I/O error opening device; RmInitAdapter failed.

Accelerated Computing CUDA CUDA Programming and Performance
1

I’m trying to get started with CUDA on a laptop which has both a GeForce 9200M GS and a GeForce 9400M G. When I try and run the deviceQuery SDK example, I get the following:

toojays@yoshi:~/NVIDIA_CUDA_SDK$ bin/linux/debug/deviceQuery

NVIDIA: could not open the device file /dev/nvidia0 (Input/output error).

cudaSafeCall() Runtime API error in file <deviceQuery.cu>, line 59 : initialization error.

Whenever I run this program, the following messages appear in my dmesg:

[ 1901.110911] NVRM: RmInitAdapter failed! (0x31:0xffffffff:1018)

[ 1901.110920] NVRM: rm_init_adapter(0) failed

I have tried some tips I saw elsewhere about adding “vmalloc=256MB” and/or “pcd=nommconf” to my boot options, but these didn’t help.

Maybe also relevant is that /proc/driver/nvidia/cards/0 does not show and Video BIOS for the 9200M:

toojays@yoshi:~$ cat /proc/driver/nvidia/cards/0 

Model: 		 GeForce 9200M GS

IRQ:   		 23

Video BIOS: 	 ??.??.??.??.??

Card Type: 	 PCI-E

DMA Size: 	 32 bits

DMA Mask: 	 0xffffffff

Bus Location: 	 02.00.0

toojays@yoshi:~$ cat /proc/driver/nvidia/cards/1

Model: 		 GeForce 9400M G

IRQ:   		 18

Video BIOS: 	 62.79.3f.00.01

Card Type: 	 PCI

DMA Size: 	 40 bits

DMA Mask: 	 0xffffffffff

Bus Location: 	 03.00.0

toojays@yoshi:~$ cat /proc/driver/nvidia/version 

NVRM version: NVIDIA UNIX x86_64 Kernel Module  180.29  Wed Feb  4 23:39:47 PST 2009

GCC version:  gcc version 4.3.3 (Ubuntu 4.3.3-3ubuntu5)

Could someone suggest how to solve this issue?

Output of nvidia-bug-report.sh is attached.
nvidia_bug_report.log.txt (177 KB)

2

Have you gotten this to work yet? I have the same problem right now. See my thread: [topic=“0”]http://forums.nvidia.com/index.php?showtopic=93854[/topic]

3

No, I was just going to wait until CUDA 2.2 is released and try again then.

By the way, I have the same laptop as you (running Ubuntu Jaunty though) and find my whole machine locks up hard occasionally when using Firefox with Compiz. So I don’t use Compiz anymore. Presumably you don’t have this problem?

4

To be honest, I havent had Compiz running long enough to give you a good answer… i had only installed the nvidia drivers specifically for CUDA, and I only did that 2 days ago.

5

Don’t know if this is still relevant.

I was facing the following problem with Nvidia 9800GT:

NVRM: loading NVIDIA UNIX x86 Kernel Module  185.18.36  Fri Aug 14 17:18:04 PDT 2009

vmap allocation for size 16781312 failed: use vmalloc=<size> to increase size.

NVRM: RmInitAdapter failed! (0x26:0xffffffff:1096)

NVRM: rm_init_adapter(0) failed

vmap allocation for size 16781312 failed: use vmalloc=<size> to increase size.

NVRM: RmInitAdapter failed! (0x26:0xffffffff:1096)

NVRM: rm_init_adapter(0) failed

The problem got SOLVED by following the “Blacklisting nouveau module” step here : http://forums.fedoraforum.org/showpost.php…amp;postcount=1

The new nouveau (open-source) module stops Nvidia proprietary driver from loading, which causes this problem.