Second GPU not being initialized

twistx · 01-25-10, 07:19 PM

Hello,
I'm having an issue in linux mint where my second GPU is not recognized or being initialized by Xorg.

Two identical cards, both appear in lspci:

Code:

03:00.0 VGA compatible controller: nVidia Corporation G71 [GeForce 7900 GTX] (rev a1)
06:00.0 VGA compatible controller: nVidia Corporation G71 [GeForce 7900 GTX] (rev a1)

Errors appear in Xorg.0.log:

Code:

(EE) Jan 25 19:48:44 NVIDIA(0): Failed to initialize one NVIDIA graphics device!
(EE) Jan 25 19:48:45 NVIDIA(GPU-1): Failed to initialize the NVIDIA graphics device PCI:6:0:0.
(EE) Jan 25 19:48:45 NVIDIA(GPU-1):     Please check your system's kernel log for additional error
(EE) Jan 25 19:48:45 NVIDIA(GPU-1):     messages and refer to Chapter 8: Common Problems in the
(EE) Jan 25 19:48:45 NVIDIA(GPU-1):     README for additional information.
(EE) Jan 25 19:48:45 NVIDIA(GPU-1): Failed to initialize the NVIDIA graphics device!

Kernel messages:

Code:

[ 3693.870527] NVRM: loading NVIDIA UNIX x86 Kernel Module  190.53  Tue Dec  8 18:51:41 PST 2009
[ 3695.324461] NVRM: RmInitAdapter failed! (0x31:0xffffffff:1039)
[ 3695.324473] NVRM: rm_init_adapter(1) failed
[ 3696.605841] NVRM: RmInitAdapter failed! (0x31:0xffffffff:1039)
[ 3696.605852] NVRM: rm_init_adapter(1) failed

I have tried the vmalloc=256MB kernel parameter but it had no affect on my issue. You can see I'm on 190.53, this is after a manual upgrade from the mint recommended 185.18.36.

My xorg.conf

Code:

Section "ServerLayout"
    Identifier     "Layout0"
    Screen      0  "Screen0" 0 0
    InputDevice    "Keyboard0" "CoreKeyboard"
    InputDevice    "Mouse0" "CorePointer"
EndSection

Section "Files"
EndSection

Section "ServerFlags"
    Option         "Xinerama" "0"
EndSection

Section "InputDevice"

    # generated from default
    Identifier     "Mouse0"
    Driver         "mouse"
    Option         "Protocol" "auto"
    Option         "Device" "/dev/psaux"
    Option         "Emulate3Buttons" "no"
    Option         "ZAxisMapping" "4 5"
EndSection

Section "InputDevice"

    # generated from default
    Identifier     "Keyboard0"
    Driver         "kbd"
EndSection

Section "Monitor"
    Identifier     "Monitor0"
    VendorName     "Unknown"
    ModelName      "Samsung SyncMaster"
    HorizSync       30.0 - 81.0
    VertRefresh     56.0 - 75.0
    Option         "DPMS"
EndSection

Section "Device"
    Identifier     "Device0"
    Driver         "nvidia"
    VendorName     "NVIDIA Corporation"
    BoardName      "GeForce 7900 GTX"
EndSection

Section "Device"
    Identifier     "Device1"
    Driver         "nvidia"
    VendorName     "NVIDIA Corporation"
    BoardName      "GeForce 7900 GTX"
    BusID          "PCI:6:0:0"
EndSection

Section "Screen"
    Identifier     "Screen0"
    Device         "Device0"
    Monitor        "Monitor0"
    DefaultDepth    24
    Option         "TwinView" "1"
    Option         "metamodes" "DFP-0: nvidia-auto-select +0+0, DFP-1: nvidia-auto-select +1600+0; 1600x1200,NULL"
    Option         "MultiGPU" "ON"
    Option         "SLI" "auto"
    SubSection     "Display"
        Depth       24
    EndSubSection
EndSection

Otherwise Xorg runs completely fine, no segfaults or instability issues.

Any advice is appreciated. Thanks in advance!

twistx · 01-26-10, 12:57 AM

Resolved this by manually updating both card's BIOS to Rev3 from mvktech.net. Created a dos boot disk with unetbootin, downloaded the latest dos compatible nvflash that would support 7900GTX cards which was 5.31, then flashed each card up to rev3. Rev4 appeared incompatible with my cards despite what the site says. Xorg can initialize both cards, and SLI is enabled as well.

**zander** · 01-26-10, 10:41 AM

FWIW, the failure in question reported is consistent with the resolution: the driver reported an error it detected attempting to POST the secondary device.

@twistx: I assume the Rev4 VBIOS you had previously flashed was not the original OEM VBIOS image?

twistx · 01-26-10, 01:20 PM

zander,
I'm not sure if it was OEM or a modified version of the vbios, but my card before flashing was at rev1. My cards are both BFG 7900GTX OC, and I found all 4 versions of the vbios on mvktech were uploaded by the same user and listed for this specific card. However, rev4 was the only vbios that had a different boardid than any of the previous 3 vbios versions. Due to this, with a -5 -6 it would not take rev4, and I didn't think it was wise to void these compatibility checks, so I stopped with rev3. If you know otherwise, and its safe to force flash to rev4 please let me know as this might prevent future problems.

vbios for my cards on this page:
http://www.mvktech.net/component/opt...erby,2/page,4/

Also, after upgrading these two cards are now collectively louder than all of my other components. There doesn't appear to be any variable speed during load or idle. Is there a way to fix this?

Last, are there any plans to enable multi-monitor sli in linux in the near future?

Thanks

01-26-10, 12:57 AM	#2
twistx Registered User Join Date: May 2006 Posts: 12	Re: Second GPU not being initialized Resolved this by manually updating both card's BIOS to Rev3 from mvktech.net. Created a dos boot disk with unetbootin, downloaded the latest dos compatible nvflash that would support 7900GTX cards which was 5.31, then flashed each card up to rev3. Rev4 appeared incompatible with my cards despite what the site says. Xorg can initialize both cards, and SLI is enabled as well.

01-26-10, 10:41 AM	#3
zander NVIDIA Corporation Join Date: Aug 2002 Posts: 3,740	Re: Second GPU not being initialized FWIW, the failure in question reported is consistent with the resolution: the driver reported an error it detected attempting to POST the secondary device. @twistx: I assume the Rev4 VBIOS you had previously flashed was not the original OEM VBIOS image?

01-26-10, 01:20 PM	#4
twistx Registered User Join Date: May 2006 Posts: 12	Re: Second GPU not being initialized zander, I'm not sure if it was OEM or a modified version of the vbios, but my card before flashing was at rev1. My cards are both BFG 7900GTX OC, and I found all 4 versions of the vbios on mvktech were uploaded by the same user and listed for this specific card. However, rev4 was the only vbios that had a different boardid than any of the previous 3 vbios versions. Due to this, with a -5 -6 it would not take rev4, and I didn't think it was wise to void these compatibility checks, so I stopped with rev3. If you know otherwise, and its safe to force flash to rev4 please let me know as this might prevent future problems. vbios for my cards on this page: http://www.mvktech.net/component/opt...erby,2/page,4/ Also, after upgrading these two cards are now collectively louder than all of my other components. There doesn't appear to be any variable speed during load or idle. Is there a way to fix this? Last, are there any plans to enable multi-monitor sli in linux in the near future? Thanks