2GB Vram The Minimum. Really?

[bhavv]

256 Mb Vram difference does not account for a 14 FPS difference in any situation. Its far more likely that the GTX 570 architecture simply handles Bf3 much better than the 560 ti does.

Your assumptions of 2 Gb 560 ti performance based on that chart are baseless as that card is not included at all in that graph you keep posting. Feel free to refer to what was said by James regarding assumptions.

Show or make a valid comparison including the following cards:

1 Gb and 2 Gb 6950
1 Gb and 2 Gb GTX 560 Ti
1280 mb GTX 570

Single GPU and crossfire / SLI results or stop over rating and speculating about Vram.

There is absolutely no way that the 2 Gb GTX 560 ti would come close to a single / dual GTX 570 1280 Mb in any game including Bf3, so why bother recommending more Vram when its clearly only the GPU that matters?

You can put 100 Gb Vram on a GTX 560 ti if you want, it still isnt going to perform any better than simply having the next GPU up.

Since you like repeating meaningless graphs which dont actually support the Vram debate, I will repeat some more accurate information:

http://www.tomshardware.com/reviews/radeon-hd-7950-overclock-crossfire-benchmark,3123-6.html
6970 2GB loses 33% frame rate on 4xMSAA application, whereas GTX570 1.25GB only loses 19.5% on 4xMSAA application.

Grunt/GPU architecture>VRAM for most games, even for games that are known to use VRAM a little over than what's available.

 

[James J]

http://www.overclock.net/t/801683/s...dia-gtx460-is-the-mainstream-card-good-enough

Benchmarks done on a system with 2x1gb 460s vs 2x2gb 460s in the same machine. Vram limiting situations used (high resolutions first, then AA added)

Food for thought.

 

[Final8y]

256 Mb Vram difference does not account for a 14 FPS difference in any situation.

Your assumptions of 2 Gb 560 ti performance based on that chart are baseless as that card is not included at all in that graph you keep posting. Feel free to refer to what was said by James regarding assumptions.

Show a valid comparison including the following cards:

1 Gb and 2 Gb 6950
1 Gb and 2 Gb GTX 560 Ti
1280 mb GTX 570

Single GPU and crossfire / SLI results or stop speculating.

There is absolutely no way that the 2 Gb GTX 560 ti would come close to a single / dual GTX 570 1280 Mb in any game including Bf3, so why bother recommending more Vram when its clearly only the GPU that matters?

Where did say anything about 2GB 560 ti or that it would beat a 570.

You got to stop continually replying to specifics with generalising and implying.
My specifics was proof of variable burden and not that a card with more Vram will absolutely beat a card with less.

If you look at the chart SLI GTX 560 ti 1024mb is already beating the single GTX 570 1280 Mb minimum without AA as soon as full AA is on the single GTX 570 minimums beat the SLI GTX 560 ti 1024mb.

And your link just proves my point that burden is not linear and even more so between different architectures.

 

[bhavv]

I was commenting entirely on what you said here:

The 570 Beats the SLI 560 minimums when with full in game AA because the 570 has a little more vram than the 560 SLI

If thats the case then surely a 2 Gb Vram GTX 560 ti will beat a 570 because it has much more Vram.

Its incredibly wrong to assume that any FPS differences between a GTX 560 ti and GTX 570 are due to Vram, they clearly arent when you are looking at different GPU architectures.

Heres some interesting reading for you:

http://www.bit-tech.net/hardware/graphics/2011/11/08/gainward-geforce-gtx-560-ti-2048mb-phantom/8

Disappointingly, the Phantom posted similar results to other stock-speed cards we've tested. It managed comfortably playable minimum frame rates of 35fps and 32fps in Arma II and Bad Company 2 respectively at 1,920 x 1,080 with 4x AA. In contrast, the Gainward Golden Sample, which is pre-overclocked but has the standard 1GB of memory, was notably quicker at the same Bad Company 2 settings.

No Bf3 tested, but thats not my problem as I didnt write the review (published 8th November 2011).

 

[kissenger]

I'm still not getting involved in the main discussion but I assume you're still talking about the GTX 560 Ti and from what I've seen 30-40% is way off the mark.

Performance Summary

For anyone who doesn't want to click the link it shows 12-17% depending on the resolution.

I think it depends on which games we're talking about and how you define 'better'. If we're talking BF3 multiplayer at ultra and we're looking at minimums as well as averages then 30% is reasonable (40% probably a slight exaggeration), I think. E.g. you will get minimums of 25 on a 2gb Ti but minimums of around 35 on a 570.

 

[kissenger]

I was commenting entirely on what you said here:



If thats the case then surely a 2 Gb Vram GTX 560 to will beat a 570 because it has much more Vram.

Check and mate.

 

[bhavv]

I've looked through 5 pages of google searches now, and cant find a single other review for a 2 Gb GTX 560 ti, and nothing at all on Bf3. The only thing that keeps coming up is the lame EVGA 2 Win dual thing, which is 2 x 1 Gb GTX 560 tis on one PCB.

 

[bhavv]

http://www.overclock.net/t/801683/s...dia-gtx460-is-the-mainstream-card-good-enough

Benchmarks done on a system with 2x1gb 460s vs 2x2gb 460s in the same machine. Vram limiting situations used (high resolutions first, then AA added)

Food for thought.

Thats using surround resolutions, this thread is clearly about 1080p as explained in the OP.

No one here has stated that higher Vram doesnt benefit at higher resolutions.

And why exactly would anyone be running surround resolutions on a pair of mid range cards in the first place? GTX 460 and 560 are strictly 1080p graphics cards, whether you are using one or two. If you game any higher than that, then you need SLI GTX 570s or Xfire 2 Gb 6950s.

