Solved quarterly now using Nvidia 570.124.04 and Cuda 12.8

NapoleonWils0n

NapoleonWils0n

Just upgraded to the latest quarterly branch

and the nvidia driver and cuda have got an upgrade

Nvidia 570.124.04 and Cuda 12.8

im going install the matching Nvidia driver in a Rocky Linux jail
with the nvidia-cuda-toolkit 12.8 and see if i can get cuda working in the jail

cuda in a jail worked with the 535 driver but not the 550 driver

Code: 
nv-sglrun nvidia-smi
/usr/local/lib/libc6-shim/libc6.so: shim init
Mon Apr  7 19:49:15 2025
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 570.124.04             Driver Version: 570.124.04     CUDA Version: 12.8     |
|-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA GeForce GTX 1650        Off |   00000000:01:00.0 Off |                  N/A |
| N/A   45C    P8              1W /   50W |       1MiB /   4096MiB |      0%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+

+-----------------------------------------------------------------------------------------+
| Processes:                                                                              |
|  GPU   GI   CI              PID   Type   Process name                        GPU Memory |
|        ID   ID                                                               Usage      |
|=========================================================================================|
|  No running processes found                                                             |
+-----------------------------------------------------------------------------------------+
 
I'm trying to get CUDA work wih pytorch. It seems the driver fails to initialize the device:

Code: 
CUDA initialization: Unexpected error from cudaGetDeviceCount()

I would try to run a small CUDA sample before trying to run a PyTorch workload. Do you know if between the packages or ports there is some tool that I can use to simulate and check if CUDA is working great ?

For example on FreeBSD nvcc does not exists for me :

marietto@marietto:/usr/home/marietto/Desktop $ nvcc
sh: nvcc: not found
Click to expand...

what port should I install to have nvcc ?
 
Code: 
CUDA initialization: Unexpected error from cudaGetDeviceCount()

thats means cuda isnt working
because it cant count the cuda device

normally you use python and torch to check if cuda is working

you could try handbrake and see if there is an nvenc encoding option listed under the hardware presets
and then try and encode a video with it
 
I suspect that they should be conducted even on FreeBSD.
Click to expand...

you need to test if cuda is working with the linuxulator
because thats what you are going to be using

Code: 
CUDA initialization: Unexpected error from cudaGetDeviceCount()

if python gives an error checking for cuda
then stable diffusion and comfy-ui wont work because they use python

trying another tool wont make any difference

you have to get get python working with cuda
 
ZioMario said:
I'm looking for some official nvidia tool to make a test to CUDA...
Click to expand...
There's nothing for native FreeBSD, as nvidia doesn't support CUDA on FreeBSD officially. And if nvidia official tool for Linux requires kmods not ported to FreeBSD (uvm for example), it shouldn't work even on Linuxulatro.
So we don't include known-not-to-work-on-FreeBSD tools in Linux version of driver packages in x11/linux-nvidia-libs. So as anything requred to build on install (as they make it far more complexed for us).
 
The whole point of the experiment is to understand if the error is caused by the python or pytorch version used or by the libc6 shim of any other mechanism used on FreeBSD that allows us to detect CUDA correctly.

Furthermore,I would like to know why I don't have nvcc :

marietto@marietto:/usr/home/marietto/Desktop $ nvcc
sh: nvcc: not found
Click to expand...

what port should I install to have it in the system ? Is nvcc supposed to run in the Linuxulator or in FreeBSD ?
 
With :

Code: 
# nv-sglrun nvidia-smi

/usr/local/lib/libc6-shim/libc6.so: shim init
Fri May  9 01:31:16 2025      
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 525.78.01    Driver Version: 525.78.01    CUDA Version: 12.0

it works correctly :

Code: 
# nv-sglrun ffmpeg -i sample_1280x720_surfing_with_audio.mkv -vcodec hevc_nvenc -acodec copy test.mkv

