Benchmark Results: Metro 2033

this one trade blows with 6950 2gb and cost about $30 more hm?

nice gpu but it's to expensive compared to 6950...

I just recently replaced an aging 8800 GTS with the 2GB Twin Frozr 560TI card. I have a large case with 7 fans and with a fan profile running the Twin Frozr at 70%, I can overclock to well past 570 performance.

Since I got the card, with a game, for $249, I'm very happy. An hour of MW3 or SC2 at max settings shows a max temp of 53C.

In the battlefield 3 tests, why aren't the testers testing the settings on Ultra? I don't care about settings on high.

thanks for great article, does anyone notice the 6950 1gb vs 2gb give same performance even tho at higher resolution ??

Because none of these cards are fast enough to run on Ultra unless you're going to drop resolution, and nobody buys this class of card to run below 1080p.

We try to make our benchmark settings realistic, not theoretical.

Wow, I hate to see the death of 1920 x 1200 resolution monitor to be replaced by 1080P. But liked this benchmark since I have a 560Ti. Always used Tom's benchmarks to help me decide on video cards.

I'm with nhat11.

I play my BF3 on Ultra settings and 1080p with the 6950 2GB. Ans this is not "theoretical". So if the framerate its 10fps everybody should know.

"the nordics".......skyrim joke please.

I always find it almost shocking that the 6950 1gb and 2gb models have basically identical framerates even at 2560x1600 in all of these super demanding games. Do we really need more than 1gb VRAM? I always think about going triple monitors, and always think my 1gb is going to be a drawback...

it really bugs me when Nvidia does this crap!!!

id like to see if this thing is any better than a Sparkle GTX 560 Ti DF Calibre

i highly doubt it


the other thing that stands out here is AMD's ever increasing performance on the HD 6970!!!

On the Zotac page you write

yet in the conclusion on the same page you write



Either I am misunderstanding this, or one of those are wrong.

I've used a GTX560ti in school with BF3 and it gets mid 50's Celcius with room temp of 60f with A/C.

My favorite part of the review was how min FPS values were included for Batman, AND the comment that the game was unplayable in the first set of tests due unacceptable min values.

Too many times do I see this sort of thing overlooked. Great job!

how the hell were those temperatures so low ?
i mean 34c is in idle mode for my msi r6950 tf3 pe/oc

2560x1600 is 4mp whilst 3 1080p monitors is 6mp, a 50% increase. This makes a significant difference especially if you enable AA options. On a xfire 5850 setup I used to run, several games simply would not run at all (Shogun 2, GTA IV, Crysis come to mind) at 3240x1920, but would run fine if I lowered the resolution. Switching to a 2x 2gb 6950 setup allowed 3240x1920 to run.

Remember, most review sites simply do no do multi monitor reviews. If the cases where they are done, like the below HardOCP article, there are very clear cases where VRAM walls are hit in triple monitor gaming. In this specific case, tri fire 6970 were able to beat tri sli 580's simply because the 580's didn't have sufficient VRAM even with 1.5gb.

Is VRAM an issue at 1080p? No. Don't bother worrying about it. However if you're using multimonitorssetups, it makes a big difference.
http://hardocp.com/article/2011/04 [...] e_review/2

Ok, here are the ingredients:

- Short lived "special" video card
- XMas season
- Option to unlock/OC to the next tier
- Limited quantity

Ok, this might not be in everybody's pleasure to read, but I think it's a sadistic way to get more green juice out of fanbois. This card will be at the GTX570 levels or more (price wise). Supply and demand tell me so.

I don't know if we'll be able to recommend this card at all =/

Cheers!

Why oh why there is never a good low end integrated or at least reasonably priced GPU given for comaprison in the charts. The numbers are quite helpless until I understand how much money I must spend to get N time the performance. People buy card only very rarely - and mostly are coming from low end or at least 2-3 generations back - never from another super card. So at least one low end comparison would be nice to show what this amount of money can do.

You'll have to define reasonably priced. For many enthusiast's a $250 card is a reasonable price. As for integrated cards, no point including them in the test and they simply will not run the majority of these tests. Adding the equivalent of a line saying zero for each of these test's is kinda silly.

To get >0 for integrated cards you'd need much lower settings which are no longer representative of the common settings used by gamers and wouldn't allow the high end GPU to distinguish amongst themselves since they wouldn't be stressed. You'll have shifted bottleneck to CPU and at that point you''ll essentially be looking at a CPU performance graph.