Nvidia confirms the GTX 580

[Vardant]

Accurate? His predictions are all over the place. First he says there will be no availability, then good availability for 6870 but not 6850. Then he says no news until next year but we know Cayman is still to be launched this year.

OBR is not what I would call a good source..

He knew the spec, the rest, I really don't care, I wasn't going to buy anything anyway. Don't forget, that he's a reviewer, so he obviously knows something.

There is no source, that leaks 100% accurate info and people post news from all around, because you never know.

The point is, I was just translating, nothing else. Or am I supposed to leave the bad things out, because some fanatics here don't like it?

[Aten-Ra]

Why people guess that GTX580 will not be available on release and HD6900 will be ok? when GTX580 is just a tweaked GF100 and Cayman is a new architecture.

[blindbox]

Why people guess that GTX580 will not be available on release and HD6900 will be ok? when GTX580 is just a tweaked GF100 and Cayman is a new architecture.

AMD has been working on that architecture for 5 years. From Radeon HD 2900 to Radeon HD 6800, it has been the same damn architecture every single time. Just optimizations. Plus, Dirk Meyers (I'm sure it's this guy..) officially said around 6 months ago that 6xxx series are coming around this time. It's not something that was said spontaneously.

Oh and, Fermi has reached its peak for 40 nm, simply put. It's too big, it's too hot. If you ask me, I certainly don't know of anything else they can improve (except probably remove everything related to programming and optimize it for gaming so that they can slap more shaders in. That's a lot of change on an architecture)

[Aten-Ra]

You read a lot of Charlie’s BS I see.

Im talking about availability and not if GF100 has reached its 40nm limits or if its too big and hot.

AMD has been working on that architecture for 5 years. From Radeon HD 2900 to Radeon HD 6800, it has been the same damn architecture every single time. Just optimizations.

I believe you should read what Eric Demers (GPG Chief Technology Officer, AMD) have said about Cayman.

http://translate.google.com/translat...d%2F137168.htm

It is a new Architecture with 4 to 1 FP64 vs 5 to 1 in Cypress (HD5800) meaning it could possible be a new 4-Way VLIW and not 5-Way VLIW that Cypress is.

[blindbox]

You read a lot of Charlie’s BS I see.

Im talking about availability and not if GF100 has reached its 40nm limits or if its too big and hot.



I believe you should read what Eric Demers (GPG Chief Technology Officer, AMD) have said about Cayman.

http://translate.google.com/translat...d%2F137168.htm

It is a new Architecture with 4 to 1 FP64 vs 5 to 1 in Cypress (HD5800) meaning it could possible be a new 4-Way VLIW and not 5-Way VLIW that Cypress is.

Whoa hold on a sec. It's logical thinking that yields are lower with bigger chips. Tell me if I'm wrong. Heck, fermi was supposed to go with the full 512 sps, why didn't they? I forgot to mention this, but like another poster said, nvidia is playing defensive (sudden GTX 580 announcement anyone?), simply replying to whatever the competitor has to offer, and not get in the lead.

Anyway, I guess my sentence was worded wrongly. AMD has been working on Cayman's architecture for 5 years is what I meant.

Read my post carefully too, I never mentioned 6900, which uses VLIW4. 6800 still uses VLIW5.

Sigh. This is what I get for responding.

[Aten-Ra]

Well my friend, I was talking about HD6900 Cayman and why GTX580 would not be available at lunch and you replied about 6800 and that GF100 is big and hot. What that has to do with my question ??

You take a lot of BS as fact and you don’t think for your self. GF100 is in production for more than 6 months now. With a tweaked redesign they could be able to unlock all the SMs of GF100 and at the same time lower heat and have good yields.

That is way easier to do vs a new architecture and that is what my question was.

[kuroikenshi]

Well my friend, I was talking about HD6900 Cayman and why GTX580 would not be available at lunch and you replied about 6800 and that GF100 is big and hot. What that has to do with my question ??

You take a lot of BS as fact and you don’t think for your self. GF100 is in production for more than 6 months now. With a tweaked redesign they could be able to unlock all the SMs of GF100 and at the same time lower heat and have good yields.

That is way easier to do vs a new architecture and that is what my question was.

Apparently there are some limitations as to what fermi (gf100) can do

[blindbox]

Well my friend, I was talking about HD6900 Cayman and why GTX580 would not be available at lunch and you replied about 6800 and that GF100 is big and hot. What that has to do with my question ??

You take a lot of BS as fact and you don’t think for your self. GF100 is in production for more than 6 months now. With a tweaked redesign they could be able to unlock all the SMs of GF100 and at the same time lower heat and have good yields.

That is way easier to do vs a new architecture and that is what my question was.

A lot of BS? One line on fermi and it's a lot of BS. It's big? Truth. It's hot? Truth. Maybe I should mention a lot of power as well.
You don't simply shrink shader size in 6 month without a die shrink either. And another round of edit. There's not much you can tweak with Fermi. Plus, where did you hear that GTX580 is a tweaked Fermi?

Anyway, my bad, let me rephrase my passage.

Cayman uarch has been in research for 5 years. Hence, that's why we can believe that Radeon HD 6900 is coming this month because they've spent YEARS on it. Another reason why we can believe it, is because they've already planned for it (Dirk Meyers said around 6 months ago it's coming at the end of this year). Their execution too, has also been flawless lately, compared to nvidia.

As for my statement from Radeon HD 2900 to Radeon HD 6800, I'm just saying that they didn't spend much doing it (same architecture for 4 generations, yeah). Nvidia used their G92 architecture for only over 2 generations.

Sorry for my bad english. Stuff happens when I don't proof read.

[Dimitriman]

Even JHH himslef was quoted a few weeks ago whem talking about kepler that response to cayman would only come january.

[blindbox]

Even JHH himslef was quoted a few weeks ago whem talking about kepler that response to cayman would only come january.

Lol that seals the deal.

[Dimitriman]

Lol that seals the deal.

only sorta. i think this time they actually have a product but not available yet. so i think instead of keeping people waiting with announcements they will paper launch and show real figures which should work better at holding people off buying ati than the little dude holding the green screen.

[kuroikenshi]

Lol that seals the deal.

wooden screw promises here we go.....

[Zloyd]

Aten-Ra I feel nvidia are getting these cards out in a hurry to showoff the 580 and put the brakes on cayman sales, it would make sense for nvidia to do so. Why do I think that the cards wont be widely available soon ... around caymans launch ? Well I believe if there was a card coming out in a few weeks or even a month or so from now there would be a lot more buzz about it, if indeed it is ready and coming out really soon then I feel its been one heck of a secret mission from nvidia and they have done it not to affect their own 480 sales.

[Dimitriman]

found my sources

this for gf110 late 2010 with 2011 availability: http://www.xtremesystems.org/forums/...id-life+kicker

and this roadmap for kepler which i guess now will be the 680gtx: http://www.tcmagazine.com/tcm/news/h...ming-next-year

[spursindonesia]

Why people guess that GTX580 will not be available on release and HD6900 will be ok? when GTX580 is just a tweaked GF100 and Cayman is a new architecture.

Charlie sucked on many fronts regarding GPU leaks, especially lately on the exact specs of the chips, but regarding the state of GPU chip fabrication in foundry (TSMC), the guy is plenty darn accurate for the past 2 years. He said that GTX 580 would be released soon, but limited in availability, i trust him more (compared to JHH parrot called Fuad Abazovic) until it's proven wrong.

Early edition of GTX 580 might be coming from the pilot or risk wafers that were limited, while broad availability might have to wait until full production batch comes over. I certainly hope that good availability will arrive sooner, so price competition would heat up, but we will see how it unfold, won't we ?

[Your_Boss]

My guess is that gf100 will EOL soon after GTX 580 comes to market, i.e. not just on paper. And of course Nvidia will make a GTX 570 from the scraps that don't hit the 580 requirements. Also a full GF104 with upped clocks is possible. That would give a full line up against Cayman GPUs.

Those are called GTX480s

[Sn0wm@n]

Even JHH himslef was quoted a few weeks ago whem talking about kepler that response to cayman would only come january.

exactly .. and lying to investors isnt a good corporate move ....

[cold2010]

ENGTX580/2DI/1536M



http://www.asus.com.cn/ContentPage.a...ontent_Id=3985

[Sn0wm@n]

is that some sort of pricing next to the numbers ????


and how much in us $ would these means if they are ???

[Metroid]

4799 CNY = 718 USD

[Sn0wm@n]

ouchhh .....

[AMDDeathstar]

700 dollars?

ENGTX580/2DI/1536M



http://www.asus.com.cn/ContentPage.a...ontent_Id=3985

[Chrono Detector]

If this card is priced at $700, hopefully not, I don't see NVIDIA will price this too long when ATI releases the 6970, assuming that the 6970 is at a cheaper price point or unless if ATI decides to play the high price gaming point.

[Gilgamesh]

Here:
http://vr-zone.com/articles/nvidia-g...th-/10222.html

say 599$.

[Sn0wm@n]

and those writting on that vr-zone article came from where exactly ??? some sort of forum???


the asus link is as real as it gets ....