Galaxy GeForce GTX 680 4 GB

[Li4m79]

Originally Posted by Tat3

Is this same card coming to Europe under KFA2 name or are there going to be some differences ?

KFA2 = Galaxy

[---TK---]

Is this card really doable with the 256bit bus?

[Agent-A01]

doubt it.. too bad the card isnt atleast 320bit

[Mike V]

Originally Posted by WhiteLightning

2gb wouldnt be enough for me , im already maxing my card with games. and thats without any AA overriding. bought the project cars game yesterday, and it was also maxing it.

Project cars has a bug in dx11 when it comes to memory usage. It isn't using as much as it reports. I remember reading something about it in the WMD forums.

Other games do come close to maxing my 1.5GB too though.

[snowdweller]

Originally Posted by slickric21

No it doesn't.

At 1080p and x2msaa it uses ~ 1700mb at x4msaa it uses ~ 1800mb.
You wouldnt know it used more as you only have 1.5gb vram.
It was the same on my 580 3gb and now with my 7970.

Im very interested in this 4gb card as a few games i like use more than 2gb at 1080p ~ Skyrim and Crysis 2 with high res textures.

Its getting very boring reading the same arguments about Vram.

I concur, I'm assuming a game will use what VRAM you have E.G.

2 identical systems only difference one has a 580 the other has a 7970.
The first 10 seconds of Crysis 2 for example (all maxed settings) on a 580 will use 1.3gig VRAM whereas the 7970 will use 1.7gig both flucuating between 40-60 fps just because it can. Wether its the card or the game doing this and the other 400mb is in RAM i don't know but this is the way I see it. IMO its like 2 different sized cups with water yet they both taste the same.

[RoadKillNL]

Originally Posted by Mike V

Project cars has a bug in dx11 when it comes to memory usage. It isn't using as much as it reports. I remember reading something about it in the WMD forums.

Other games do come close to maxing my 1.5GB too though.

My Skyrim goes over 1,5GB with just the Official HD Pack and some "Minor" Texture addons.

That's with 4x MSAA, 16x AA and 2x SuperSample in 1680 x 1050.

[Year]

Originally Posted by Agent-A01

doubt it.. too bad the card isnt atleast 320bit

http://www.neoseeker.com/Articles/Ha...TX_680/10.html

"The GTX 680 dominates in our Batman benchmark, with the most interesting results displayed at 2560x1600. With a 256-bit frame buffer, we expected the HD 7970 to pull closer when testing at the highest resolutions, but that just isn't the case in all games.

320 sounds better than 256 and 256 sounds better than 128 that's for sure..

however (as an example) the AMD AthlonXP 2000+ @ 1.7ghz was faster than the Intel 2.4ghz, people who assumed more mhz automatically meant more speed were wrong because in the end its the architecture of the cpu that dictated the advantage, same thing with graphic cards, there's more than sheer mhz or in this case "bits" alone.

in many cases a good 256bit videocard can be much faster than a 320/384 bit card.

Originally Posted by Darren Hodgson

The only game I've seen so far on my GTX 680 that eats up all 2 GB of its memory is Aliens vs. Predator running under DX11 with everything maxed, including 4xMSAA, at 1920x1200. The memory usage is 2,000 MB but this is the first time I've been able to play that game smoothly, without a single hitch or judder, from resuming a save as it was a stuttery mess on my GTX 580 on those same settings.

i was wrong about AVP not using all of the memory of the card, i just tested and indeed the game used the entire video memory (2GB)

i was right however that more memory don't always translate into more performance, it really depends on how the game use the memory.

so it's not enough to just fill the entire vram apparently.

http://www.fudzilla.com/reviews/item...tested?start=5



as for the 4GB debate (talking in general)..

if BF3 put your video ram to good use and 90% of games don't, then 4gb is pointless unless you got money to burn and a multi-monitor setup, can't base a purchase on a single game (BF3) and by the time games start using all the vram for real there will be cards twice as fast than what's available now, coming with 3gb or 4gb vram default at more affordable prices.

going for the 2GB today is not really a compromise, maybe a year from now, just not today.

the GTX680 4GB success will be based mostly on the price.

[poo417]

Originally Posted by slickric21

No it doesn't.

At 1080p and x2msaa it uses ~ 1700mb at x4msaa it uses ~ 1800mb.
You wouldnt know it used more as you only have 1.5gb vram.
It was the same on my 580 3gb and now with my 7970.

Im very interested in this 4gb card as a few games i like use more than 2gb at 1080p ~ Skyrim and Crysis 2 with high res textures.

