, RV620 (HD3000) which lead to the highly succesful RV770/740/730/710 (HD4000) and now the HD5000 series.. all from the carcass of the nearly still-born R600.

 

When does the statute of limitations expire on NV30? Or is there no such thing in this industry?

 

Wow. Someone who agrees with me on the R600. This is a historical date. I still to this day don't think the R600 was that bad. Specially when you look at how AMD designs have developed since it.

 

The kickass performance is pretty meaningless until competition arrives. NV30 would have seemed kickass if the 9xxx series from ATI did not exist.

 

same time when the whole "Ati drivers Suck!" or the whole "Quack" thing expire I would assume.