/usr/local/lib/libc6-shim/libc6.so: shim init
ffmpeg version 6.1.2 Copyright (c) 2000-2024 the FFmpeg developers
  built with FreeBSD clang version 18.1.6 (https://github.com/llvm/llvm-project.git llvmorg-18.1.6-0-g1118c2e05e67)
  configuration: --prefix=/usr/local --mandir=/usr/local/share/man --datadir=/usr/local/share/ffmpeg --docdir=/usr/local/share/doc/ffmpeg --pkgconfigdir=/usr/local/libdata/pkgconfig --disable-static --disable-libcelt --enable-shared --enable-pic --enable-gpl --cc=cc --cxx=c++ --disable-alsa --disable-libopencore-amrnb --disable-libopencore-amrwb --enable-libaom --disable-libaribb24 --disable-libaribcaption --enable-asm --enable-libass --disable-libbs2b --disable-libcaca --disable-libcdio --disable-libcodec2 --enable-libdav1d --disable-libdavs2 --disable-libdc1394 --disable-debug --enable-htmlpages --enable-libdrm --disable-libfdk-aac --disable-libflite --enable-fontconfig --enable-libfreetype --enable-frei0r --disable-libfribidi --disable-gcrypt --disable-libglslang --disable-libgme --enable-gmp --enable-gnutls --enable-version3 --disable-libgsm --enable-libharfbuzz --enable-iconv --disable-libilbc --disable-libjack --enable-libjxl --disable-libklvanc --disable-libkvazaar --disable-ladspa --enable-libmp3lame --enable-lcms2 --disable-liblensfun --disable-libbluray --enable-libplacebo --disable-librsvg --enable-libxml2 --enable-lto --disable-lv2 --disable-mbedtls --disable-libmfx --disable-libmodplug --disable-libmysofa --enable-network --disable-nonfree --enable-nvenc --disable-openal --disable-opencl --disable-opengl --disable-libopenh264 --disable-libopenjpeg --disable-libopenmpt --disable-openssl --disable-libopenvino --enable-optimizations --enable-libopus --disable-pocketsphinx --disable-libpulse --disable-librabbitmq --disable-librav1e --disable-librist --enable-runtime-cpudetect --disable-librubberband --disable-sdl2 --enable-libshaderc --disable-libsmbclient --disable-libsnappy --disable-sndio --disable-libsoxr --disable-libspeex --disable-libsrt --disable-libssh --enable-libsvtav1 --disable-libtensorflow --disable-libtesseract --disable-libtheora --disable-libtwolame --disable-libuavs3d --enable-libv4l2 --enable-vaapi --disable-vapoursynth --enable-vdpau --disable-libvidstab --enable-libvmaf --enable-libvorbis --disable-libvo-amrwbenc --disable-libvpl --enable-libvpx --enable-vulkan --enable-libwebp --enable-libx264 --enable-libx265 --disable-libxavs2 --enable-libxcb --disable-libxvid --disable-outdev=xv --disable-libzimg --disable-libzmq --disable-libzvbi
  libavutil      58. 29.100 / 58. 29.100
  libavcodec     60. 31.102 / 60. 31.102
  libavformat    60. 16.100 / 60. 16.100
  libavdevice    60.  3.100 / 60.  3.100
  libavfilter     9. 12.100 /  9. 12.100
  libswscale      7.  5.100 /  7.  5.100
  libswresample   4. 12.100 /  4. 12.100
  libpostproc    57.  3.100 / 57.  3.100
Input #0, matroska,webm, from 'sample_1280x720_surfing_with_audio.mkv':
  Metadata:
    COMPATIBLE_BRANDS: isomiso2avc1mp41
    MAJOR_BRAND     : mp42
    MINOR_VERSION   : 512
    ENCODER         : Lavf58.45.100
  Duration: 00:03:03.13, start: 0.000000, bitrate: 3111 kb/s
  Stream #0:0: Video: h264 (High), yuv420p(progressive), 1280x720 [SAR 1:1 DAR 16:9], 23.98 fps, 23.98 tbr, 1k tbn (default)
    Metadata:
      HANDLER_NAME    : VideoHandler
      ENCODER         : Lavc58.91.100 libx264
      DURATION        : 00:03:03.061000000
  Stream #0:1(eng): Audio: vorbis, 48000 Hz, stereo, fltp (default)
    Metadata:
      HANDLER_NAME    : Stereo
      ENCODER         : Lavc58.91.100 libvorbis
      DURATION        : 00:03:03.129000000
Stream mapping:
  Stream #0:0 -> #0:0 (h264 (native) -> hevc (hevc_nvenc))
  Stream #0:1 -> #0:1 (copy)
Press [q] to stop, [?] for help
Output #0, matroska, to 'test.mkv':
  Metadata:
    COMPATIBLE_BRANDS: isomiso2avc1mp41
    MAJOR_BRAND     : mp42
    MINOR_VERSION   : 512
    encoder         : Lavf60.16.100
  Stream #0:0: Video: hevc (Main), yuv420p(progressive), 1280x720 [SAR 1:1 DAR 16:9], q=2-31, 2000 kb/s, 23.98 fps, 1k tbn (default)
    Metadata:
      HANDLER_NAME    : VideoHandler
      DURATION        : 00:03:03.061000000
      encoder         : Lavc60.31.102 hevc_nvenc
    Side data:
      cpb: bitrate max/min/avg: 0/0/2000000 buffer size: 4000000 vbv_delay: N/A
  Stream #0:1(eng): Audio: vorbis (oV[0][0] / 0x566F), 48000 Hz, stereo, fltp (default)
    Metadata:
      HANDLER_NAME    : Stereo
      ENCODER         : Lavc58.91.100 libvorbis
      DURATION        : 00:03:03.129000000
[out#0/matroska @ 0x85fa174c0] video:42344kB audio:2371kB subtitle:0kB other streams:0kB global headers:4kB muxing overhead: 0.222003%
frame= 4389 fps=669 q=17.0 Lsize=   44815kB time=00:03:03.11 bitrate=2004.9kbits/s speed=27.9x

