RTX 3080 on PCI-e 3.0 (x16 vs x8)

[JOSHSKORN]

I have a Rampage V Extreme that I want to keep for awhile still, but upgrade my GTX 980 to the RTX 3080. It's got PCI-E 3.0, and obviously the 3080 is meant for PCI-E 4.0. I read somewhere there's literally a couple of percentage points difference (Gen 3 being lower).

Have there been any benchmarks on the 3080 in terms of the slot types (x8 vs x16)? Basically, I have a Noctua NH-U14S which is basically covering my top x16 slot, so I'm wondering how much more of a loss I would take using the x8 slots.

Anyone with knowledge on this specific board? The ASUS Forums in this section are virtually dead and no help. The board IS six years old.

I basically am looking to upgrade the GPU without having to upgrade everything else, which still works fine. Vertical GPU mounting is possibly an option, if I absolutely have to use the top PCI-E slot.

 

[defaultluser]

You will lose 5% performance. Not a big deal.

PCIe 3.0 running at x8 = PCIe 2.0 running at x16, so use the red line.

 

[Ready4Dis]

Some games see almost no difference, some games see a bit more. The average as mentioned was a few % points. If you have some specific games, they did a good amount of testing honestly.

First game starts here: https://www.techpowerup.com/review/nvidia-geforce-rtx-3080-pci-express-scaling/3.html

Last game is on page 25, and average/summary (as seen above) on page 26/27. It also compares 1080/1440/2160. The one posted above is for 1080 and is the largest gap. It closes at higher resolutions since there is less commands/traffic being sent across PCIE bus due to lower frame rates (more time spent rendering, less time sending commands).

 

[Hakaba]

GN covered this a day or two ago.



BLUF: You are 99.9% not going to notice a difference between Gen3/4. Well outside of benchmarks.

 

[JOSHSKORN]

Some games see almost no difference, some games see a bit more. The average as mentioned was a few % points. If you have some specific games, they did a good amount of testing honestly.

First game starts here: https://www.techpowerup.com/review/nvidia-geforce-rtx-3080-pci-express-scaling/3.html

Last game is on page 25, and average/summary (as seen above) on page 26/27. It also compares 1080/1440/2160. The one posted above is for 1080 and is the largest gap. It closes at higher resolutions since there is less commands/traffic being sent across PCIE bus due to lower frame rates (more time spent rendering, less time sending commands).

And of course, they skip the one gaming review I'm looking for, which is Microsoft Flight Simulator 2020 for the 3080. Have you seen a review of the 3080 with that game (ok, "Simulator") ?

 

[Ready4Dis]

And of course, they skip the one gaming review I'm looking for, which is Microsoft Flight Simulator 2020 for the 3080. Have you seen a review of the 3080 with that game (ok, "Simulator") ?

I honestly haven't tried to find them, sorry. It uses dx11 currently which sends a lot of commands through the pcie bus (compared to dx12/vulkan) and is inherently single threaded, they are supposed to be switching to dx12 at some point though, which should reduce pcie load. What resolution are you playing at? I would imagine it's be similar to other dx11 games @ 1080p, around 5-10%% difference. 1440p would probably be only 2-4% while 4k you wouldn't even notice (maybe 0.5-1%). These are just my guesses as I don't have any real data to go by for this game, sorry I can only suggest these guess without knowing for sure.

Ps. The only thing that has me a little concerned is the live streaming/updating as that requires "constant" data through pcie, so maybe 10% as a safe guess @1080p, 5-6% @1440 and still 1% for 4k.