PCIe Lanes – PCIe 8x vs 16x in SLI

Reading 1 Minute

PCIe lanes have, surprisingly enough, a pretty big impact on CPU prices. So, is there a tangible benefit to purchasing a CPU with more lanes when running, let’s say, 4 NVIDIA GTX 980s in SLI?

Detailed in-game settings:

Sponsor link:

Support us:

Join our community forum:

Intro Screen Music Credit: Adhesive Wombat – Check out his channel here:

Outro Screen Music Credit: Approaching Nirvana – Sugar High

31 Replies to “PCIe Lanes – PCIe 8x vs 16x in SLI”

  1. Enlightening, thank you, what really matter is GPU processing power, dropping from 16 to 8 is marginal. I hope AMD and Intel someday give us the possibility to eliminate the bridge and just communicate through the PCI-e lanes and to do this without degrading the performance I'd say give support for 10-12 lanes per video card with 4 video cards on the PC AND have USB and M.2 drives working with additional PCI-e lanes and better PCI-e raiser cards, of course would be optimal to have this platform with each video card at 16X not only to process in parallel but to co-process, sharing resources, for efficiency, having 4 copies of a game or 3D app on 4 video cards is a big waste IMO

  2. I've been trying to tell people this for years since 3.0 came out. That even though you may only have a certain number of lanes, cutting the lanes in half or by 3 doesn't really mess with performance; that yes the CPU does drive only a certain number of lanes but the SLI/Crossfire tech makes up for on the backend .. and I'm glad you finally proved that…

  3. Did you not use the M.2. slot during these tests? Or did you use a regular sata ssd with Windows?
    Because I am in doubt whether I would do the the same, but then with GTX 1080 videocards. But I also use an m.2 ssd which takes 4 PCI-e lanes. Now, there are still 8 lanes left, but I am not even sure whether USB 3.0 ports or the SATA-Express ports take some of those lanes on my motherboard. So, I am in serious doubt whether to buy a second GTX 1080.

  4. Average FPS really doesn't tell me much about the performance of the different cards in different configurations. As a gamer, I'm more concerned with what the outcome will be when multiple particle effects happen in a 3D world that create heavy FPS drop. This is where I think the extra graphics cards come into play, when max performance is needed to ensure no lag. We know average FPS will be about the same because there arent many situations where large particle effects happen in games that really test how good a GPU is when it needs to perform at its best to ensure smooth game play. I wonder if some of the different configurations can handle that hardcore frame rate hit better than others, because what you demonstrated is fairly meaningless in terms of performance when consolidated into an average. That 1% difference in average FPS can be extremely more meaningful at the right moment if one SLI configuration is at 30 FPS while another is at 60 FPS when an explosion is happening in a game and you're trying to snipe someone.

  5. I have the 5820k processor and have since it came out a few years ago and it still is beast mode! The best thing I think personally is that it has MASSIVE OC potential, i have mine currently running stable at 4.7Ghz @ 1.32Volts & my temps dont go above 65C' on load. Not just a lucky lotto winner either, its actually quite typical of the 5820K to have said 'headroom', not so much is said for the more expensive 5930k bahahahaha