Thanks for the review. I can't see myself getting 9xx SLI setup. The 4k results are disappointing, and below that point a single card is more than enough. So if one's gaming on 2560x1440 or 1920x1080, a single 970 or 980 should be a nice card to get. 4k - we'll have to wait for the next offerings from NV or AMD.

Thanks for the review. I can't see myself getting 9xx SLI setup. The 4k results are disappointing, and below that point a single card is more than enough. So if one's gaming on 2560x1440 or 1920x1080, a single 970 or 980 should be a nice card to get. 4k - we'll have to wait for the next offerings from NV or AMD.

Depends on your monitor, those of us with Korean 1440p overclockable monitors would disagree (or pretty much any 120/144hz monitor). Mine will do 118hz but the 7970 can't keep up. Now if only the wife would let me drop some $$$ on a new build.......

Thanks for the review Hilbert.

Depends on your monitor, those of us with Korean 1440p overclockable monitors would disagree (or pretty much any 120/144hz monitor). Mine will do 118hz but the 7970 can't keep up. Now if only the wife would let me drop some $$$ on a new build.......

That's a good point. Thank you for this valuable comment. Now when I think about it, 970/980 SLI would be great for ASUS Rog Swift... dang... is it time to change my monitor?

That's a good point. Thank you for this valuable comment. Now when I think about it, 970/980 SLI would be great for ASUS Rog Swift... dang... is it time to change my monitor?

It would certainly help driving 1440p / higher refresh rates, at the moment I can pretty much only run @ 118hz on the desktop / light games, anything else is too much.

Thanks for the review Hilbert. Those 4K benchmarks are looking pretty good. Hopefully in the next 2 years or so, we can achieve those same numbers off 1 card instead of 3

Cool my 290 tri x crossfire can keep up with the 980 sli but for 4k we need to wait for the 20 nm nod big maxwell or GCN 2.0

Hmmm, I might end up going 2x970s in the end for my 1440p monitor after reading this, but then again I still want to wait for the 980 Lightning and then decide if it is worth it.

Thanks for the review, guess i wont be running 3x 1440p yet in newer games, getting close though. Looks good for my decision to get 2x 970 though, except seems to occasionally drop off more at 4k compared to the 980 for some reason. Also on page 2 the load temperature bar graph, I see 680 2x sli is highlighted instead of the single 980?

Also on page 2 the load temperature bar graph, I see 680 2x sli is highlighted instead of the single 980?

Ah fixed, thanks.

Hi Hilbert if possible could you do the 970`s in 3 way sil bench please.

Very good scaling on 3way sli on the "good" engines 🙂 Id like to see how 4way scales on frostbite and cryengine.

Hi Hilbert if possible could you do the 970`s in 3 way sil bench please.

Not possible, the initial cards are returned and the other I have are all AIB cards with different coolers , e.g. the 3x SLI connector will not fit with three difference boards.

Another excellent article, well done Hilbert ! Iam now even more excited for my two 970's to arrive sometime in the next two weeks knowing that I'll be able to play bf4 on 1080p with 150 fps and above 🙂 Oh and do you plan on doing benchmarks with the upcoming games like shadow of mordor and evil within too? I don't know if evil within will have a 60 fps lock like both of the previus games on id5 had and thats why people who do benchmarks don't use either of those games for tests and benchmarking.

Nice review, thanks for doing both 3way and 2way gaming! I still think I'll go 980SLI rather than 970, because of the horsepower behind it. And probably for the TitanX, or 980Ti... 🙄

Awesome review, thanks once again! I wonder if the FCAT measurement would change if you use vSync or even gSync. Could it be due to a cpu limitation or any other IO limitation? These frame drops look similar to the IO bottleneck you had here: http://www.guru3d.com/articles-pages/fcat-benchmarking-review,6.html At 5 ms per frame does that mean there is more data to be processed, or is the data stream always constant no matter the frame time?

Nice review, thanks for doing both 3way and 2way gaming! I still think I'll go 980SLI rather than 970, because of the horsepower behind it. And probably for the TitanX, or 980Ti... 🙄

If you have the money, then it's a nobrainer, quite frankly, Iam going for a two 970's because of the excellent price they go for 😉 Oh and I also read somewhere that there will most likely be a 960 and 990 before 980Ti will be released.

I also wonder how the benchmarks in games are performed. For example the Battlefield 4 time demo, do you control the player yourself each time or is there some prerecorded demo you can replay?

All of these AMD Gaming Evolved titles on a 3-way Nvidia SLI setup can't be giving good results. Using Tomb Raider and Thief for FCAT results is pretty much a worst case scenario for these cards. Even so, despite the fact that 7 of the 8 gaming benchmarks are Gaming Evolved titles, it's interesting that the one Nvidia game, Metro: LL was the one that didn't work in 3-way SLI.

Aah, a quick word on AMD Gaming Evolved titles. Its not that I choose these deliberately, but most triple A titles have AMD plastered all over them these days. Secondly, I try to use scripted benchmarks as much as I can for the best measurements. So that's Metro, Hitman, Tomb Raider, BioShock, Thief and then the two 3DMark's already. I would like to exclude it, but have to BF4 as everybody wants to see the results, I throw in MOH, Crysis 3 and Metro to balance things a little. I ditched Ubisoft titles for their DRM issues and the impossible to use UbiPlay. EA / Origin is getting on my nerves as well (an update each day these days) but a necessity. next to that I try and use as much Steam titles as I can use the games on multiple machines, that saves this office a lot of cash opposed to re-purchasing the game for every test system. I will look out for a couple of pre-NV games as well, but I just hate Batman, I purchased Titanfall but that was online only ... well and so on. AMD is all over the AAA games, NVIDIA not that much. So the Gaming Evolved titles are not a deliberate choice, it's just the most popular AAA games that people use + they need have ability to be scripted or measured with FRAPS/FCAT in a fast enough session (benchmarking takes a lot of time).