We will see if it will work good even with a greater nvidia driver version...
 
Code: 
# nv-sglrun nvidia-smi
/usr/local/lib/libc6-shim/libc6.so: shim init

Fri May  9 01:59:10 2025     
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 565.77                 Driver Version: 565.77         CUDA Version: 12.7     |
|-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA GeForce GTX 1060 3GB    Off |   00000000:01:00.0  On |                  N/A |
| 55%   38C    P0             26W /  120W |     131MiB /   3072MiB |      0%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+
|   1  NVIDIA GeForce RTX 2080 Ti     Off |   00000000:02:00.0 Off |                  N/A |
| 31%   42C    P8             21W /  250W |       2MiB /  11264MiB |      0%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+
                                                                                       
+-----------------------------------------------------------------------------------------+
| Processes:                                                                              |
|  GPU   GI   CI        PID   Type   Process name                              GPU Memory |
|        ID   ID                                                               Usage      |
|=========================================================================================|
+-----------------------------------------------------------------------------------------+

# nv-sglrun ffmpeg -i sample_1280x720_surfing_with_audio.mkv -vcodec hevc_nvenc -acodec copy test.mkv

/usr/local/lib/libc6-shim/libc6.so: shim init
ffmpeg version 6.1.2 Copyright (c) 2000-2024 the FFmpeg developers
  built with FreeBSD clang version 18.1.6 (https://github.com/llvm/llvm-project.git llvmorg-18.1.6-0-g1118c2e05e67)
  configuration: --prefix=/usr/local --mandir=/usr/local/share/man --datadir=/usr/local/share/ffmpeg --docdir=/usr/local/share/doc/ffmpeg --pkgconfigdir=/usr/local/libdata/pkgconfig --disable-static --disable-libcelt --enable-shared --enable-pic --enable-gpl --cc=cc --cxx=c++ --disable-alsa --disable-libopencore-amrnb --disable-libopencore-amrwb --enable-libaom --disable-libaribb24 --disable-libaribcaption --enable-asm --enable-libass --disable-libbs2b --disable-libcaca --disable-libcdio --disable-libcodec2 --enable-libdav1d --disable-libdavs2 --disable-libdc1394 --disable-debug --enable-htmlpages --enable-libdrm --disable-libfdk-aac --disable-libflite --enable-fontconfig --enable-libfreetype --enable-frei0r --disable-libfribidi --disable-gcrypt --disable-libglslang --disable-libgme --enable-gmp --enable-gnutls --enable-version3 --disable-libgsm --enable-libharfbuzz --enable-iconv --disable-libilbc --disable-libjack --enable-libjxl --disable-libklvanc --disable-libkvazaar --disable-ladspa --enable-libmp3lame --enable-lcms2 --disable-liblensfun --disable-libbluray --enable-libplacebo --disable-librsvg --enable-libxml2 --enable-lto --disable-lv2 --disable-mbedtls --disable-libmfx --disable-libmodplug --disable-libmysofa --enable-network --disable-nonfree --enable-nvenc --disable-openal --disable-opencl --disable-opengl --disable-libopenh264 --disable-libopenjpeg --disable-libopenmpt --disable-openssl --disable-libopenvino --enable-optimizations --enable-libopus --disable-pocketsphinx --disable-libpulse --disable-librabbitmq --disable-librav1e --disable-librist --enable-runtime-cpudetect --disable-librubberband --disable-sdl2 --enable-libshaderc --disable-libsmbclient --disable-libsnappy --disable-sndio --disable-libsoxr --disable-libspeex --disable-libsrt --disable-libssh --enable-libsvtav1 --disable-libtensorflow --disable-libtesseract --disable-libtheora --disable-libtwolame --disable-libuavs3d --enable-libv4l2 --enable-vaapi --disable-vapoursynth --enable-vdpau --disable-libvidstab --enable-libvmaf --enable-libvorbis --disable-libvo-amrwbenc --disable-libvpl --enable-libvpx --enable-vulkan --enable-libwebp --enable-libx264 --enable-libx265 --disable-libxavs2 --enable-libxcb --disable-libxvid --disable-outdev=xv --disable-libzimg --disable-libzmq --disable-libzvbi
