Hybrid nVidia - modprobe says "No such device" [SOLVED]

/dev/zero · 2013-10-13 21:32:35

My Toshiba Satellite has an intel/nvidia optimus graphics card (see lspci output below). Everything is working well by running my X environment on the Intel card but I would like to try using the nVidia GPU for CUDA programming. I get the below error whether I use the standard nvidia drivers from pacman or nvidia-custom.

$ sudo modprobe -vv nvidia
modprobe: INFO: custom logging function 0x409c10 registered
modprobe: INFO: Failed to insert module '/lib/modules/3.11.4-1-ARCH/extramodules/nvidia.ko.gz': No such device
modprobe: ERROR: could not insert 'nvidia': No such device
modprobe: INFO: context 0x24481e0 released
insmod /lib/modules/3.11.4-1-ARCH/extramodules/nvidia.ko.gz

True enough, if I run lsmod after this, I see that nvidia hasn't been loaded. I also have bumblebee installed but the error persists regardless of whether that service is running.

I don't get the error when I modprobe nouveau. Nouveau gives me other problems when I try using optirun/bumblebee but that's independent, something to do with KMS not initialising correctly. So for the moment I want to try and make this work with the nvidia drivers. I currently have nouveau blacklisted.

Some hardware info:

$ uname -a
Linux GLaDOS 3.11.4-1-ARCH #1 SMP PREEMPT Sat Oct 5 21:22:51 CEST 2013 x86_64 GNU/Linux

$ lspci -nnvk | grep -A10 "VGA\|NVIDIA"
00:02.0 VGA compatible controller [0300]: Intel Corporation 3rd Gen Core processor Graphics Controller [8086:0166] (rev 09) (prog-if 00 [VGA controller])
	Subsystem: Toshiba America Info Systems Device [1179:fa43]
	Flags: bus master, fast devsel, latency 0, IRQ 45
	Memory at d3000000 (64-bit, non-prefetchable) [size=4M]
	Memory at c0000000 (64-bit, prefetchable) [size=256M]
	I/O ports at 4000 [size=64]
	Expansion ROM at <unassigned> [disabled]
	Capabilities: <access denied>
	Kernel driver in use: i915
	Kernel modules: i915

--
01:00.0 3D controller [0302]: NVIDIA Corporation GK208M [GeForce GT 740M] [10de:1292] (rev a1)
	Subsystem: Toshiba America Info Systems Device [1179:fa43]
	Flags: fast devsel, IRQ 16
	[virtual] Memory at d2000000 (32-bit, non-prefetchable) [size=16M]
	Memory at a0000000 (64-bit, prefetchable) [size=256M]
	Memory at b0000000 (64-bit, prefetchable) [size=32M]
	I/O ports at 3000 [size=128]
	[virtual] Expansion ROM at b2000000 [disabled] [size=512K]
	Capabilities: <access denied>
	Kernel modules: nouveau, nvidia

$ lsmod | grep -i "video\|nvidia\|nouveau\|i915"
uvcvideo               72804  0 
videobuf2_vmalloc       3272  1 uvcvideo
videobuf2_memops        2335  1 videobuf2_vmalloc
videobuf2_core         27797  1 uvcvideo
videodev              110188  2 uvcvideo,videobuf2_core
media                  11591  2 uvcvideo,videodev
i915                  611170  3 
intel_agp              10872  1 i915
intel_gtt              12664  2 i915,intel_agp
drm_kms_helper         35854  1 i915
drm                   236092  5 ttm,i915,drm_kms_helper
video                  11380  1 i915
button                  4669  1 i915
i2c_algo_bit            5391  1 i915
i2c_core               23720  6 drm,i915,i2c_i801,drm_kms_helper,i2c_algo_bit,videodev
usbcore               178119  7 keucr,uvcvideo,usb_storage,ehci_hcd,ehci_pci,usbhid,xhci_hcd

Some relevant output from pacman:

$ pacman -Qs nvidia | grep "^[[:alpha:]]"
local/bumblebee 3.2.1-3
local/libcl 1.1-3
local/libvdpau 0.7-1
local/nvidia-custom 331.13-1 (nvidia-cuda custom)
local/nvidia-utils-custom 331.13-1 (nvidia-cuda custom)
local/opencl-nvidia-custom 331.13-1 (nvidia-cuda custom)

I don't currently have any xorg configs set up for nvidia; I believe that would muddy the issue given that I don't need to run any graphical software on the nvidia card, it is just for GPU programming. (See, for example, GWIS - model-free, fast and exhaustive search for epistatic interactions in case-control GWAS: "Exhaustive analysis of studies containing approximately 450,000 SNPs and 5,000 samples requires only 2 hours using a desktop CPU or 13 minutes using a GPU (Graphics Processing Unit).") I would expect that xorg configs only become relevant after the card has been successfully modprobed and you want the card to interface with X, correct me if I'm wrong?

If anyone has any ideas on how I can get modprobe to successfully see the nvidia device, I'd really appreciate it!

Edit: PS - Of course I tried STFWing but found no love anywhere.

Last edited by /dev/zero (2013-10-14 02:18:40)

ngoonee · 2013-10-14 01:34:09

I get the same error on my optimus laptop, but optirun/primusrun have always worked for me for games. I'm not sure exactly what the bumblebee wrapper (optirun) does, but maybe you could, for example, run

optirun bash

A quick test on that in my system shows that the resulting shell has access to the nvidia card (glxgears shows the nvidia framerate, not the lower intel framerate). lsmod shows that the module is loaded.

/dev/zero · 2013-10-14 02:17:35

ngoonee wrote:

I get the same error on my optimus laptop, but optirun/primusrun have always worked for me for games. I'm not sure exactly what the bumblebee wrapper (optirun) does, but maybe you could, for example, run
optirun bash
A quick test on that in my system shows that the resulting shell has access to the nvidia card (glxgears shows the nvidia framerate, not the lower intel framerate). lsmod shows that the module is loaded.

This is embarrassing. The problem seems to be fixed now after a reboot. I already rebooted multiple times in the last 24 hours and I must have lost track of needing to do the most recent reboot. So I will mark as solved o_O.

If anyone finds this thread via Google, make sure you reboot every time you reinstall the nvidia drivers! <facepalm />

Thanks ngoonee for inspiring me to finally find the answer

Arch Linux

#1 2013-10-13 21:32:35

Hybrid nVidia - modprobe says "No such device" [SOLVED]

#2 2013-10-14 01:34:09

Re: Hybrid nVidia - modprobe says "No such device" [SOLVED]

#3 2013-10-14 02:17:35

Re: Hybrid nVidia - modprobe says "No such device" [SOLVED]

Board footer