Archives Discussions

[Skysnake]

Yeah, and a HD7970 is something about 5% faster with PCI-E 3.0 16x than with PCI-E 2.0 16x in a wide range gamebenchmark... :rollleyes:

So what do you want to say?

[avk]

You mean - in the real games? Well, I'd like to see your proof, then .

[Skysnake]

I was not abel to find the test i remember, but i have the luck, that there is a new PCI-E 2.0 vs PCI-E 3.0 test out there.

So just look at the Benchmarks for the DX11 Games.

Battlefield 3, Dirt Showdon and Project Cars benefit slightly from PCI-E 3.0 even with 16x. With only 8x in Multi-GPU-Setups the benefit is bigger.

http://www.computerbase.de/artikel/grafikkarten/2013/bericht-pcie-3.0-vs.-pcie-2.0/

[avk]

Ok, what do we have here (2560*1440, 4xAA, 16xAA) ?:

1) Alan Wake: 42.0 vs 41.8 fps, i.e. 0.47%;

2) Anno 2070: 38.0 vs 37.2 fps, i.e. 0.53%;

3) Battlefield 3: 37.8 vs 37.8 fps, i.e. 0.00%;

4) Dirt Showdown: 55.9 vs 53.7 fps, i.e. 4,09%;

5) Dragon Age 2: 28.3 vs 28.3 fps, 0.00%;

The only noticeable gain we can see is in Dirt Showdown (4,09%), whereas all other games show almost nothing from exchange PCI 2.0 to 3.0. Ergo, PCI 2.0 is sufficient for a single high-end class GPU like Radeon HD 7970, whereas PCI 3.0 can show its real (hmm...) advantage (within 5.5%) in Crossfire/SLI modes only.

[brutalix]

Actually HARDOCP found it to be in the order for 5-10 %  diff between pcie 2.0 and 3.0. and this was without CFX or SLI, but in a single card configuration.  But here one can object that the test was done on two different plattforms, ivy brigde and sandy brigde.

http://m.hardocp.com/article/2012/07/18/pci_express_20_vs_30_gpu_gaming_performance_review/2

Kind regards

Brutalix

Your friendly linux neighbor.

[avk]

Different CPUs - different results .