Kepler Nvidia GeForce GTX 780

[Chickenfeed]

How does that adaptive vSync differ from a simple 60 FPS cap? I was hoping for something that would help with multi-gpu micro-stutter, but it looks to me this simply turns vsync off <60 FPS.

At least 4 monitor Surround seems to be confirmed. Hopefully it's also possible to use mixed monitors.

With a 60 fps cap alone tearing can still be a problem as you can still be out of sync. With this adaptive solution it basically will reduce the slow downs youd normally encounter at sub refresh rate frequencies as well as making triple buffering less of a necessity but you still have the benift of no tearing when performance allows. Basically the best of both worlds.

I was hoping this would come along on PCs sooner than later and here we go :-) FYI many consoles games have been using adaptive vsync for a while now.

[GoldenTiger]

Release can't come soon enough... glad to have seen all these leaks though, helps take the edge off waiting. Hopefully we'll see some vendors in the US place up early listings with pricing so we get a better idea of what it'll end up priced at here.

[Callsign_Vega]

This adaptive Vsync says nothing about getting rid of the input lag associated with VSync. That was one of the key items for the Lucid MVP.

[DarthShader]

Check out the Steam hardware survey. High end GPUs are MUCH more popular than ultra high resolution monitor setups. So it goes to reason that more people using high end cards for lower resolutions than for higher resolutions.

There are some problems with the Steam survey. First, is it an unbiased representation of all the gamers all over the world? In other words, is every gamer in the world equally likely to have a steam account and be included in the survery, regardless of where he lives and what rig he has and how many games he can afford? Very unlikely. Second, how is SLI/Crossfire being calculated? Seems like such systems are treated exactly like different persons having single cards. So the number of people having high end cards would be to high from this side, otoh people with slower cards but in SLI/crossfire are not counted in. Add in the non-strict definition of what high end is and you can't really determine how many gamers actualy do have such rigs. Thirdly, looking at the plethora of weird multimonitor resolutions it would appear having different bezel compensation settings results in a different resolution being registred, which dilutes the scores there. Bottom line: I wouldn't be so hasty with the conclusions.

My point is, I am not against benching at 1080p with 8xAA. I always look with interest at the data points, if a review provides them (like hardwarecanucks does). But if that is the only data point provided, then I smell a marketing rat. There are lies, damn lies and statistics.

[GoldenTiger]

There are some problems with the Steam survey. First, is it an unbiased representation of all the gamers all over the world? In other words, is every gamer in the world equally likely to have a steam account and be included in the survery, regardless of where he lives and what rig he has and how many games he can afford? Very unlikely. Second, how is SLI/Crossfire being calculated? Seems like such systems are treated exactly like different persons having single cards. So the number of people having high end cards would be to high from this side, otoh people with slower cards but in SLI/crossfire are not counted in. Add in the non-strict definition of what high end is and you can't really determine how many gamers actualy do have such rigs. Thirdly, looking at the plethora of weird multimonitor resolutions it would appear having different bezel compensation settings results in a different resolution being registred, which dilutes the scores there. Bottom line: I wouldn't be so hasty with the conclusions.

My point is, I am not against benching at 1080p with 8xAA. I always look with interest at the data points, if a review provides them (like hardwarecanucks does). But if that is the only data point provided, then I smell a marketing rat. There are lies, damn lies and statistics.

Do you have a better cross-section sample data of PC gamers than Steam does? Please do provide it if so. In the meantime, it's the best we know offhand, and using your own common sense along with it results in what can be a very good analytical tool.

[Manicdan]

the 680 is impressive in a few areas
they added some decent features that might actually serve a purpose
and they supposedly are faster than a 7970, but also very small, sub 300mm2 vs the 7970s 365mm2

the unimpressive part
is that it took so long
and that the performance gain over 7970 is psudo (as in the turbo feature is used vs a stock competing part, while OCed it gets smashed). this will require a review or two to find out the case.

[Vardant]

First, is it an unbiased representation of all the gamers all over the world?

It's free, random and available all over the world with some of the most played games in the history of gaming and one of them can be played even on a freakin Adroid phone. How much more unbiased can it really get?

[GoldenTiger]

EDIT: Deleted double-post.

[GoldenTiger]

the 680 is impressive in a few areas
they added some decent features that might actually serve a purpose
and they supposedly are faster than a 7970, but also very small, sub 300mm2 vs the 7970s 365mm2

the unimpressive part
is that it took so long
and that the performance gain over 7970 is psudo (as in the turbo feature is used vs a stock competing part, while OCed it gets smashed). this will require a review or two to find out the case.