Its getting very boring reading the same arguments about Vram.

That's what I was getting at. On my 480's it is hitting the max of the ram. I do see the full 1500 occasionally. However with three oc'd 480's I get 80 - 190 fps at that res but prefer to use the higher res so min everything instead and a much better gaming experience.

It is weird that I get no stuttering you would think if the game uses that much on cards with 3 gig it would start to stutter when it hits the vram wall. I did briefly pick up a pair of 3 gig 580 but sent them back. They still did not have the power (or drivers) to make any large difference over the sli 480's so did not bother getting the third.

I will wait and see what the results of the 4 gig cards are. If they are similar to the 2 gig cards in surround I will just wait on the full kelpler to come our instead.

[Redemption80]

I'm not convinced games will not just use what is available, whether it brings any performance benefits or not.

When the 2GB 256bit GTX680 is keeping up, and alot of the time even beating with the 3GB 7970 in resolutions like 5760x1200 it's hard for me to worry that if i get a 680 then the 2GB won't be enough for 1920x1080.

[Tat3]

Originally Posted by Li4m79

KFA2 = Galaxy

I know that, but there are some differences between cards you can buy from Galaxy and from KFA2. KFA2 does offer GT520 2GB with Passive cooling while Galaxy does not have it. Some other stuff too, but I'm feeling too lazy to find it.

[slickric21]

Originally Posted by poo417

That's what I was getting at. On my 480's it is hitting the max of the ram. I do see the full 1500 occasionally. However with three oc'd 480's I get 80 - 190 fps at that res but prefer to use the higher res so min everything instead and a much better gaming experience.

It is weird that I get no stuttering you would think if the game uses that much on cards with 3 gig it would start to stutter when it hits the vram wall. I did briefly pick up a pair of 3 gig 580 but sent them back. They still did not have the power (or drivers) to make any large difference over the sli 480's so did not bother getting the third.

I will wait and see what the results of the 4 gig cards are. If they are similar to the 2 gig cards in surround I will just wait on the full kelpler to come our instead.

Oh okay mate i see what you mean now.

I had a 580 1.5gb and then a 3gb version so was quite lucky to be able to test them both in the same games.

Both cards at stock yielded the same fps in Crysis2, & AvP however the 3gb card felt smoother when stood on the spot and turning around for example. I noticed this effect in GTA 4 aswell but didnt spend that long to check it.

So imho fps doesnt tell the whole story, fps doesnt = smoothness all the time.

Id love to see Techreport, who also use frametimes to assess gpu performance (aswell as traditional fps) do a test on identical card's but one with more vram and see whats going on with the frametimes.

[BlackZero]

Originally Posted by slickric21

Oh okay mate i see what you mean now.

I had a 580 1.5gb and then a 3gb version so was quite lucky to be able to test them both in the same games.

Both cards at stock yielded the same fps in Crysis2, & AvP however the 3gb card felt smoother when stood on the spot and turning around for example. I noticed this effect in GTA 4 aswell but didnt spend that long to check it.

So imho fps doesnt tell the whole story, fps doesnt = smoothness all the time.

Id love to see Techreport, who also use frametimes to assess gpu performance (aswell as traditional fps) do a test on identical card's but one with more vram and see whats going on with the frametimes.

Consider this, all hypothertical.


You run/see a benchmark.


Max FPS = 70
Min FPS = 30

Avg. FPS = 60


You think great. But..


Imagine you are standing in a spot in the game world where you are seeing 65 fps, now say you suddenly turn and a lack of vram causes the fps to drop to 35 momentarily. This jumping effect can happen on a number of occasions but as the fps satyed above the 30 fps minimum and was only momemtarily lower than 65 the bechmark will still show the same results but you sure will feel it.

[dchalf10]

Originally Posted by BlackZero

Consider this, all hypothertical.


You run/see a benchmark.


Max FPS = 70
Min FPS = 30

Avg. FPS = 60


You think great. But..


Imagine you are standing in a spot in the game world where you are seeing 65 fps, now say you suddenly turn and a lack of vram causes the fps to drop to 35 momentarily. This jumping effect can happen on a number of occasions but as the fps satyed above the 30 fps minimum and was only momemtarily lower than 65 the bechmark will still show the same results but you sure will feel it.

I find that extremely annoying in games. That's when you realise that no benchmark will show you what kind of stuttering or frame rate fluctuations you will experience.

If you see min frames of 30 from a bench you might think that the 30 is an odd occurrence during loading or something, but that doesn't mean the game won't drop to 31 frames 10 other times during gameplay.