damie158f02e3
New member
- Joined
- Feb 9, 2025
- Messages
- 4
Hi I made this because a lot of the threads about this are concerning specific motherboards and I believe this to be a problem over multiple ranges x870e if not the entire 870/850 range I have seen posts from tomahawk owners, some 850 ones that looked very similar, and my own x870e gaming + wifi etc.
I note all the bios updates msi has put out for the gaming + wifi I mention above has NOT solved the problem but several things have led me to believe I have a notion of what may be happening (or at least the general location)
First the issue most have described
Graphics cards get 'stuck' at 16x1 instead of 16x4 on some boots. Mine seems to jump around at the moment.
Other issues described on other posts I believe are related
First I use a sata raid array from a pcie card, the card is insert correctly as is the graphics card (3090).
I first noted the odd behaviour in some games, when processing the shaders, the whole system would grind to a halt, something my old am3 board did not do. This was extending into gameplay where the pcie of the graphics card would regularly (with increasing frequency as time went on) drop back to pcie16x1 and at that time (just after the dropback) audio would drop out. This would get worse and worse until I couldn't hear anything, it would also affect all sound on the system, not just the ingame sound. So discord would go down, everything.
I also noted that things were becoming slower to load from desktop. I updated my motherboards bios and the problems remained.
I also started noting the increasingly large number of error events in the windows logs, with conflicts occurring and the 2gbe lan getting an error every second now.
After about a week, windows started to report errors on on of the sata drives connected to a PCIe card. These are drives less than a year old with a 10 year warranty on them and are built for doing these middling ranges of raid, they have never had bad blocks before. As mentioned above, if you plug all of these cards into another motherboard, everything works.
Why I think it is PCI bus/channel comms issue or power issue
The lan starting to get hardware errors could be caused by voltage irregularities, but also if the incorrect data is being returned I would not be surprised to see an error like this.
The graphics cards randomly throttling down could be caused by voltage issues, OR could be caused by miscommunications, I have noticed some graphical artifacts on occasion.
The hard drive having a detected bad block could be a hdd on the way out (unlikely for a prosumer drive but not unheard of) or could be caused by voltage issues, or could be caused when bad data is written to the drive on that channel. Now I am getting event logs indicating conflicts when multiple things try to use the pci bus at the same time.
It could also be data coming from memory is corrupted, this could be backed up by the voltage changes to RAM causing further issues or less issues for some people.
Comms/voltage etc. also explains why it is so hard to pinpoint the issue as they would be occurring irregularly and in a way that may not be easily testable.
Other forums with the similar or the same issues:
The biggest with the most info - MSI shrugs and says we can't figure it out -
https://forum-en.msi.com/index.php?threads/nvidia-4090-gpu-stuck-at-pcie1-1x16.387014/
There are many more out there including one on LTT, but I just added the ones I could find immediately.
No power saving features enabled 4
I note all the bios updates msi has put out for the gaming + wifi I mention above has NOT solved the problem but several things have led me to believe I have a notion of what may be happening (or at least the general location)
First the issue most have described
Graphics cards get 'stuck' at 16x1 instead of 16x4 on some boots. Mine seems to jump around at the moment.
Other issues described on other posts I believe are related
- Instability using EXPO
- 2gbe (or other) lan hardware IO error
- PCIe cards doing weird things
- voltage changes making stability differences beyond the amount of stability difference that should be happening (using in-range voltages and having things go weird anyway)
- Putting all system cards on a different brand of board removes all issues, putting other cards on the problem board and the issues still occur.
- What I think is happening
- Either the voltage regulator has a lot more variation in it with the PCIe bus or something is causing bad PCIe data to be sent through the bus, this could be a knockon effect from another power issue, but seems to cover the bases.
First I use a sata raid array from a pcie card, the card is insert correctly as is the graphics card (3090).
I first noted the odd behaviour in some games, when processing the shaders, the whole system would grind to a halt, something my old am3 board did not do. This was extending into gameplay where the pcie of the graphics card would regularly (with increasing frequency as time went on) drop back to pcie16x1 and at that time (just after the dropback) audio would drop out. This would get worse and worse until I couldn't hear anything, it would also affect all sound on the system, not just the ingame sound. So discord would go down, everything.
I also noted that things were becoming slower to load from desktop. I updated my motherboards bios and the problems remained.
I also started noting the increasingly large number of error events in the windows logs, with conflicts occurring and the 2gbe lan getting an error every second now.
After about a week, windows started to report errors on on of the sata drives connected to a PCIe card. These are drives less than a year old with a 10 year warranty on them and are built for doing these middling ranges of raid, they have never had bad blocks before. As mentioned above, if you plug all of these cards into another motherboard, everything works.
Why I think it is PCI bus/channel comms issue or power issue
The lan starting to get hardware errors could be caused by voltage irregularities, but also if the incorrect data is being returned I would not be surprised to see an error like this.
The graphics cards randomly throttling down could be caused by voltage issues, OR could be caused by miscommunications, I have noticed some graphical artifacts on occasion.
The hard drive having a detected bad block could be a hdd on the way out (unlikely for a prosumer drive but not unheard of) or could be caused by voltage issues, or could be caused when bad data is written to the drive on that channel. Now I am getting event logs indicating conflicts when multiple things try to use the pci bus at the same time.
It could also be data coming from memory is corrupted, this could be backed up by the voltage changes to RAM causing further issues or less issues for some people.
Comms/voltage etc. also explains why it is so hard to pinpoint the issue as they would be occurring irregularly and in a way that may not be easily testable.
Other forums with the similar or the same issues:
The biggest with the most info - MSI shrugs and says we can't figure it out -
https://forum-en.msi.com/index.php?threads/nvidia-4090-gpu-stuck-at-pcie1-1x16.387014/
There are many more out there including one on LTT, but I just added the ones I could find immediately.
No power saving features enabled 4
Last edited: