TechReport Multi-GPU RoundUp - SLI vs. CFX : GeForce Tends To Be Better

[caveman-jim]

Techreports' Scott Wasson compares a multitude of titles and cards to give a more comprehensive picture of the value and performance of the DirectX 11 cards currently on sale:

The latest GeForces tend to be better values at present than AMD's 5000-series Radeons. That's a clear reversal of fortunes since the GeForce GTX 400 series' somewhat underwhelming debut, when the GTX 480 was really no faster than the Radeon HD 5870. The Fermi architecture still quite new, and Nvidia has extracted enough additional performance through driver tuning to put the GTX 470 on equal footing with the 5870—and the GTX 480 well ahead. That's true across a range of games, not just those where the GeForces have an apparent advantage, like Metro 2033 and Borderlands. The addition of the cheaper and more architecturally efficient GeForce GTX 460 has further solidified Nvidia's value leadership.

As for the question of the hour, yes, a multi-GPU solution can provide better value than a higher-end single card. Dual GeForce GTX 460s of either flavor, 768MB or 1GB, will set you back less cash and deliver higher frame rates than the GeForce GTX 480 AMP!, as the scatter plot shows. In fact, the 460 SLI options are among the best values we tested.

Read the full article @ Techreport.com

[noko]

Makes me want to double up on my 5870 except that would require new MB, CPU, Ram, OS .

Now the results where way different then HardOCP, well in this case it appears to be a much more thorough review at TechReport. In anycase the 5870 in CFX toasted the 460 SLI in most things but not all things as in Metro2033 where the GFX are just plain awesome. I thought it was interesting how well 3 5770s did in CFX, I am getting the impression that ATi 3 card setups are way faster then Nvidia 3 card setups, that will be an interesting thing to look at.

Best bang for buck has to go to two 460 1gb in SLI, they beat any single GPU card out there significantly at at times, the 3 CFX 5770 maybe next but what a hassle maybe that will be, I wouldn't know though since I never done any multiple graphic card setups.

[jlpktnst]

How many people will buy single geforces now just because of this test... even though amd still has better bang/buck in all categories but the 200€ price range. Meh.

Duno, 0.1% of gamers on sli/cfx?

I will never understand why comparison of fiesta turbo vs yaris turbo matters when you are buying non-turbo version?

[spajdr]

While 5870CFX is faster, there is no point to buy 5870 in CFX now, due to high price difference of two GTX460 (when when you add few more dollars you get factory OCed version and difference is then zero to 5870 CFX, but high price difference stays).

[kaarel]

I would like to see also a chart with power draw/perf. Current summer made me very aware how important is heat output when gaming.

[Neoauld]

I would like to see also a chart with power draw/perf. Current summer made me very aware how important is heat output when gaming.

yup, and the extra power drain lowers its value to
plus, with summer here, ive had to use my AC a bit more with my new PC while gaming, imagine 2 Nvidia cards..ouch

Though, in winter id probably want them nvidias :P

[DaNdo]

I will never understand why comparison of fiesta turbo vs yaris turbo matters when you are buying non-turbo version?

me neither.
But, perhaps it's just our balkan's mentality speaking.

[Dr. Zhivago]

When it comes to hardware reviews, I think Scott Wasson is the most reliable, thorough, unbiased, skilled, and entertaining reviewer out there.

When reviews such as the one from [H] come out, I roll my eyes and move along. I don't give [H] any credence at all really. I think they are biased and don't do a good job of hiding their biases. For instance, they cherry picked 3 games where nVidia does extremely well to prove their point and said, "See? Told ya so!" Plus, they write like 7th graders...

Scott used a much wider variety of games, logged all kinds of detailed data, provided easy to read charts and graphs to back up his analysis and peppers the review with humor and well founded insights and acknowledges where the shortcomings and strengths of each product are.

[H] does none of this...

My .02

[Hapatingjaky]

The way I look at it now is the following:

1.) HardOCP has the Galaxy Advertising plasterd all over its site. With no ATI advertising.

2.) Techreport has XFX and Asus ATI variant's plastered all over its site. With no Nvidia Advertising.

We've all seen this in the past.

Its a good read from both sites but after reading the techreport site, I find it interesting they would use a driver thats not even ledgit from Nvidia and a driver thats a release old for ATI. I agree that when someone purchases a card from either Nvidia or ATI the first thing most consumers would do is use the driver on the CD or go to the companies website and download the latest driver from them. Only a handful like us will browse forums and find the best suitable driver to use.

Each review is biased, better to just read the forums and find out what works best.

[@md_Guy]

While 5870CFX is faster, there is no point to buy 5870 in CFX now, due to high price difference of two GTX460 (when when you add few more dollars you get factory OCed version and difference is then zero to 5870 CFX, but high price difference stays).

Actually I would say there's no point in buying either the GTX460/470 or the 5850/5870 at the moment given the upcoming price reductions and subsequent HD6000 release. IF you MUST have something soon, wait for the price drops.. if you can wait 4 months get the HD6000 or if one is a total fanboi of green, wait for Nov and the subsequent GTX480 price drops.

[caveman-jim]

Each review is biased, better to just read the forums and find out what works best.

Right, because no users are biased or unrepresentative under the cover of anonymity...

The actual answer is to read many reviews, and many opinions, and see if you can conclude for yourself where it fits.

[Dr. Zhivago]

I use ad-blocking software. I don't see ads. Ever.

I just know that from years of reading Scott's reviews that there isn't any funny business going on.

[EmpTe]

I wish more sites would use average AND minimum frames per second as the basis of assessing performance - a high average is pointless if the minimum is unplayable or noticeably affects the gameplay. Top marks to them for a more in-depth article.

[Riou]

I wish more sites would use average AND minimum frames per second as the basis of assessing performance - a high average is pointless if the minimum is unplayable or noticeably affects the gameplay. Top marks to them for a more in-depth article.

It depends where the the minimum occurs though. Minimums that only occur at the beginning of a level once is negligible. However if the frame rate is constantly dipping back towards the absolute minimum, that is bad.

I am not a fan of just posting minimums because it does not tell you much about the distribution of frame rates. A Frame rate vs. Time plot graph gives more information.

[nycdarkness]

Looking at xfire scaling is just depressing.