EVGA GeForce GTX 560 Ti 2Win Dual GPU Review

[BababooeyHTJ]

Im not going to notice the difference between 19 FPS on a 1 Gb card, and 24 FPS on a 2 Gb card am I?

Neither case would be playable - your link only proves that 2-3 Gb cards arent yet powerful enough to run games that need more than 1 Gb Vram, which is the point that others are making.

Skyrim is out soon, and someone with the same opinions as yours has reported it to be using over 1 Gb Vram at 1920x1200. I cant wait to see 'lag spikes' on my PC



But you're exactly the same with your opinions which over rate Vram

We'll see what happens when the first texture pack gets released for Skyrim. Do you remember what people used to say about QTP3 for Oblivion?

[sniper_sung]

We'll see what happens when the first texture pack gets released for Skyrim. Do you remember what people used to say about QTP3 for Oblivion?

He will insist that he can play with the texture pack fine with 1GB vram per GPU silky smooth, no lags no freezes; or he could just blacklist the texture packs accusing memory leaks or whatever.

[eXa]

We need some proper GTX560TI 1gb vs 2gb SLI testing using the games that actually do benefit from more than 1gb ram.
And no, i dont mean cherry picking games but there is no point in focusing on games that do not have high vram usage because no one is saying that more than 1gb is beneficial for all games but there are some that do.

[ExodusC]

We need some proper GTX560TI 1gb vs 2gb SLI testing using the games that actually do benefit from more than 1gb ram.
And no, i dont mean cherry picking games but there is no point in focusing on games that do not have high vram usage because no one is saying that more than 1gb is beneficial for all games but there are some that do.

I think it would clear it up if we saw GTX 560 Ti 1GB vs 2GB vs 2Win benchmarks on several games at both 1920x1080 and 2560x1440. Maybe throw in a GTX 570 and/or GTX 570 SLI as well, just to see.

That would hopefully clear up this issue. I'm not taking either side, since am really unfamiliar with multi-GPU setups, but I am genuinely curious to know what the results show. I'm happy with my GTX 570's performance as is.

[sniper_sung]

We need some proper GTX560TI 1gb vs 2gb SLI testing using the games that actually do benefit from more than 1gb ram.
And no, i dont mean cherry picking games but there is no point in focusing on games that do not have high vram usage because no one is saying that more than 1gb is beneficial for all games but there are some that do.

Sadly NVIDIA's drivers have been intentionally, actively and aggressively throttling the vram usage in certain popular games regarding GPU model (i.e. aggresive thrashing at driver-level), regardless of vram size, making non-reference cards with large vram almost useless, causing unnecessary lag spikes / stuttering on them. For example, the vram usage in BF3 and Metro 2033 with MSAA 4X 1080p is only around 1000MB on GTX560 Ti 2GB, while it can easily reach 1500MB on GTX580s. Does this introduce problems? Of course yes. For example, users have been reporting that when they open sniper scopes in BF3 there would be a lag on GTX560s (both 1GB and 2GB), loading the textures of objects far away, which indicates that the driver has been cheating by reducing the view distance in stealth, in order to reduce vram usage. NVIDIA will never admit this, otherwise they admit 1GB vram is inferior, which would affect their sales.

I honestly have no idea why people are so persistent to debate this simple question. For those who is not smart enough to understand what is lag spikes / stuttering, they can only read this kind of plots:

[Mungri]

For the limitation of 1GB vram in BF3, a good review should tackle the vram hungry scenes and use MSAA:

Those graphs that HWC did are completely wrong and are not mirrored by other reviews of SLI GTX 560 ti performance in BF3 (TBH it looks like SLI was accidentally disabled on the 560 tis, or incorrect drivers / profiles used).

Theres a video recording right here of BF3 on ultra settings and 4x MSAA at 1080p on GTX 560 ti SLI:



I see no lag spikes. There are other reviews out there which show SLI GTX 560 ti performing as well as it should in BF3, but I cant locate those right now but will continue looking for them.

Several people that I have heard from that are playing BF3 on SLI GTX 560 tis are handling it wonderfully at 60-100 FPS at ultra settings without lag spikes or stuttering.

DA2 and GTA4 are very poor console ports that have not been optimised to run well on PC hardware. I dont see how people can take a comparison of DA2 seriously when it looks worse than games which only require 512 Mb Vram.

We need some proper GTX560TI 1gb vs 2gb SLI testing using the games that actually do benefit from more than 1gb ram.

I fully agree with you, but shouldnt it be the responsibility of the people who claim that Vram is so significant to make those comparisons, rather than people who already know that 1 Gb Vram is yet a limitation to waste their time doing so?

[Mungri]

Here is the correct performance of GTX 560 ti, and this EVGA dual GPU card in BF3:

http://www.guru3d.com/article/evga-g...2win-review/17



I fail to see how SLI GTX 560 Tis with 1 Gb Vram are limited in this game in anyway, unless you look at false graphs.

