Unsolved
1 Rookie
•
1 Message
0
2069
September 24th, 2022 21:00
Aurora R12, PCIe training error
Alienware Aurora R12
Received an error in a diagnostics run,
PCIe - Training error PCI tag - 0200 VendorID-10DE Device ID 2204 Svid-1028 SDid-3880 Bus 02: Link Degraded, Max width x16, Neg width x 4 PCIE Slot 01
Anyone else experienced this issue? As far as I'm aware everything is up to date according to the Alienware update, Windows 11 update and Nvidia updater.
Thanks.
No Events found!



ProfessorW00d
4 Operator
•
2.3K Posts
0
September 24th, 2022 21:00
you are beta testing Windows 11
Vanadiel
6 Professor
•
7K Posts
0
September 25th, 2022 03:00
That looks like an RTX3090. Did you install an aftermarket video card, or is that the original OEM video card?
Check your power settings. Link speed goes down to X4 pending on the power settings. It should bump up to X16 once you put load on the video card.
If you want it always at X16 you have to disable PCI express link speed power management under the advanced tab of the windows power plan settings.
Vanadiel
6 Professor
•
7K Posts
0
September 25th, 2022 08:00
Yours does not stay linked at X16 with Link State Power Management set to off?
ProfessorW00d
4 Operator
•
2.3K Posts
0
September 25th, 2022 08:00
ProfessorW00d
4 Operator
•
2.3K Posts
0
September 25th, 2022 08:00
That is still not working for me. Did you ever come across your alternate solution?
ProfessorW00d
4 Operator
•
2.3K Posts
0
September 25th, 2022 09:00
I don't know how to tell if it is X16, but I am going by my low idle MHZ
Vanadiel
6 Professor
•
7K Posts
0
September 25th, 2022 09:00
GPU-Z will show you the link speed. I think the GPU frequency being low is a separate power savings feature.
ProfessorW00d
4 Operator
•
2.3K Posts
0
September 25th, 2022 14:00
Bus Interface?
Vanadiel
6 Professor
•
7K Posts
0
September 25th, 2022 15:00
Yes sir. You have something going on because you are linked at X16 but at PCIE standard 1.1 instead of 4.0.
I have seen this before: 1.1 PCIE speeds
I don't know if there is a BIOS setting that affects this on the R13. On my R10 there's no settings and it is all regulated by the Windows Power plan.
I do know that AWCC tends to override the power settings, so that is another area to look at.
Also, you have REBAR disabled. If I am not mistaken the R13 supports REBAR.
If that is a Dell RTX 3080 than we have the same card. Mine does not change link or PCIE speeds when disabling Link State Power Management.
ProfessorW00d
4 Operator
•
2.3K Posts
0
September 25th, 2022 17:00
PC is Area-51 R5 and RTX 3080 is Founders Edition. If I start the render test the 1.1 changes to 3.0. I'm not sure what resizable bar is/does?
Tesla1856
8 Wizard
•
17.1K Posts
0
September 25th, 2022 17:00
With what software?
Is the ePSA (outside of Windows) or instead ... something you found inside Windows?
Vanadiel
6 Professor
•
7K Posts
0
September 26th, 2022 03:00
I thought we were talking about an R13. On an R5 you will not have REBAR support. REBAR = Resizable bar
Resizable bar support
The 3080 can use PCIE version 4.0, but on that motherboard it will be restricted to PCIE 3.0.
Could be an issue with that. It's for sure a power savings feature that is enabled somewhere. Not sure what that BIOS has for power savings options.
KenEglinton
12 Posts
0
May 14th, 2023 01:00
I'm sure we can do better, the link state power management tab has nothing to do with lane/link speed. Although I've turned that off since win98 and it will effect gpu when idle-his gpu needs to be under load for full pcie link-its automatic-he is having link training issue because support assist believes those are x16 pcie slots-they are not-its well known dell uses x8 lanes since the R7, R8, R9 greedfest that eluded major media exposure-same mobo same everything for R9 new look-m.2 drives not pcie 4.0 on R10-3rd slot at x2-I installed a pcie 4 ssd in the gpu slot and achieved 7,000 only to yank it for oem dell 6800xt facebook has been paying them thru the nose-using gpu-z, downloading the directx3d driver should get things going-there are also command profiles pre-made for OC and CPU-the folks who provide us with.
I bricked a R7 modding the bios-I brought it back to life with a hacked bios on purpose to prompt bios guard to reinstall the backup-thats how I came up with idea of letting dells own program insert the cpuid code and power update into the nvram-Ive been sporting a 9th gen in an R7 for 2 years-its a crazy amalgamation of a bios file from china, dells inhouse xps 8930 that I merged into R7 bios that booted and ended 3 hours later with my win10 pro wiped and installed under some tag from china lol-
it ran for a few months and like a dummy I didnt disable UEFI updates in bios and windows slid in firmware and no more post-I vowed to get back later and using an R8 chip,hardware programmer and some nvram hotsauce back in business.
so 1 pcie 4.0 cpu slot R10, if more than 3 m.2 ssd installed the pch slot runs at 8gts and intel fake raid a no go but software raid not bad but fast as weakest m.2 slot so switch cards good but no dual 4.0 cards work either which I believe the misrepresentation of actual pcie link speeds-even hwinfo gets it wrong.
Vanadiel
6 Professor
•
7K Posts
0
May 14th, 2023 07:00
Who are you talking to?