NVIDIA Kepler speculation thread

[A1xLLcqAgt0qc2RyMz0y]

Kepler NDA Expiration

I read somewhere (can't find link currently) that the NDA on Kepler expires on Thursday March 22 at 6:00am PDT.

Is that still correct?

 

[trinibwoy]

lol, i thought it was Trinibwoy approXimation AA

Nah that doesn't have a nice ring to it.

 

[Man from Atlantis]

Nah that doesn't have a nice ring to it.

How about Temporal AA?

 

[Blazkowicz]

I ran temporal AA on 3dfx voodoo5. yes, about when ATI implemented it in its radeon drivers, quite many years ago, someone went and made the same for the old 3dfx beast.
I didn't thought it looked any good, yes it was a nice hack to make 2x look like 4x but the flickering was very apparent even at high refresh rate and framerate. tested on quake 1.

 

[Devildoll]

I read somewhere (can't find link currently) that the NDA on Kepler expires on Thursday March 22 at 6:00am PDT.

Is that still correct?

seems possible.

one of swedens biggest gaming/hardware stores are having an event on the 22'n.

with the hint that the main sponsor has a fondness for the color green.

link( google translate )

 

[Man from Atlantis]

 

[Man from Atlantis]

another 640M review

Spoiler

Power consumption ~40 W

http://pc.watch.impress.co.jp/docs/topic/feature/20120316_519144.html

 

[Psycho]

http://www.abload.de/img/sdfsdf0xow.png

Would be more believable if 7970 and 580 actually had the same power usage. Now I just read: "we have different ways of measuring power, but in reality 7970 and 680 is using the same".

 

[Arty]

PR reaching new levels of stupid by fudging TDP numbers now.

 

[leoneazzurro]

Not only, but that slide (if not fake) seems also to mislead about projected performances, as of course the 7970 and the GTX580 haven't the same performance (and I wonder what the performance should be if AMD really choose to have the same real power envelope of the 580's)

 

[CarstenS]

PR reaching new levels of stupid by fudging TDP numbers now.

How's the mentioning of official numbers for GTX 580 (244 watts) and HD 7970 (250 watts) fudging? Because one (the former) is labelled TDP and one (the latter) maximum board power? Those have been used interchangeably in the past too.

Won't comment on GTX 680's alleged 195 watts for now, though.

 

[Kaotik]

How's the mentioning of official numbers for GTX 580 (244 watts) and HD 7970 (250 watts) fudging? Because one (the former) is labelled TDP and one (the latter) maximum board power? Those have been used interchangeably in the past too.

Won't comment on GTX 680's alleged 195 watts for now, though.

Because at least the 580 can use over it's supposed "TDP" even in games

 

[Arty]

How's the mentioning of official numbers for GTX 580 (244 watts) and HD 7970 (250 watts) fudging? Because one (the former) is labelled TDP and one (the latter) maximum board power? Those have been used interchangeably in the past too.

Won't comment on GTX 680's alleged 195 watts for now, though.

If you are going by maximum board power then the GTX680 should be 225W, shouldnt it?

 

[xDxD]

Not only, but that slide (if not fake) seems also to mislead about projected performances, as of course the 7970 and the GTX580 haven't the same performance (and I wonder what the performance should be if AMD really choose to have the same real power envelope of the 580's)

7970 is more efficient than gtx580, but it has a 250w official tdp, if this had never reached in actual use, why put 6+8 pin power?

 

[CarstenS]

Because at least the 580 can use over it's supposed "TDP" even in games

Hm, I see. So that would make the GTX 680 even more appealing in comparison, is it that what you're suggesting?

If you are going by maximum board power then the GTX680 should be 225W, shouldnt it?

I don't know - why would it, assuming the slide we're talking about is legit?

 

[Kaotik]

Hm, I see. So that would make the GTX 680 even more appealing in comparison, is it that what you're suggesting?

Compared to 580 sure, 7970 on the otherhand seems to stay under 190W even as peak value in Crysis 2. (Though, 580 doesn't reach it's TDP in this game either, but Xbit where it went over it's TDP doesn't have similar test done for 7970 so can't do direct comparison there)

 

[CarstenS]

Maybe then the sample size of only 1 game measured is a bit too small to draw conclusions? I mean, I did some measurements for our whole test parcours last year - not wait, 2010 it was - and there was wild variety all over the place. The game we had used prior to that in the older generation of cards was with the new generation suddenly not very representative any more: Among the top power drawers with one architecture and among the lowest ones with the other architecture.

If you are bottlenecked by some part of your architecture in one game, measuring power there does not make much sense, wouldn't you agree?

 

[lanek]

How's the mentioning of official numbers for GTX 580 (244 watts) and HD 7970 (250 watts) fudging? Because one (the former) is labelled TDP and one (the latter) maximum board power? Those have been used interchangeably in the past too.

Won't comment on GTX 680's alleged 195 watts for now, though.

Question is the " rumored 195W" is with this "turbo mode " or without ?

 

[Arty]

I don't know - why would it, assuming the slide we're talking about is legit?

Because they're using PCIe pin plugs to illustrate their point.

 

[SimBy]

How's the mentioning of official numbers for GTX 580 (244 watts) and HD 7970 (250 watts) fudging? Because one (the former) is labelled TDP and one (the latter) maximum board power? Those have been used interchangeably in the past too.

Won't comment on GTX 680's alleged 195 watts for now, though.

That 250W is with +20% powertune right? And even then it consumes less power than
GTX580. Not to mention its actually faster too.

And I seriously doubt enthusiasts are gonna be very enthusiastic about 2x6 power plug.

Unless of course this thing has no OC headroom at all.