You are not logged in.
I am trying to use an application that needs OpenCl, but I cannot get it to see the nvidia card I have installed.
Some info:
1. Checking that which openCl implementations are active on my system as per the wiki page:
stefano@gorgias $ ls /etc/OpenCL/vendors
mesa.icd nvidia.icd
2. Checking data about the installed card:
stefano@gorgias $ nvidia-smi
Sat Feb 15 11:48:48 2020
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 390.132 Driver Version: 390.132 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 Quadro 2000 Off | 00000000:03:00.0 On | N/A |
| 37% 73C P0 N/A / N/A | 803MiB / 962MiB | 5% Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| 0 926 G /usr/lib/Xorg 197MiB |
| 0 102233 G /usr/bin/kwin_x11 7MiB |
| 0 102237 G /usr/bin/plasmashell 152MiB |
| 0 102835 G ...uest-channel-token=15412631400218406066 333MiB |
| 0 105401 G /usr/bin/krunner 11MiB |
| 0 414013 G python2 1MiB |
| 0 867482 G konqueror 4MiB |
| 0 1132879 G python2 73MiB |
3. Checking that packages are installed:
stefano@gorgias ~ $ pacman -Qs nvidia
local/lib32-nvidia-390xx-utils 390.132-1
NVIDIA drivers utilities (32-bit), 390xx legacy branch
local/libvdpau 1.3-1
Nvidia VDPAU library
local/libxnvctrl-390xx 390.129-1
NVIDIA NV-CONTROL X extension, 390xx legacy branch
local/nvidia-390xx 390.132-26
NVIDIA drivers for linux, 390xx legacy branch
local/nvidia-390xx-settings 390.129-1
Tool for configuring the NVIDIA graphics driver, 390xx legacy branch
local/nvidia-390xx-utils 390.132-2
NVIDIA drivers utilities
local/nvidia-cg-toolkit 3.1-5
NVIDIA Cg libraries
local/opencl-nvidia-390xx 390.132-2
OpenCL implemention for NVIDIA
4. And yet, clinfo cannot find the nvidia platform or device:
stefano@gorgias ~ $ clinfo
Number of platforms 1
Platform Name Clover
Platform Vendor Mesa
Platform Version OpenCL 1.1 Mesa 19.3.3
Platform Profile FULL_PROFILE
Platform Extensions cl_khr_icd
Platform Extensions function suffix MESA
Platform Name Clover
Number of devices 0
NULL platform behavior
clGetPlatformInfo(NULL, CL_PLATFORM_NAME, ...) Clover
clGetDeviceIDs(NULL, CL_DEVICE_TYPE_ALL, ...)
clCreateContext(NULL, ...) [default] No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_DEFAULT) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CPU) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_GPU) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_ACCELERATOR) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CUSTOM) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_ALL) No devices found in platform
ICD loader properties
ICD loader Name OpenCL ICD Loader
ICD loader Vendor OCL Icd free software
ICD loader Version 2.2.12
ICD loader Profile OpenCL 2.2
What am I missing?
Last edited by stefano (2020-02-20 16:06:38)
Offline
Just a thought: Have you tried to remove opencl-mesa?
Offline
Just a thought: Have you tried to remove opencl-mesa?
I hadn't but I tried and it does not help:
stefano@gorgias ~ $ sudo pacman -Rss opencl-mesa
[sudo] password for stefano:
checking dependencies...
:: ocl-icd optionally requires opencl-driver: packaged opencl driver
Packages (2) libclc-0.2.0+589+9aa6f35-1 opencl-mesa-19.3.3-1
Total Removed Size: 77.03 MiB
:: Do you want to remove these packages? [Y/n]
:: Processing package changes...
(1/2) removing opencl-mesa [###################################################] 100%
(2/2) removing libclc [###################################################] 100%
:: Running post-transaction hooks...
(1/1) Arming ConditionNeedsUpdate...
Now I have no platforms at all:
stefano@gorgias ~ $ clinfo
Number of platforms 0
It definitely looks like nvidia is not recognized
Last edited by stefano (2020-02-15 21:48:02)
Offline
I have no real idea, but it would for sure be helpful, if you could post your Xorg.0.log. And maybe it is also useful to know which application you are trying to run...
Last edited by Thorsten Reinbold (2020-02-15 22:01:53)
Offline
No idea if it matters, but libxnvctrl and nvidia-390xx-settings are an older version then the others .
390.129 vs 390.132 .
Can you remove those 2 older ones for testing purposes ?
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
(A works at time B) && (time C > time B ) ≠ (A works at time C)
Offline
I have no real idea, but it would for sure be helpful, if you could post your Xorg.0.log. And maybe it is also useful to know which application you are trying to run...
Here is the pastebin of the Xorg log:
https://pastebin.com/gkxG0hVa
I am really bad with X, so I don't know what I should be looking for.
As for the app: eventually I want to have a go at TensorFlow, but my current tests are with leela-zero, which is a deep-learning Go|Baduk|Weiqi playing engine: https://aur.archlinux.org/packages/leela-zero/
Also, should X be restarted after every package addition/removal or every tweak?
Offline
No idea if it matters, but libxnvctrl and nvidia-390xx-settings are an older version then the others .
390.129 vs 390.132 .Can you remove those 2 older ones for testing purposes ?
I tried, it seems to have no effect. I still get:
"Number of platforms 0"
Offline
I am not sure how the issue sorted itself out, but after reinstalling cuda and going through a couple of reboots I now have both opencCl and cuda working
Last edited by stefano (2020-02-20 16:07:03)
Offline