Why isin't gtx 690 on the top?

[Monte Cristo]

I was looking at benchmarks and noticed that gtx 680 is better than gtx 690, why?

[David (PassMark)]

The GTX 690 is a dual GPU card. Which means it uses SLI.
But SLI only works well in a limited number of applications (mostly the AAA games).

[Monte Cristo]

Now i understand, but its preety mistleading becouse people will think the gtx 680 is alot faster

[David (PassMark)]

It is faster with some applications.
It is only faster when used in SLI mode with some AAA games, that are optimized for SLI in the drivers. And even then SLI only comes into play in some configurations (e.g. very high resolutions with a fast CPU)

[fayo]

It is precisely the people looking for this type of cards to play this type of games, I think than GTX 690 It should be on the top rank, as said Monte Cristo, the ranking is misleading at the time of purchase or choose.
And, I can´t belive than GTX 690 only got 4,843 pts in 3D Mark, really?, this used only one GPU?, Greetings from Mexico DF

[David (PassMark)]

To a large degree it is up to the device driver to use both GPUs in a dual GPU card. We only have limited control over it.

There is some deeper analysis here,
http://www.passmark.com/forum/showth...ormanceTest-V8

As noted in the linked post, manually re-configuring your SLI device driver setting can give better results for some the 3D tests. If you aren't happy with this, then please contact nVidia and ask they fix up their drivers. The AMD crossfire drivers do a significantly better job.

[fayo]

Thanks, and what do you think about Asus Ares II card?, this one has 2 gpu's ati 7970 (crossfire), is a monster like GTX 690 too, do you going to include in your rank?, I guess that is the same case isn't it?

And it is not possible to include a score with both GPU's working?, For a better and fair evaluation of the multi-GPU cards or, make a rank for multi-GPU cards.... It is only an idea or suggestion

[David (PassMark)]

Any cards people submit get included in the charts. The 7970 card has been included in the charts for a while now. The 7970 is a popular high end card.

Our preference is always for single GPU cards. The multiCPU cards have a number of issues, like the need for profiles in device drivers, incompatibility with some applications and micro stuttering. Not to mention the extra power noise and heat from SLI / Crossfire setups. Other reviewers, like Anand over at AnandTech make more or less the same points as well.

[wittzi]

Any cards people submit get included in the charts. The 7970 card has been included in the charts for a while now. The 7970 is a popular high end card.

Our preference is always for single GPU cards. The multiCPU cards have a number of issues, like the need for profiles in device drivers, incompatibility with some applications and micro stuttering. Not to mention the extra power noise and heat from SLI / Crossfire setups. Other reviewers, like Anand over at AnandTech make more or less the same points as well.

This is just misleading rubbish. YOUR preference is single GPU cards, however the world has moved on and in 2015 most decent games support them. Those that don't should receive poor reviews because they've been poorly written, or are games that do not require powerful GPU's.

Seeing some of the GPU's and their position in your chart is a joke. The obvious one that stands out is the 295X2. The fact that it's an enthusiast's card and not all games support Crossfire (but it's 2015 - almost all do) should have no bearing on a bassline benchmark test.

[David (PassMark)]

Those that don't should receive poor reviews because they've been poorly written

Clearly you have never had the job of writing any 3D code. To a large degree SLI/Crossfire optimisation is it out of control of the developer.

If you have a look in the release notes for the video card device driver releases, you'll see that AMD and nVidia are constantly modifying their drivers to work with the games. This shouldn't be the way the world works, but unfortunately it is. It would be better if the driver was stable and exposed all the functionality of the hardware for the developer to use. AMD and nVidia only bother fixing their drivers for the big name games. Other 3D software doesn't get much attention.

Having said that, the 295X2 result is low. Other crossfire setups did better in the past. So I suspect there is a driver issue holding back the performance. We'll contact AMD about it (as it is really out of our control).

This might change in DirectX12, as the developer will get more visibility of the hardware available and control over what functions run on which cards.