It's 10 weeks, nothing in market terms or even real-life terms...
Performance gain is stock-to-stock, both can be OC'd, but variable amounts as always. You can't compare OC-to-OC as it always varies. The 680 has a turbo feature that is done automatically aka stock however, it's not oc'ing.

[DarthShader]

Do you have a better cross-section sample data of PC gamers than Steam does? Please do provide it if so. In the meantime, it's the best we know offhand, and using your own common sense along with it results in what can be a very good analytical tool.

You have missed my point. It is in the second, the shorter, paragraph.

I have never claimed to have better data than Steam provides, furthermore, my first argument against Steam data shows why it is practicaly impossible to have any better data. While the second and third argument show that common sense can decieve you when looking at the data. While it is certainly true that it is "the best" we have, it doesn't mean it's "good".

It's free, random and available all over the world with some of the most played games in the history of gaming and one of them can be played even on a freakin Adroid phone. How much more unbiased can it really get?

It can get more unbiased, if Steam becomes the only source of acquiring games, there's no freeware independent sources and there is no possible piracy either. So every gamer in the world is forced to have a steam account to play games. We are far from it.

[DarthShader]

...

[highoctane]

I'm not overly concerned with any so called turbo feature if it works anything like a cpu, you basically run balls out until an extreme thermal or tdp limit is reached at which point things throttle down to stay within the limits. Actually turbo is a bit of a funny marketing ploy anymore with parts already dynamically clocking up and down as needed, the turbo naming just adds that marketing spin to something that has essentially been happening for a while now.

Who's to say the turboboost frequency isn't the originally intentioned frequency, everything is turboboosted from idle clocks, turboboost blueray playback, etc...

I can't imagine that the turbo would run so close to the ragged edge that it would induce throttling outside of a furmark type utility.

[Olivon]

[spicypixel]

AMD's zerocore 3w idle evidently wins here for idle use but not so much under load...

[jam2k]

They fixed BF3 benchmark:

[Lanek]

Lol...

Still i dont know how they use 8xAA on BF3...

[SubZero.it]

They fixed BF3 benchmark:

Mmm...shouldn't be BF3 where this card performs better?This just makes me think one thing..

[SubZero.it]

They fixed BF3 benchmark:

Mmm...shouldn't be BF3 where this card performs better?This just makes me think one thing..

[Manicdan]

It's 10 weeks, nothing in market terms or even real-life terms...
Performance gain is stock-to-stock, both can be OC'd, but variable amounts as always. You can't compare OC-to-OC as it always varies. The 680 has a turbo feature that is done automatically aka stock however, it's not oc'ing.

it might not be a stock to stock comparison
if they are using that turbo feature in their slides then its possibly showing us the limit of the chip. if you buy an OC edition of a gpu, you get generally a solid % gain in most cases. but for nvidia its like they took that option away from people. they say its possible to OC and let turbo run, which makes no sense because the turbo is a TDP limited feature, and would have done just as good if you didnt touch the clocks. no where did they show us an example that when overclocking you get more than using stock+turbo. they only showed us that you get less turbo benefit.

i would like to be proved wrong because that turbo thing is nice for people who want a faster gpu, but we all know that performance comes at a price, and we oc to get more for free, this might stop us from being able to OC, and thus a very bad thing to many folks here.

[Vardant]

Just think for a second. Nobody is ever going to release and mass produce a chip, that has to always run on the edge of its capabilities. That would be insane and insanely stupid.

Boost might be limited to fit to a certain power envelope only if you don't OC for example. Problem solved.

[aussie-revhead]

1000mhz default? YES

But how well will it overclock? This will be an important factor.

[Vardant]

NVIDIA's New AA Algo is TXAA, Adaptive V-Sync and New 3DVision Surround Detailed

The new 3D Vision Surround is said to work in conjunction with Adaptive V-Sync to ensure the center display has higher frame-rate (since it's at the focus of your central vision), at the expense of the frame-rates of the two side displays (since they're mostly at your peripheral vision). This ensures there's a balanced, high-performance experience with multi-monitor gaming setups.

[Chickenfeed]

Just think for a second. Nobody is ever going to release and mass produce a chip, that has to always run on the edge of its capabilities. That would be insane and insanely stupid.

Boost might be limited to fit to a certain power envelope only if you don't OC for example. Problem solved.

This.

[OCX600RR]

So hold up, can somebody straighten this out for me. Maybe I'm not reading it right, but this new TXAA feature, will it be available to be used in any game through the Nvidia Control panel, or will it only work in games that support it? Mainly asking because I'm heavily into flight simming and not sure if this new TXAA would work in a game that came out roughly 6 years ago (MS FSX)...

[boxleitnerb]

TXAA needs game support. It will be exclusive to Kepler at first, but later it may be available on older chips.