• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

GPU Running with PCIe x8 instead of x16 help

Duz

New Member
Joined
Sep 14, 2022
Messages
4 (0.00/day)
Hello, recently noticed that for some reason my GPU was running with PCIe x4, tried older drivers (I was running beta drivers thought it could be that) now it's "stuck" at x8 even under load doesn't get to x16, if anyone knows what the issue might be any help would be appreciated

Btw, I'm pretty sure it's on the correct slot and I do not have any NVMe on the M2_1 slot or any at all for that matter.
1663176036723.png
 
system specs for the SSD/HDD, CPU & Mobo? Also anything else in your other PCIe slots?
 
Last edited:
What CPU/MOBO do you have? If for example you have a 3200g your card would be stuck in X8 mode. We need more details before we can troubleshoot.
 
2 reasons, your cpu doesn't support it or you have a card in the pcie slot that shares lanes with the gpu.
 
For those asking the specs (PC has a few years but I just noticed this now, pretty sure it wasn't doing that before... I think, and like I said when I used another set of drivers it went from x4 to x8):

MOBO: Asus b350-f gaming
CPU: Ryzen 5 1600
SSD: Samsung evo 860

And no the other PCI slots are empty.

NOTE: If I check in the BIOS it says x8 Native, I can share a pic if needed, doesn't really say anything else.
 
I would try reseating the card in the pcie socket, maybe dusting the socket out a bit with some compressed air or similar.
 
I'd also reseat the CPU, since that is what provides the PCIe lanes to the GPU.
 
I'd also reseat the CPU, since that is what provides the PCIe lanes to the GPU.
That'd be unnecessary as if it has a cooler it should have sufficient pressure on it.
 
I guess reseating the GPU and cleaning the PCIe socket is my next option. Will do that tomorrow and post an update.
 
That'd be unnecessary as if it has a cooler it should have sufficient pressure on it.

It's precisely because of the pressure that you would want to reseat it, too much pressure can cause some pins to disconnect. This is rare, but it can occur, especially on cheaper motherboards with a lower quality PCB/socket.
 
It's precisely because of the pressure that you would want to reseat it, too much pressure can cause some pins to disconnect. This is rare, but it can occur, especially on cheaper motherboards with a lower quality PCB/socket.
I doubt pins on AMD AM4 CPUs can "disconnect" from too much socket pressure.
CPU is simply pushed more into plastic after all.
Maybe traces that go to the socket can be bend, but that would require a certain pressure in localized area (where PCIe pins are) for OP's problem to occur.
 
Reseating is not the solution imho. Try to look at bios menu and sub menus
 
Check for damage near the connector area, broken off components
 
are you using a riser cable ?.
 
Bios up to date? a quick check of its support page indicates this board has had 28 bios updates since release.
 
This shouldn't be a problem, but I believe the two SATA ports up near the RAM are pulling off of the CPU so if you're using those instead of the ones towards the bottom of the board perhaps that's where the conflict is.

FWIW the x8 vs x16 shouldn't make any performance difference with that card since your system is running it in PCIe 3.0 and it doesn't need the extra bandwidth.
 
Anything in this slot highlighted pink forces the top slot to 8x mode, even a 1x card

1663236137337.png


Top slot is 16x
Middle slot is 8x - but drops top slot to 8x when in use
bottom slot is x4 2.0

The bottom x4 slot if used, will disable two x1 slots
Using NVME or SATA drives in the M.2 port will disable SATA 5 and 6
 
Hello everyone, update: I guess reseating the GPU and cleaning the PCIe slot did the trick, finally running at x16
Thanks everyone for the help and good tips, I hope I don't have to reseat it again in a while hahaha.
1663240542833.png


@thestryker6 It might be placebo, but I noticed fewer stutters in some games, then again I might be imagining things
 
Stutters are another issue, and theres a lot of common fixes.


Obviously, physical things in your PC can cause it. Anything overheating and thermal throttling, mechanical drives being slow (or even just spinning up/down and stalling the system), running out of RAM

Then you get into software (quit everything you can... ugh goddamned wallpaper engine) and work on driver/software/game settings... all i can suggest there is the usual FPS cap 3FPS below your Vsync, or an FPS cap at double your refresh rate without Vsync if you can handle the tearing (I'm not confident your RX580 can achieve this in modern titles)

Long story short, 60Hz = 16.67ms each frame and triple buffering can have your CPU with three frames ready before the GPU is, meaning you can get 50ms old images every time you have CPU power to spare with a maxed out GPU. 3 FPS before the limit prevents that weirdness from happening and lets you game tearing free with Vsync off, and i find this advice tends to resolve a lot of peoples input lag issues they perceive as stutter.
 
I doubt pins on AMD AM4 CPUs can "disconnect" from too much socket pressure.
CPU is simply pushed more into plastic after all.
Maybe traces that go to the socket can be bend, but that would require a certain pressure in localized area (where PCIe pins are) for OP's problem to occur.

It can happen, and even worse: with PGA designs, overpressure can cause pins to break inside the socket. I killed my X6 1090T that way, as I attempted to install a heatsink without the "support" around the socket... and tightened it a bit too much in the process. Yes, of course, this is my fault and my fault alone, but it's a possibility... and don't try to work around the mounting support bracket, it's not worth it :P
 
Last edited:
It's precisely because of the pressure that you would want to reseat it, too much pressure can cause some pins to disconnect. This is rare, but it can occur, especially on cheaper motherboards with a lower quality PCB/socket.
LoL it happens on very expenssive boards very often as well, it depends on your skill to mount the coolers :D
Most think it has to be screwed tight and in most case that turns out to be too much
To be honest you do not need X16 as most cards today are not even able to push X8 to it's limits
Yes not even the latest toys on the market.... those who think X16 are wrong.
Else buy a very expenssive workstation board then you will have real X16 with PRO cards and you can push the PCI slots to it's limits
Never ever think you need X16 as home user as the gpu will never use more than X8
 
Last edited:
@thestryker6 It might be placebo, but I noticed fewer stutters in some games, then again I might be imagining things
With your solution being a physical one I could easily see that making a difference, because bad contact can easily lead to all sorts of issues and I'm glad you got it resolved. I had my own x8/x16 issue in the past and I definitely kept going until I figured out what it was even if the performance wouldn't matter because who knows what else the culprit could be impacting.
 
i find hell of a difference between pci 3 x8 and x16 with VR games, at x8 they stutter, screen Tare and just run crap i do play heavy games though the link cable mainly open world stuff like no mans sky, fallout vr, skyrim VR with around 100 mod ive only a 2080 but dos its best. glad you have sorted your x16 out i had a duff riser cable that caused mine to run in x8 for a while.
 
On the pins breaking, the socket cant do it - i've got a detached one from ali express and they 'pass through' with room to spare
(I use it when lapping AM4 CPU's, and pins dont come out the back or anything like that)

Unlike on intel this doesnt lead to a situation where pins can avoid contact, but simple dust, thermal paste or random cat hair certainly can - reseating is definitely on the todo list with any issues like this, since the lanes do directly come from the CPU
 
Back
Top