Discussion in 'Reviews' started by W1zzard, Apr 29, 2012.
To read this review go to: http://www.techpowerup.com/reviews/Intel/Ivy_Bridge_PCI-Express_Scaling/
Why don't you test SLI/Crossfire?
They require more bandwight from PCIe and you can see real difference between PCIe 3.0 and 2.0 in big resolution like 5760x1080
we'll see about that. given the data in this article i seriously doubt higher resolution in multi gpu will need more bandwidth.
i'm planning to do multi gpu testing once hd 7990 is out (using 680 sli, 690, 7970 cf, 7990)
multi-gpu is only a minority of users so i thought it would be more useful to test single card first
What about compute benchmarks? I saw a couple that showed quite a big difference between PCI-E 2.0 16x and PCI-E 3.0 16x.
I always thought the whole PCI-Ex 3.0 'double the bandwidth' thing was more of a gimmick just to sell more boards then anything else.
Oh well.... since IB is not such a big leap in performance from SB and PCI-E 3.0 being nothing but a waste of time, theres really no reason to upgrade to IB from a decent SB set up and this review just proves it.
does GTX690 on PCI-E 2.0 x8 will be similar like 2 GTX680 each on 2.0 x4? ,assume performance of GTX690=2x GTX680 in SLI
Thanks for the review. Would it be possible to test minimum frame rates (drops) in the next article about scaling with interface speed? Thanks.
Would be nice if you could add x1 performance as well (Maybe just a few of them) at http://forum.notebookreview.com/gaming-software-graphics-cards/418851-diy-egpu-experiences.html we are interested in x1 performance until Thunderbolt gear is ready for external GPUs for our laptops
Thank you for the hard work, it sure is a hassle to test it in detail.
this is really great review wizzard
Just goes out to show that PCIe 3.0 is pure marketing. Maybe it'll have more of a difference when you run quad-SLI/Xfire with 8x/8x on PCIe 2.0 vs. PCIe 3.0.
It will be usefull, not now, but later and would you rather have it beforehand or have to wait for it and therefore limiting your GPU? Another field that will benefit is PCIe based SSD cards, these are growing massively in speed
This review proves once again, AMD>NVIDIA when it comes to lower PCIE bandwidth
I wonder why you guys think its pure marketing gimmick, it delivered on its promise to double bandwidth. Psersonally I would rather them release 3.0 before we hit a bottleneck on the 2.0 (and thus showing you no performance advantage right now between the two) rather than wait until 2.0 becomes a bottleneck before making 3.0 a standard (where they actually show a difference between the two during launch day). They can release it slightly later, so people will not feel "cheated" but why release something tommorow when you can do it today?
Outstanding review W1zz! I don't think people understand how much time you spend on your reviews. Hell reviewing one card would be a pain in the ass Seems like 3.0 is not worth the worry at all.
Excellent review, greatly appreciated!
No one can afford those, so it's pretty irrelevant.
There is another way to look at it... PCIe 3.0 means you can ditch x16 slots. We can start making cheaper MBs, cheaper sockets, and cheaper GPUs by only needing x4 or x8.
For mainstream, 3.0 x4 is all that is needed. For enthusiast, 3.0 x8 is more than enough. The extra 1% performance for x8 to x16 is far outclassed by the next GPU upgrade, or a better CPU etc.
gpu compute seems to be a waste of time. video encoders have laughable quality. are there any other applications that anyone uses?
Some seriously questionable comments made here.
AMD performs better at the new spec than nVIDIA.
PCIe 3.0 is marketing.
Make boards with x8 instead of x16 to save money on motherboards and because PCIe 3.0 isn't saturated.
love it. everything else seems professional until that crops up.
Amazing review to have - thanks!
Good to know that PCI-e 3.0 is just for future tech or for crazy multi-card configs on non-enthusiast motherboards right now.
wow....what a joke....seems like a real difference is still years away....i cant believe the pcie 1.1 numbers....holy shit
2000 individual test awesome
i think it will be useful later
Separate names with a comma.