If you compare 1Gb to 2Gb GTX 560 tis in an apples to apples comparison (I am waiting for this, and it should be the people who claim that there will be any difference to carry out the test), you will not see any differences at 1200p resolution.

[sniper_sung]

I'm still saying the same again: 4 reviews using non-vram-hungry scenes in BF3 does not necessarily mean there is no vram-hungry scene that can bring 1GB down to knees. Of course you can enjoy BF3 facing a wall all the day. I suspect that you might even get a K/D ratio of zero (0) in 64-player maps in BF3, if you dare use MSAA 4X and Ultra at 1920x1200 on your 1GB cards, which would make you easy points for farming.

Furthermore, lag spikes / stuttering is not for people like you to understand, because you have no ability to read scales of y-axis and do quantitative analysis of data. Your claims have been based on subjective feelings all the time, without giving any reasonable proof.

[Mungri]

I understand lag spikes and stuttering perfectly fine when I see them with my own eyes in a video game thanks, and if I got them I would upgrade my graphics cards to eliminate them. It doesnt need any kind of higher capacity to understand other than being able to see them while playing a game. However I dont have any lag spikes yet, nor do I have a single game that actually requires me to use graphics cards with more Vram for smooth and lag free performance.

The graphs which I have seen do not show any indication, nor proof of lag spikes which anyone could actually see or notice while playing a game.

The Guru 3D review above was done across a whole level of BF3, also you can see in the GTX 560 ti video of BF3 that I posted above that the player is going around the whole level playing normally, and not just standing against a wall as you suggest.

[cold2010]

How do you explain this

HD 6870 1GB

[sniper_sung]

The Guru 3D review above was done across a whole level of BF3, also you can see in the GTX 560 ti video of BF3 that I posted above that the player is going around the whole level playing normally, and not just standing against a wall as you suggest.

I never bothered to watch the video you posted until just now I clicked play and realised that it already has stuttering in there, not having to mention that a video transcoded by youtube down to 24fps or 30fps has already lost much information on framerates.

[Mungri]

But it looks far from being unplayable as you claim it should be with only 1 Gb Vram.

[sniper_sung]

How do you explain this

HD 6870 1GB

Please refer to #83 of this thread.

[Mungri]

Your argument in #83 is weak, and merely your own personal opinion, not a fact.

[cold2010]

Please refer to #83 of this thread.

I do not need it because you did not answer my question Why did not occur for HD 6870 1GB, as happened to the HD 6950 1GB, suffocation, but you did not answer the question, "as usual"
I love you man

[BababooeyHTJ]

Your argument in #83 is weak, and merely your own personal opinion, not a fact.

The same can be said about your arguments.

This dude is a pro troll, try to google "bhavv troll" and there you go

He will insist that he can play with the texture pack fine with 1GB vram per GPU silky smooth, no lags no freezes; or he could just blacklist the texture packs accusing memory leaks or whatever.

That really is a lot of trolling.

I think it would clear it up if we saw GTX 560 Ti 1GB vs 2GB vs 2Win benchmarks on several games at both 1920x1080 and 2560x1440. Maybe throw in a GTX 570 and/or GTX 570 SLI as well, just to see.

That would hopefully clear up this issue. I'm not taking either side, since am really unfamiliar with multi-GPU setups, but I am genuinely curious to know what the results show. I'm happy with my GTX 570's performance as is.

Its true that more vram is not necessary for most games. No one is arguing that but to say that there is no benefit to more than 1GB of vram is ridiculous. The vast majority of the games that I play use about 1GB of vram most of the time. I also like to use high resolution texture packs which can really eat up the vram usage and this is in games that do not cache memory. I don't have the means to test this with my modded Fallout 3 install but I can say that with a card with more vram I don't see the hitching that I once did when loading textures.

[Mungri]

Troll:

a troll is someone who posts inflammatory,[2] extraneous, or off-topic messages in an online community

I dont think that you will find that any of the posts I make on whichever forums I use fall into any of those categories. On topic posts which oppose your beliefs and opinions on a topic =/= troll posts.

I can say that with a card with more vram I don't see the hitching that I once did when loading textures.

But your GTX 570 has a much more powerful GPU than any 1 Gb card out there, so you cannot use this as a valid comparison or assumption that the improvement was caused by the mere 256 Mb exra Vram when it is far more likely that your improvements are caused by having a much more powerful GPU.

The only way to test it would be to compare a 1 Gb to a 2 Gb card of the same GPU - the HD 6950 or GTX 560 ti would be the best ones to use as they would clearly show more difference due to Vram as opposed to comparing a 1.5 Gb to a 3 Gb GTX 580. Any Vram comparison should also have a sufficient amount of system ram - at least 8-12 Gb to allow the possibility of cheaper to obtain shared memory to alleviate any lag spikes or hitching problems.

