[H] Tests Crossfire V SLI with Mantle

[VulgarDisplay]

Mantle itself. Read the entire article.

What do you mean by lower end? I'm using a mid range 7970 is and it's great.

 

[el etro]

Amazing how many people are now interested in 1080p/0xAA benchmarks...

Because is much more CPU bound. Mantle effectiveness is in CPU-bound situations.
And 1080p benches don't need to be made with a 290x. Put a HD 7870 on a i3 and test at the setting.

 

[frozentundra123456]

What do you mean by lower end? I'm using a mid range 7970 is and it's great.

Is "read the entire article" somehow ambiguous? It specifically states which cards they tested that did not work properly with mantle, and gave better results with D3D.

 

[krumme]

H testing methology is, unlike most other situations when they test gfx, not very good here.

They test for around the average of 60fps. On a 4.6ghz ib.

Its to low average fps for my taste as the min go to low then. And a bit to high ghz on the cpu compared to what i and most other use.

Others might have it differently or oposite.

But in this case they run into a situation where gpu and cpu frame to frame time is about the same at 60fps on the maps they test.

What that means is if you eg want 70fps as an average and have a 4.2ghz cpu then suddenly mantle is 30% faster than dx on the same maps and using the same testing pattern. Its imho a far more realistic scenario for a compettitive fps like bf4.

So h is as much testing their own preferences and oc than mantle itselves.

I tend to prefer higher than 60fps on average using an i5 ib at 4ghz and on the 64 player maps there is a huge difference. People just need to know their own preferences.

As a sidenote. I completely agree with h that mantle should just have been postponed a month until at least the second beta driver. And bf4 itself should have been released a month later. There is no need for all that rush, the engine is bounds a leaps beyond anything else. It could wait half a year if that was needed. Mantle have been in work for several years. A month is nothing.