NVIDIA Kepler speculation thread

The folks at Anandtech's forum seem to believe that the review is fake because it's not possible to run 8x MSAA in BF3.

Yeah, I'm looking at these with huge ball busting grains of salt for now.

We are of different opinions then :smile:. I don't know of many who game that often in a 24 hour period where the difference in load would matter that much. However, in 2D mode when some can go days without playing a game is a different situation.

Add to that situations like mine (which admittedly might be rare) where I have to leave the computer on for days (sometimes weeks) at a time and long idle power makes a quite significant difference.

Especially with multimonitor setups (which isn't rare, lots of people have 2x monitors) where neither company can clock down memory speeds in idle like you can with signle monitor setups. Although AMD can clock down memory if all monitors are the same model and brand. Not sure if Nvidia has that or not. Long idle in this case is an absolutely huge difference in power consumption between 7xxx and other AMD or Nvidia cards.

Regards,
SB
 
The folks at Anandtech's forum seem to believe that the review is fake because it's not possible to run 8x MSAA in BF3.


I dont know if it is fake but this is not a review as we can intend by several sites, there's a lot of error and mismatch:

- BF3 settings ( effectively the max is 4xMSAA+FXAA in ultra )
- 3Dmark11 too low in conjunction with an I7 3960K for the 7970 ( score with the cpu oc at 5ghz are even lower in gpu test for the 680 it is reported on the benchmark slides ). For give you an example, a I7 2600Kghz + 7970 give 11K in Physic test, and with an I7 3960K @5ghz give 16K )

- LostPlanet2: Test A or B? strange numbers ..

- Reviewers have the card, but not this driver and they dont know what is the driver they have use ( and will be happy if someone can give it to them.... lol )

Or they have take different scores from a previous test, or they have take it from an older sites reviews and add what they have find or they just report scores given by other ... i dont know.
 
it's not indicating the idle power usage, it indicates previous power state than 3D clocks.. so you cant conclude a result from it.. GF114 is bigger and manufactured on a worse process, now there is GK104 is smaller and manufactured better process tech.. Tahiti is much bigger, yet draws less than Barts.. i dont expect anything above 15W idle.. and anything less than 15W shouldnt be a problem for money or heat..

chart shows 200W idle according to your say, seriously 200W idle ??
All this based on the observations on the power consumption chart at the start/end of it's run? It's apparent we won't see eye to eye on that one because I don't consider that relevant. :D Again, it is what it is...for now. Lets see what official reviews show.

Especially with multimonitor setups (which isn't rare, lots of people have 2x monitors) where neither company can clock down memory speeds in idle like you can with signle monitor setups. Although AMD can clock down memory if all monitors are the same model and brand. Not sure if Nvidia has that or not. Long idle in this case is an absolutely huge difference in power consumption between 7xxx and other AMD or Nvidia cards.

Good point to bring up. We would need to see how this develops once the official benchmarks are released. There are few things we have to consider:
-retail card or something else
-old drivers or release drvers
-etc
So there is still some other variants still in play.
 
Force it by driver ? it is not usable by Ingame settings.. max is 4xMSAA and add FXAA ... who in a review will force the AA by driver ? instead of use the setting in game ?

Hmm, is 8xAA done different between CCC and NCP?
 
Both have 8xMSAA available, though the performance hit might be different from the usual in-game implementation, depending on the game.
 
I really dont see anyone want to try to force 8xAA ( or 8xMSAA ) when the game is made for use 4xMSAA +FXAA or only FXAA ....

This is the max setting you can set ( outside resolution ofc )


But i believe this is mostly a typo in both case... ( they have just do one time the 1920x1080 test as settings used )
 
Hmm, is 8xAA done different between CCC and NCP?

If working at all they are completely different (and slower) from the ingame msaa (which is only applied at the right surfaces etc). The overclock3d numbers are most probably just a mistake in the graph.
(and btw, BF3 is a fine example where you can't replace msaa with fxaa - as the fxaa doesn't cure the actual aliasing on distant objects/details)


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).[/B] This ensures there's a balanced, high-performance experience with multi-monitor gaming setups

Apart from the inherent problems of getting this to work in the games (some frames only rendering parts of the frame..), it the peripheral vision that is the fast one ( like we went through recently: http://forum.beyond3d.com/showpost.php?p=1596364&postcount=6 ). And what does it have to do with the 60hz frame capping? Sounds like BS.
 
Last edited by a moderator:
Then there's the new 3D Vision Surround, bolstered by a redesigned display logic, which addresses the two-display limitation of NVIDIA GPUs. You can now connect as many as four monitors to a GeForce Kepler GPU, enabling 3-monitor HD 3D Vision Surround setups. You no longer need more than one GeForce GPU to connect more than two monitors. 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.

TPU​

I had a similar idea way back when I first tried out eyefinity, except I assumed that the peripheral screens would just have lower IQ. I wonder how this will work though, lower framerates at the sides could be annoying enough even though you aren't really looking at them.
 
lanek, I do see your point of view here. I do have to wonder why all stock IQ settings weren't used? Also even though they call it 4xMSAA in Bf3 isn't it some sort of AA deferred?



