[Solved] eGPU issue

wangcow · 2018-10-31 21:10:47

Hello there,

I have been trying this past week to get this eGPU setup going. To start, the PSU is jumped to be constantly on. I initially had no luck booting with the eGPU inserted into the Express Card slot. So with the PERST# Delay switch set to 2 and the CLKRUN switch set to 1, I booted into the OS w/o it recognizing the card. Long story short, I figured out that if I plug in the express card anytime after seeing the initial Lenovo screen with both switches to 1, I could boot into the OS and have it recognize the GPU.

Now, I have had no luck getting the NVIDIA drivers (except one time, below) to work. Dmesg show's this error:

[ 1791.537139] nvidia-nvlink: Nvlink Core is being initialized, major device number 236
[ 1791.537474] NVRM: This PCI I/O region assigned to your NVIDIA device is invalid:
NVRM: BAR0 is 0M @ 0x0 (PCI:0000:04:00.0)
[ 1791.537475] NVRM: The system BIOS may have misconfigured your GPU.
[ 1791.537480] nvidia: probe of 0000:04:00.0 failed with error -1
[ 1791.537493] NVRM: The NVIDIA probe routine failed for 1 device(s).
[ 1791.537494] NVRM: None of the NVIDIA graphics adapters were initialized!
[ 1791.537630] nvidia-nvlink: Unregistered the Nvlink Core, major device number

And lspci -v:

04:00.0 VGA compatible controller: NVIDIA Corporation GM200 [GeForce GTX 980 Ti] (rev a1) (prog-if 00 [VGA controller])
Subsystem: Micro-Star International Co., Ltd. [MSI] GM200 [GeForce GTX 980 Ti]
Physical Slot: 1
Flags: fast devsel, IRQ 18
Memory at <unassigned> (64-bit, prefetchable) [disabled]
Memory at <unassigned> (64-bit, prefetchable) [disabled]
I/O ports at 4000 [size=128]
[virtual] Expansion ROM at f1400000 [disabled] [size=512K]
Capabilities: [60] Power Management version 3
Capabilities: [68] MSI: Enable- Count=1/1 Maskable- 64bit+
Capabilities: [78] Express Legacy Endpoint, MSI 00
Capabilities: [100] Virtual Channel
Capabilities: [250] Latency Tolerance Reporting
Capabilities: [258] L1 PM Substates
Capabilities: [128] Power Budgeting <?>
Capabilities: [420] Advanced Error Reporting
Capabilities: [600] Vendor Specific Information: ID=0001 Rev=1 Len=024 <?>
Capabilities: [900] Secondary PCI Express <?>
Kernel modules: nouveau, nvidia_drm, nvidia

04:00.1 Audio device: NVIDIA Corporation GM200 High Definition Audio (rev a1)
Subsystem: Micro-Star International Co., Ltd. [MSI] GM200 High Definition Audio
Physical Slot: 1
Flags: bus master, fast devsel, latency 0, IRQ 19
Memory at f1480000 (32-bit, non-prefetchable) [size=16K]
Capabilities: [60] Power Management version 3
Capabilities: [68] MSI: Enable- Count=1/1 Maskable- 64bit+
Capabilities: [78] Express Endpoint, MSI 00
Capabilities: [100] Advanced Error Reporting
Kernel driver in use: snd_hda_intel
Kernel modules: snd_hda_intel

I've tried many solutions with this dmesg error to no avail. (Using echo to disconnect PCI devices and rescan, pci=nocrs kernel flag, nouveau.modeset=0 kernel flag, etc...).

So earlier today I decided to give it one last go before tapping out. I'm not sure what my line of thinking was, but I believe I modprobe'd the nouveau drivers, rebooted, plugged in expresscard @ the Lenovo screen, and while in grub set the kernel flag of pci=nocrs. It worked! I had the external display up with my laptop as a secondary display, delved into some games to see what the fps would be like.. it was great. Went to plug in the speakers from the monitor to the laptop and bumped the card out. No matter what I do, I can't replicate whatever it is I did. I keep getting the errors above. This tells me it's not a hardware issue, and is probably something so simple... it's killing me.

Based on the error, it seems like a memory issue... BIOS is up-to-date. On windows I got an error 12 but got the BSOD trying to do the manual DSDT override. I guess I'll be reinstalling while I await a reply.

My hardware:
- ThinkPad x230t
- i7 - 3520M
- 8 GB RAM

-PE4C v3.0
-MSI 980 ti
-Corsair CX 500

Last edited by wangcow (2018-11-14 04:37:17)

Slithery · 2018-10-31 21:16:19

Please remove the redundant cry for help from your thread title.

Code of Conduct - How to post

wangcow · 2018-11-03 12:19:05

Copy that m8

wangcow · 2018-11-06 04:35:32

So I got it working in Windows.. Followed the Code 12 guide on egpu.io and bought nando4's setup tool (Boot into Menu Based, Reset startup.bat, PCI Compaction 36-bit, Run on eGPU only, Reboot, Choose automated startup using startup.bat). Averages in Valley and Heaven were around 70 FPS. Also worth noting, with Windows 7 installed on an mSATA and using grub on my main SSD, you have to remove the SSD to access nando4's tool.

Really stuck on where to go with this BAR0 error. I guess this Windows test proves it's possible.

wangcow · 2018-11-10 18:23:05

ALRIGHT! Got it figured out. To my future x230t egpu peeps... All switches set to 1. Boot the laptop WITH the expresscard inserted, and monitor UNPLUGGED. This avoids all DSDT errors and all improper memory allocation errors in Linux. Works flawlessly.

Slithery - not sure I could've done this without your help. Thanks for all the assistance. Professionalism and just being kind goes a long way in life if you're ever struggling.

Last edited by wangcow (2018-11-10 18:34:54)

Arch Linux

#1 2018-10-31 21:10:47

[Solved] eGPU issue

#2 2018-10-31 21:16:19

Re: [Solved] eGPU issue

#3 2018-11-03 12:19:05

Re: [Solved] eGPU issue

#4 2018-11-06 04:35:32

Re: [Solved] eGPU issue

#5 2018-11-10 18:23:05

Re: [Solved] eGPU issue

Board footer