  libavutil      58. 29.100 / 58. 29.100
  libavcodec     60. 31.102 / 60. 31.102
  libavformat    60. 16.100 / 60. 16.100
  libavdevice    60.  3.100 / 60.  3.100
  libavfilter     9. 12.100 /  9. 12.100
  libswscale      7.  5.100 /  7.  5.100
  libswresample   4. 12.100 /  4. 12.100
  libpostproc    57.  3.100 / 57.  3.100
Input #0, matroska,webm, from 'sample_1280x720_surfing_with_audio.mkv':
  Metadata:
    COMPATIBLE_BRANDS: isomiso2avc1mp41
    MAJOR_BRAND     : mp42
    MINOR_VERSION   : 512
    ENCODER         : Lavf58.45.100
  Duration: 00:03:03.13, start: 0.000000, bitrate: 3111 kb/s
  Stream #0:0: Video: h264 (High), yuv420p(progressive), 1280x720 [SAR 1:1 DAR 16:9], 23.98 fps, 23.98 tbr, 1k tbn (default)
    Metadata:
      HANDLER_NAME    : VideoHandler
      ENCODER         : Lavc58.91.100 libx264
      DURATION        : 00:03:03.061000000
  Stream #0:1(eng): Audio: vorbis, 48000 Hz, stereo, fltp (default)
    Metadata:
      HANDLER_NAME    : Stereo
      ENCODER         : Lavc58.91.100 libvorbis
      DURATION        : 00:03:03.129000000
File 'test.mkv' already exists. Overwrite? [y/N] y
Stream mapping:
  Stream #0:0 -> #0:0 (h264 (native) -> hevc (hevc_nvenc))
  Stream #0:1 -> #0:1 (copy)
Press [q] to stop, [?] for help
Output #0, matroska, to 'test.mkv':
  Metadata:
    COMPATIBLE_BRANDS: isomiso2avc1mp41
    MAJOR_BRAND     : mp42
    MINOR_VERSION   : 512
    encoder         : Lavf60.16.100
  Stream #0:0: Video: hevc (Main), yuv420p(progressive), 1280x720 [SAR 1:1 DAR 16:9], q=2-31, 2000 kb/s, 23.98 fps, 1k tbn (default)
    Metadata:
      HANDLER_NAME    : VideoHandler
      DURATION        : 00:03:03.061000000
      encoder         : Lavc60.31.102 hevc_nvenc
    Side data:
      cpb: bitrate max/min/avg: 0/0/2000000 buffer size: 4000000 vbv_delay: N/A
  Stream #0:1(eng): Audio: vorbis (oV[0][0] / 0x566F), 48000 Hz, stereo, fltp (default)
    Metadata:
      HANDLER_NAME    : Stereo
      ENCODER         : Lavc58.91.100 libvorbis
      DURATION        : 00:03:03.129000000
[out#0/matroska @ 0x8648f94c0] video:42256kB audio:2371kB subtitle:0kB other streams:0kB global headers:4kB muxing overhead: 0.222391%
frame= 4389 fps=341 q=17.0 Lsize=   44726kB time=00:03:03.11 bitrate=2000.9kbits/s speed=14.2x
 
Code: 
marietto@marietto:/usr/home/marietto/Desktop $ nv-sglrun clpeak

/usr/local/lib/libc6-shim/libc6.so: shim init

Platform: NVIDIA CUDA
  Device: NVIDIA GeForce RTX 2080 Ti
    Driver version  : 565.77 (FreeBSD)
    Compute units   : 68
    Clock frequency : 1545 MHz

    Global memory bandwidth (GBPS)
      float   : 505.42
      float2  : 533.59
      float4  : 549.18
      float8  : 551.64
      float16 : 491.24
 
I think I have passed all CUDA tests now :

Code: 
marietto@marietto:/usr/local/lib/libc6-shim $ env __NV_PRIME_RENDER_OFFLOAD=1 __GLX_VENDOR_LIBRARY_NAME=nvidia LD_LIBMAP="`nv-sglrun printenv LD_LIBMAP | grep -v libGL`" blender
/usr/local/lib/libc6-shim/libc6.so: shim init

Screenshot_2025-05-09_02-33-31.jpg


so,if pytorch+cuda will not work,it will not depend about the wrong configuration of the shim and or of the nvidia-driver or the CUDA libraries,but it may probably be a problem of pytorch and or python. Right ?
 