If working at all they are completely different (and slower) from the ingame msaa (which is only applied at the right surfaces etc). The overclock3d numbers are most probably just a mistake in the graph.
(and btw, BF3 is a fine example where you can't replace msaa with fxaa - as the fxaa doesn't cure the actual aliasing on distant objects/details)
Thanks for the information. I know there was an explanation about that sometime ago but couldn't find it.
 
If working at all they are completely different (and slower) from the ingame msaa (which is only applied at the right surfaces etc). The overclock3d numbers are most probably just a mistake in the graph.
(and btw, BF3 is a fine example where you can't replace msaa with fxaa - as the fxaa doesn't cure the actual aliasing on distant objects/details)

FXAA cures more aliasing than MSAA in BF3, MSAA misses most edges altogether on BF3.

FXAA blurs the image though, of course.
 
I had a similar idea way back when I first tried out eyefinity, except I assumed that the peripheral screens would just have lower IQ. I wonder how this will work though, lower framerates at the sides could be annoying enough even though you aren't really looking at them.

Same, Im really curious to see what will be the result.
 
lanek, I do see your point of view here. I do have to wonder why all stock IQ settings weren't used? Also even though they call it 4xMSAA in Bf3 isn't it some sort of AA deferred?




Thanks for the information. I know there was an explanation about that sometime ago but couldn't find it.


Im sorry it is in french, yes it is 4xmsaa deferred, and the second line under it is FXAA ( low, medium, high quality ) ... instead of Battlefield BC2 ( who was use standard MSAA ( 8xMSAA ), this one use a real specific AA system..
you dont want to force any other AA setting in the CCC or NVP for override them, in the best case you dont have any AA applied, in the worst the game is just unplayable )

Still i think they have just copy the text in the graph saying 8xAA, but thier performance numbers dont match at all with thoses max AA settings ( who kill the performance by nearly 30% )
 
Apple still doesn't support Flash.

yes Apple is at war against scripting languages except the slowest of all, javascript.
they don't want flash because you can make "apps" with it (with access to camera, microphone)

they even banned a children's programming language and a C64 emulator for the same reason.
you must follow the Apple way else you're not welcome on the ipad.
 
Force it by driver ? it is not usable by Ingame settings.. max is 4xMSAA and add FXAA ... who in a review will force the AA by driver ? instead of use the setting in game ?

I admittedly never tried, but does expand the ingame 4x-setting to 8x not work? That would be the easiest explanation besides a typo.
 
I had a similar idea way back when I first tried out eyefinity, except I assumed that the peripheral screens would just have lower IQ. I wonder how this will work though, lower framerates at the sides could be annoying enough even though you aren't really looking at them.

Lower fidelity/IQ would be somewhat OK for the side monitors except that oftentimes people can and do turn their heads to focus on something briefly that their peripheral vision caught, rather than turning their entire ingame viewpoint to it. But I think it could in certain games be a net benefit (racing games for instance).

Lower framerates would just be bad. Across the board bad. With absolutely zero redeeming features, IMO. Even if it's your peripheral vision, having it update once per 2 frames in main vision is going to be distracting.

Now if you move it to any type of strategy game (Civ 5 for instance) you'll immediately notice the effect on your mouse as it'll get less responsive the moment you move from primary monitor to the side monitors.

And then how are they going to deal with 3x2 setups? Or 5x1 setups?

If this truly is something Nvidia are putting in. Hopefully they are smart and making it an option that you opt in on, rather than the default.

In other words, the only good thing that could come of lower framerates on side monitors is to boost benchmark scores. It would be absolutely useless in actual gameplay.

Regards,
SB
 
Im sorry it is in french, yes it is 4xmsaa deferred, and the second line under it is FXAA ( low, medium, high quality ) ... instead of Battlefield BC2 ( who was use standard MSAA ( 8xMSAA ), this one use a real specific AA system..
you dont want to force any other AA setting in the CCC or NVP for override them, in the best case you dont have any AA applied, in the worst the game is just unplayable )

Still i think they have just copy the text in the graph saying 8xAA, but thier performance numbers dont match at all with thoses max AA settings ( who kill the performance by nearly 30% )

Agreed. It still possible that those slides aren't real as mentioned before. When playing BF3 the game is supposed to be tailored for deferred geometry pass renders with MSAA, not just MSAA. So that would bring a few things up:
-What's the IQ like if working?
-Is MSAA actually working correctly directly from the drivers?
-Typo?
 
Agreed. It still possible that those slides aren't real as mentioned before. When playing BF3 the game is supposed to be tailored for deferred geometry pass renders with MSAA, not just MSAA. So that would bring a few things up:
-What's the IQ like if working?
-Is MSAA actually working correctly directly from the drivers?
-Typo?

too much question, lets wait official reviews ...
 
Back
Top