PCIe drops from x16 3.0 to 1.1 and stays there under load

5150Joker

Supreme [H]ardness
Joined
Aug 1, 2005
Messages
4,445
So to start, my hardware:

Gigabyte Z390 Aorus Master with the latest F11c bios
Intel 9900K
Asus Strix 2080 Ti OC and it's using a Phanteks PCI-e riser (gpu is placed upright)
Ram is Corsair Vengeance RGB Pro 32 GB modules (4 x 8) 3200MHz C16 (XMP enabled in bios)
Latest NVIDIA driver (452.06)

The problem I noticed was my FPS in Warzone had become atrocious and even teammates pointed it out. Last night at 1080p with almost everything on low I was barley getting 80-90 fps and a buddy of mine on an old 4790k and gtx 1080 was getting 60-70 fps and saw my stream and pointed it out. I had chalked it up to bad game optimization until I started poking around some more and realized it was my system. I had this issue before and did a clean windows install recently and even now I ran DDU and got rid of the driver and installed the latest one.
gpuz.png
B04F6946-71FB-4BAE-A596-880D84B73373.jpeg
So I poked around Googling what the problem could be and someone with a similar problem mentioned the PCIe x 16 power state was running at 1.1 and to check GPU-Z. Now I know at idle the GPU powers down and will run at 1.1, that would be fine as long as it went back to 3.0 during load but it doesn't. Instead, when I turn on a game after a clean boot, it drops from x16 3.0 to x16 1.1 and I'm at a loss for figuring out what it could be.


Things I've tried:
1. Clean driver install
2. Reset bios to optimized defaults
3. Turned off PCIe link state power management in the power plan
4. Reseated the Phantek Riser Cable

I just cannot figure out why it's doing this but there's a clear loss of FPS happening and it's pretty massive. Earlier I managed to fire up Warzone while it would still run at x16 3.0 and my FPS was nearly double what it usually is until it dropped to x16 1.1. I'm hoping someone here has experience with it and can help me solve it.

Edit: I want to point out that my suspicion is it could be the Phantek cable but why would it now give issues when it didn't from Jan-May?

UPDATE 1:

I reset the CMOS again, pulled the riser cable out and reseated again, loaded optimized defaults and now I've got Warzone up and it's staying at x16 3.0 so far. Hopefully it stays there after I re-apply my OC.
 
Last edited:

5150Joker

Supreme [H]ardness
Joined
Aug 1, 2005
Messages
4,445
Try it without the riser.

In the UEFI set that slot to PCI-e 3.0

I haven't found anything in the UEFI to change the PCIe setting. It does show x16 1.1 there though in the information section just like GPU-Z does.
 

5150Joker

Supreme [H]ardness
Joined
Aug 1, 2005
Messages
4,445
Update: I reset the CMOS again, pulled the riser cable out and reseated again, loaded optimized defaults and now I've got Warzone up and it's staying at x16 3.0 so far. Hopefully it stays there after I re-apply my OC.
 

mda

[H]ard|Gawd
Joined
Mar 23, 2011
Messages
1,902
Certainly seems odd. Hope it stays.

Does your OC mess with the BCLK? If it does, then that may be a reason if the PCIE is downclocking.
 

5150Joker

Supreme [H]ardness
Joined
Aug 1, 2005
Messages
4,445
Certainly seems odd. Hope it stays.

Does your OC mess with the BCLK? If it does, then that may be a reason if the PCIE is downclocking.

I leave the bclk at auto. Unfortunately it reverted back to 1.1 again. If it was a bad cable I’m thinking it wouldn’t work at all so it has to be something else.
 

mda

[H]ard|Gawd
Joined
Mar 23, 2011
Messages
1,902
A bad cable can probably cause 1.1 due to bad contacts or it autodetecting insufficient bandwidth. As much as it's hard to test, it looks like the easiest and most conclusive way to verify at this point.

If you have a spare GPU to see if it will also drop to 1.1, that will indicate a likely problem with the riser or the slot.
 

5150Joker

Supreme [H]ardness
Joined
Aug 1, 2005
Messages
4,445
A bad cable can probably cause 1.1 due to bad contacts or it autodetecting insufficient bandwidth. As much as it's hard to test, it looks like the easiest and most conclusive way to verify at this point.

If you have a spare GPU to see if it will also drop to 1.1, that will indicate a likely problem with the riser or the slot.

I have that Asus cable on order so I'll probably just rip the 2080 Ti out if it causes me more problems. Right now I managed to get the OC back to 5 GHz stable and the PCIe is staying at x16 3.0 but this thing is pretty unpredictable so it might not work tomorrow. Thx for the assist.
 
Last edited:

sirmonkey1985

[H]ard|DCer of the Month - July 2010
Joined
Sep 13, 2008
Messages
22,223
I have that Asus cable on order so I'll probably just rip the 2080 Ti out if it causes me more problems. Right now I managed to get the OC back to 5 GHz stable and the PCIe is staying at x16 3.0 but this thing is pretty unpredictable so it might not work tomorrow. Thx for the assist.

it's likely dropping to 1.1 when it's trying to swap a massive amount of textures like a map change or something but yeah i'm with everyone else that it's likely an issue with the riser cable, maybe one of the internal wires broke and so when it's making contact it's working and then it starts heating up it fails. just my guess.
 

5150Joker

Supreme [H]ardness
Joined
Aug 1, 2005
Messages
4,445
it's likely dropping to 1.1 when it's trying to swap a massive amount of textures like a map change or something but yeah i'm with everyone else that it's likely an issue with the riser cable, maybe one of the internal wires broke and so when it's making contact it's working and then it starts heating up it fails. just my guess.

Yeah if it fails again I’ll remove it. The ASUS cable arrives on Sunday too.
 
Last edited:

somebrains

[H]ard|Gawd
Joined
Nov 10, 2013
Messages
1,360
Did the problem happen after that last Win10 update?
I’ve noticed some wonky power state issues in other devices, not just pcie.
 
Top