Is it true I would need a 6 core and up CPU to get more speed from my GTX 1080 SLI cards?

no need to upgrade, you would need to overclock though in some situations


that is if your motherboard supports at least 8x 8x pcie 3.0 slots

I only read the conclusion but it seems more like they are referring to the 4.4Ghz clock speed than the 6 cores. Only games that can take advantage of more than 4 threads will gain an advantage from 6 cores and so far there are very few of those around. Your CPU should be fine.

My advice, stick with your current CPU until the day you can't get playable framerates from a game you want to play and worry about it then. It doesn't matter which component in your PC is the bottleneck in a game as long as all of them let you get to smooth framerates. GPU capable of 200 fps in a game and CPU limiting you to 150? so what you can't tell the difference anyway unless you are benchmarking.

Until you start dropping below 60 fps it's not that important. Even 30 fps+ is smooth enough for a lot of games. Then again if you can afford SLI 1080's and want the best that's up to you. I just game at 1080P with a 6700K and a single 980 and find it overkill.

If you look at some of the game benchmarks you should notice the systems hit a CPU bottleneck for them at 1080P and to a lesser extent at WQHD and gain no benefit from SLI 1080's but the 4K benchmarks show a massive gain from the SLI setup because the GPU becomes the bottle neck.

You should have a similar result, your CPU will limit framerates at lower resolutions but at 4K because it's such a massive graphics load it's less likely to. It will vary from game to game though and I still think your CPU should be good enough. Do you overclock at all or is it stock?

If the game is demanding and can take advantage of multiple threads then you can prevent a bottleneck that way, but that's all you can do. If the cores are still too slow, then there's nothing you can do. The GPU's are getting faster not the other way around. Each new CPU generation we see an increase in FPS paired with the same GPU.

No that's of course not true. If a game can't take advantage of the extra threads then it's not going to do anything, in which case a newer architecture and higher clock speed wins.

The CPU they used was overclocked to 4.4Ghz. I think their comments about the speedier CPU refer more to that than the extra 4 cores personally.
As you go up in core count the stock clock speeds go down. The Stock clock speed of a 5960X is 3Ghz and turbo up to 3.5Ghz so 4.4 is a fairly good overclock.

Check this arcticle out http://www.eurogamer.net/articles/digitalfoundry-2016-what-is-the-fastest-gaming-cpu where a 6700K beats the 6 and 8 core CPU's in every game they tested with a single card and lost by less than 5 fps on a couple of titles with SLI. Admittedly it's a small sampling and slower graphics cards but I've not heard of more than 1 or 2 games that get much advantage from more than 4 cores and none that aren't perfectly playable with 4.

By coincidence I watched a video on the 980 Ti and the 6700K, the 4790K, and a six core CPU. All at 4.4GHz. (Paul's hardware on youtube.)

The conclusion was there was about 1-2 fps between the CPUs in benchmarks. Only in 3D Mark Firestrike was the six core CPU ahead. That was purely because the CPU did some graphics or PhysX or something, on the test. (Which normally the GPU would do.)

However you're asking what two 1080s will need. That sort of frame rate is barely seen.

What I think is that the CPU doesn't do the kind of work that the GPUs are dependant on. CPUs only do stuff like manage movement of NPCs etc, put sound to NPCs and so on. They don't factor how fast your GPU will go.

In cases where the two GPUs need to communicate with each other to a larger extent, 2011-3 based boards can net a noticeable gain over 4 core CPUs like the i7-6700. But it’s not really about the extra cores at this point (until Vulcan and dx12 become more of a norm).

Right now it has more to do with the number of PCIe lanes from the CPU. That’s why there is a jump in performance in very demanding graphical settings on high end graphics cards going from even an i7-6800 (28 lanes = 16x+8x) to an i7-6850 (40 lanes = 16x+16x) in SLI. In most occasions PCIe 3 divided in half (8x+8x) over two cards like with i7-6700 is enough. But in very high resolution graphics as well as, presumably, in compatible VR games (soon to come hopefully) there is a noticeable difference given multi card setups. It doesn’t just potentially net higher fps but less stutter as well.


I once read an article while deciding what system to buy given my then new three screen Nvidia surround setup. The article and subsequent tests were about why PCIe speed is enough while at the same time why it really isn’t (yeah, pretty contradictory). The article dealt with the myth that PCIe speed is stellar and doesn’t need any improvement since “no graphics card can make use of that much bandwidth anyway”. The end results as well as conclusion were very interesting and it prompted me to do my own testing based on said article.
So I did a test on a Sandy Bridge 2600k based system using two different mobos (both running PCIe 2). Everything else was equal and only the boards where shifted between tests. One board had to split the 16 PCIe lanes into 8x+8x in SLI while the other board had an NF200 chip which doubles the number of PCIe lanes yielding 16x+16x while at the same time (unfortunatelly) inducing some extra latency between the GPUs.
In all tests running “normal” resolutions the 8x+8x system won by a few fps since the 16x+16x system didn’t see any benefit from the extra PCIe lanes while at the same time experiencing some non wanted latency due to the NF200 doubling chip.

But moving to three screens in 5760x1080 resolution made the more simple and now bandwidth starved 8x+8x system take a considerable performance hit compared to the 16x+16x system. That’s even though the latter was experiencing this extra communication latency due to the NF200 chip. Either way… running this three screen 5760x1080 resolution setup had the 16x+16x system beat the 8x+8x system with anything between 10-20 fps instead of losing against it by 1-3 fps. That’s PCIe bandwidth for you when you truly need it.

I have found my i5-4690 topping out in some games. Gaming at 1080p.

Borderlands: The Pre Sequel (B:TPS) puts a lot of demand on the CPU when the GPU is doing max PhysX. It caused frame rate drop under 60 fps and some stutter.
Assassin's Creed: Syndicate (AC:S) tops out the CPU in intense crowd scenes, and pushed frame rate under 60fps.

I am wondering about buying a 4790K, hoping the extra clock speed and hyper-threading will help me. However I doubt there are many other games right now that will max out a quad just on CPU alone. The Witcher 3 comes close. Crysis 3 can get quite high but usually about 50%.

It's therefor best to keep background app or browsers closed while gaming in these instances.

I only use a single GTX 980. I hope this helps you decide, though it probably doesn't. I figured post my findings anyway. I think B:TPS putting mass load on the CPU might mean one thing. It's best to get the best CPU you can get your paws on. (However some say B:TPS is poorly PhysX optimised.) Folks with i7 and AC:S reported only 50-60% CPU usage.

At the moment the i7 6700k is regarded as the best gaming CPU.

That is for a reason.

You can chose to ignore that or not.