View previous topic :: View next topic |
Author |
Message |
kelvin34501 n00b
Joined: 05 Oct 2018 Posts: 9
|
Posted: Sat Oct 06, 2018 3:46 am Post subject: [SOLVED] CUDA sample fails |
|
|
I tried to run some cuda samples, but it returns with following error message:
Code: | /opt/cuda/sdk/bin/x86_64/linux/release $ ./deviceQuery
./deviceQuery Starting...
CUDA Device Query (Runtime API) version (CUDART static linking)
cudaGetDeviceCount returned 30
-> unknown error
Result = FAIL |
I installed these packages:
Code: |
$ equery list 'nvidia*'
* Searching for nvidia* ...
[IP-] [ ] dev-util/nvidia-cuda-sdk-9.2.88:0
[IP-] [ ] dev-util/nvidia-cuda-toolkit-9.2.88:0/9.2.88
[IP-] [ ] x11-drivers/nvidia-drivers-396.54:0/396 |
The result of nvidia-smi:
Code: |
$ nvidia-smi
Sat Oct 6 11:40:18 2018
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 396.54 Driver Version: 396.54 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 GeForce GTX 1080 Off | 00000000:1C:00.0 On | N/A |
| 8% 49C P0 45W / 250W | 313MiB / 8116MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| 0 4365 G /usr/bin/X 159MiB |
| 0 4535 G kwin_x11 28MiB |
| 0 4543 G /usr/bin/krunner 2MiB |
| 0 4545 G /usr/bin/plasmashell 106MiB |
| 0 4760 G /usr/bin/akonadi_archivemail_agent 2MiB |
| 0 4782 G /usr/bin/akonadi_mailfilter_agent 2MiB |
| 0 4790 G /usr/bin/akonadi_sendlater_agent 2MiB |
| 0 4996 G /usr/lib64/firefox/firefox 2MiB |
| 0 5267 G /usr/lib64/firefox/firefox 2MiB |
| 0 5432 G /usr/lib64/firefox/firefox 2MiB |
+-----------------------------------------------------------------------------+
|
nvcc version:
Code: |
$ nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2018 NVIDIA Corporation
Built on Wed_Apr_11_23:16:29_CDT_2018
Cuda compilation tools, release 9.2, V9.2.88
|
Are there any solutions? or where can I get more help for this?
Last edited by kelvin34501 on Sun Oct 14, 2018 11:33 am; edited 1 time in total |
|
Back to top |
|
|
Tyler_Durden Apprentice
Joined: 27 Jul 2004 Posts: 189 Location: Germany
|
|
Back to top |
|
|
Aiken Apprentice
Joined: 22 Jan 2003 Posts: 239 Location: Toowoomba/Australia
|
Posted: Sun Oct 07, 2018 8:34 pm Post subject: |
|
|
The tool and driver version are the same as what I am using with a gtx1060. If you do a strace you may find it is not finding /sys/devices/system/node which is related to numa. It barfs when that directory is not found. I enabled
CONFIG_NUMA=y
CONFIG_CPUSETS=y
After a driver upgrade breaking cuda for me those 2 kernel options got it working again. _________________ Beware the grue. |
|
Back to top |
|
|
Chiitoo Administrator
Joined: 28 Feb 2010 Posts: 2573 Location: Here and Away Again
|
|
Back to top |
|
|
kelvin34501 n00b
Joined: 05 Oct 2018 Posts: 9
|
Posted: Sun Oct 14, 2018 11:32 am Post subject: |
|
|
It seems CONFIG_NUMA=y doesn't come with the default kernel config.
After enable that option everything works. |
|
Back to top |
|
|
|