 

[Final8y]

I was commenting entirely on what you said here:



If thats the case then surely a 2 Gb Vram GTX 560 to will beat a 570 because it has much more Vram.

Again your not reading and i will be repeating for the 3rd time.
Nowhere am i claiming that a card with more Vram will absolutely beat another more powerful card with less, you really like to put words in peoples mouth.

Its seems that you just cant comprehend shifting variables and see everything as absolute back and white and you treat every comment as so.

And im not going to repeat myself again.

 

[bhavv]

But you very clearly said that a GTX 570 only beats a GTX 560 ti's minimums by an additional 14 FPS simply because it has an extra 256 Mb Vram, theres no going back on that, you said it, I responded to it and quoted it.

If a 2 Gb GTX 560 ti wont beat a 1280 Mb GTX 560 ti 448, then what is the point in buying the first one when they both cost the same?

 

[Final8y]

But you very clearly said that a GTX 570 only beats a GTX 560 ti's minimums by an additional 14 FPS simply because it has an extra 256 Mb Vram, theres no going back on that, you said it, I responded to it and quoted it.

If a 2 Gb GTX 560 ti wont beat a 1280 Mb GTX 560 ti 448, then what is the point in buying the first one?

I said GTX 570 beating SLI 560 because of the 256 Vram and in that specific bench under those variables nothing more and nothing less.
Your reply is under completely different variables.

 

[bhavv]

Then under the same variables which according to your description are due to a Vram limitation, 2 Gb would beat 1.2 Gb!

Simples.

You are wrong in any case where you assume that the difference in performance between two different GPUs is caused by the amount of vram. Vram differences can only ever be fairly compared without any bias on cards with the same GPU / Bitrate.

 

[Final8y]

Then under the same variables which according to your description are due to a Vram limitation, 2 Gb would beat 1.2 Gb!
You are wrong in any case where you assume that the difference in performance between two different GPUs is caused by the amount of vram. Vram differences can only ever be fairly compared without any bias on cards with the same GPU / Bitrate.

Simples.

No that is your problem that its not that simple, because i said already that you must take the GPU Grunt Variable and the Vram Variable into consideration together, its not one or the other black and white.
The SLI 560 has the advantage in that bench with grunt and combined bitrate and still looses in mins with full in game AA to the single 570.

Your truly just wasting time to win an argument and i have better things to do.

 

[Surveyor]

I think it depends on which games we're talking about and how you define 'better'. If we're talking BF3 multiplayer at ultra and we're looking at minimums as well as averages then 30% is reasonable (40% probably a slight exaggeration), I think. E.g. you will get minimums of 25 on a 2gb Ti but minimums of around 35 on a 570.

You didn't mention any specific circumstances for your 30-40%.

The link I provided is far more representative of the difference over a wide range of tested games.

I'm sure you could find specific circumstances where card X is 100% faster than card Y, but it would be entirely misleading to say card X is typically 100% faster than card Y.

 

[andybird123]

Guys bhavv is clearly trolling, he doesnt even have the game he is arguing his cards can play

he is a blind man arguing that the sky is green

 

[tommybhoy]

Attempting to move the goalposts or clouding the situation now?

In regards to BF3@1080p,

Single v single
570>560 ti 1Gb/2Gb
570>1Gb/2Gb 6950
570>6970

Single 570 v SLi
570 min fps>1Gb 560 SLi
2Gb 560 SLI min fps>570

No one is arguing that in a single card scenario 570 is slower than a single 2Gb card.
The argument is about 1Gb 560 SLi falling flat in it's face regarding minimums in the case of ultra@1080p.

Show or make a valid comparison including the following cards:

1 Gb and 2 Gb 6950

Already posted it:

Same architecture/gpu, 1Gbv2Gb, BF3:

Clear effects of vram limitations in BF3!

 

[Admiral Huddy]

Ok I've had a few complaints about this thread.

If anyone feels that any individual is trolling then please don't feed them by engaging or debating with them. Simply ignore and report the post accordingly. If people start quoting and discussing such posts within the thread, it makes any cleanup nearly impossible which is why we tend to remove.



Thanks

 

[Lunawolfy]

Jeez we may as well revive all the harmony METRO2033 threads.

1) 1-1.5Gb is plenty at 1080p

2) If you've got more than 2Gb VRAM then bully for you

3) Nowadays you are rarely going to be crippled by VRAM shortages as the systems are smart enough to compensate, it is slightly more pronounced on SLI/X-Fire. However this is again based on many other factors which may affect it further (SLI 580s on a stock 775 quad with 5400RPM green drives for example)

4) If people in these forums spent the time they spent moaning about VRAM playing games their skill would advance to such a level that you would no longer blame your deaths on VRAM, making the entire argument moot.

 

[bhavv]

Clear effects of vram limitations in BF3!

Not at all, take notice of the scaling above the 2 Gb 6950 on faster GPUs with less Vram.

Going from 17 to 22 min FPS doesnt make the game playable, and its only a difference of 5. Its better to simply buy a 1280 Mb GTX 570 which massively outperforms a 2 Gb 6950 or 560 ti despite having so much less vram.

I've just downloaded the HD texture pack for Skyrim, its running perfectly smooth and lag free with just 1 Gb Vram with ultra settings and 4x AA, 16x AF, so i'm not limited in anyway by Vram at 1920x1200 resolution.

but it would be entirely misleading to say card X is typically 100% faster than card Y.

No, its not misleading at all to state that a GTX 570 will always be faster than a GTX 560 ti. In BF3 the difference is 18 > 31 min FPS, which is a 40% increase.

 

[andybird123]

You dont even have bf3 which basically makes any comment you make on bf3 void