Then there is also the problem with skewed / flawed results due to improperly set up PCs, as the ones that Sniper Sung keeps on posting which are proven to be completely incorrect in other similar benchmarks.

but to say that there is no benefit to more than 1GB of vram is ridiculous.

No one is saying that, we are saying that the difference is so small that it would be unnoticeable.

[eXa]

There there, this is just an forum discussion, no need to get harsh or personal.

SKYMTL, if you are still following:
Your own testing of this EVGA card shows exactly what i have been arguing the whole time.
http://www.hardwarecanucks.com/forum...-review-4.html (same graphs sniper_sung posted earlier in this thread)
1920x1200 with 0 MSAA 2 GTX560TI is well above a singel GTX580, but with 4x MSAA they run out of vram and performance plunge way below GTX580. With 2gb per gpu it would still be above GTX580 and plenty playable.

[SKYMTL]

Those graphs that HWC did are completely wrong and are not mirrored by other reviews of SLI GTX 560 ti performance in BF3 (TBH it looks like SLI was accidentally disabled on the 560 tis, or incorrect drivers / profiles used).

Wrong. Do some research and then test on the same mission we did.

All of the other reviews seem to have taken the easy way out by either using the first mission or using an intentionally "light" area of the game. The area we took is the heaviest in the game that I could find.

[Mungri]

I would like to test the game myself, but I dont play FPS's so it would be a waste of cash for me to buy the game just to test one level.

Your site seems to be down atm, could I ask how much ram the test PC had?

[SKYMTL]

I would like to test the game myself, but I dont play FPS's so it would be a waste of cash for me to buy the game just to test one level.

Your site seems to be down atm, could I ask how much ram the test PC had?

Site is up.

I can't remember off the top of my head how much memory but I'll find out ASAP after SB-E testing is done and I set back up my gaming computer.

Alternately, I would be happy to send someone my Save Game file and they could have a go at it.

You can see all sequences here: http://www.hardwarecanucks.com/forum...-review-2.html

[sniper_sung]

I would like to test the game myself, but I dont play FPS's so it would be a waste of cash for me to buy the game just to test one level.

Your site seems to be down atm, could I ask how much ram the test PC had?

They used 6GB system memory at 1600MHz. As I said, they did a good job by including some zooming in of sniper scopes in some relatively busy areas, which stresses vram by requiring pre-caching of some textures far away. For cards with small vram size it is more likely to get lag spikes / stuttering when zooming in with sniper scopes.

Don't waste your money on this game. You don't know how to prove it's stuttering-free with appropriate data to back up but only with your subjective feelings. Do I have to repeat again that a video clip transcoded to 24 fps or 30 fps would lose much information regarding framerates and stuttering?

[Mungri]

It would be great if anybody could test that same sequence again using:

- 8 or 12 Gb ram (to minimize performance loss due to not enough ram for both system use and shared video ram)
- 1 Gb vs 2 Gb GTX 560 ti SLI.

Though I doubt anyone will, but this would be all that is required to find out how much extra Vram would benefit that scene, and whether or not Vram is the definite cause.

In fact, a whole review with the most demanding games and 'Vram eating' situations at 1920x1200 and 2560x1600 res with 4 and 8x MSAA at both resolutions would make the most accurate comparison for Vram.

Without that, it still remains nothing but assumption and baseless opinion that Vram alone would make a significant enough difference in that case.

[sniper_sung]

It would be great if anybody could test that same sequence again using:

- 8 or 12 Gb ram (to minimize performance loss due to not enough ram for both system use and shared video ram)
- 1 Gb vs 2 Gb GTX 560 ti SLI.

Though I doubt anyone will, but this would be all that is required to find out how much extra Vram would benefit that scene, and whether or not Vram is the definite cause.

In fact, a whole review with the most demanding games and 'Vram eating' situations at 1920x1200 and 2560x1600 res with 4 and 8x MSAA at both resolutions would make the most accurate comparison for Vram.

Without that, it still remains nothing but assumption and baseless opinion that Vram alone would make a significant enough difference in that case.

You still don't understand BF3 would be a bad choice for GTX560 Ti 2GB? I've already said the current NVIDIA driver would only use around 1000MB on a GTX560 Ti 2GB in BF3, which does not take any advantage of the additional vram at all, while GTX580 can use around 1500MB, taking advantage of the pre-caching. Try other tests that are NOT known to be prone of this problem, e.g. try Shogun 2 with script overriding of 8AA, or try Rage with config file of 16k textures.

[Mungri]

I've already said the current NVIDIA driver would only use around 1000MB on a GTX560 Ti 2GB in BF3, which does not take any advantage of the additional vram at all

I dont believe that at all.