We don't have good news :

Hi @Marietto2008,

This definitely looks like an interesting experiment! As you're aware of, FreeBSD is not a platform support by conda, and additionally, the Pytorch Anaconda channel has officially been deprecated. All of this means that you should take up what issues you're having with Pytorch itself with the official Pytorch team and those who maintain the packages on PyPI.

I hope you understand. I will now be closing this issue as "off-topic" because is it not relevant to this project, and there's nothing more for us to add.
Click to expand...

Source :

github.com

The installation of Stable Diffusion and / or ComfyUI on FreeBSD using the Miniconda Installer and the Linuxulaor fails with a CUDA error · Issue #14809 · conda/conda

Checklist I added a descriptive title I searched open reports and couldn't find a duplicate What happened? Hello. I've installed the nvidia driver and the linux-nvidia-libs vers. 565.77 and I start...
github.com github.com

PS :

Who maintains the packages on PyPI ?
 
We don't have good news :
Click to expand...

there's no we
we refers to more than one person

you havent passed any tests
because python using the linuxulator doesnt work with cuda ( except with the 535 driver )

testing with blender or ffmpeg is irrelevant

if python running using the linuxulator doesnt work with cuda
then stable diffusion and comfy-ui wont work
 
Yes,I've worked on the Popov's suggestion. Everything went good until I got this error :

Code: 
marietto@marietto:~ $ ai-ml-env-python stable-diffusion-sample.py

/compat/linux/bin/python3: error while loading shared libraries: libpython3.9.so.1.0: cannot open shared object file: No such file or directory

science/linux-ai-ml-env introduced there seems to be what you want, isn't it ?
Click to expand...

Not sure. I need to understand if it allows me to do what I want,that's to run ComfyUI using the 3D acceleration of CUDA on my GPU (RTX 2080 ti).

I don't understand the reason of the error,because the missing libraries are there :

Code: 
marietto@marietto:~ $ find /compat/linux -name libpython3.9.so.1.0

/compat/linux/usr/lib/libpython3.9.so.1.0
/compat/linux/usr/lib64/libpython3.9.so.1.0
 
im closing this thread

because ZioMario started acting like a bell end in a private chat

i said i spent several weeks trying all the different permutations

different drivers, versions of linux, using jails, the linuxulator
and the errors i got, and none of the combinations worked

the only driver that ever working with Cuda in a jail or the lixulator was the Nvidia 535 driver
i havent tested the 570 driver because it crashed my machine

but none of the other divers worked

he then spent the next week trying all the things i said didnt work
and asking why he was getting all the errors i had already mentioned

i then explained i managed to get Davinci Resolve working with Cuda
using the linuxulator, and Ubuntu Jail and a Rocky Linux jail

using the X11 socket to forward the display from the linuxulator or jail,
and the pulseaudio socket for the audio

and then using creating desktop launcher that worked on freebsd to launch Davinci Resolve
from the linuxulator or a jail, forwards the display and start the pulseaudio server for audio

so i have a lot of experience on getting cuda working using the linuxulator or a jail

ziomario gave me once piece of code for manually installing the Nvidia driver
which i has already seen during my searches, which i thanked him for

after i mentioned i was the first person to get Davinci Resolve working with Cuda on Freebsd
using the linuxulator and a jail

he then said thats rubbish we collaborated on getting Davinci Resolve working
as i said he gave me one bit of code

you can check the threads the old threads
one of then is about 30 pages of me responding to myself

i should also mention he tried very instantly to give a csh config file for running Resolve in the linuxulator
because he used the csh shell, and i was using the export command with a sh script

but he didnt know how to set variables in the shell he was using

csh example

Code: 
setenv  DISPLAY :1

where as he was trying

Code: 
setenv  DISPLAY=:1

so i said.
no equals

meaning you dont use the = to set the variable with csh shell

if someone doesnt know the most basic thing you do configuring your shell
which is setting variables, then that tells you a lot about their level of coding ability

you will find lots of his thread, where he doesnt read the man page
or even google for errors

so its a bit rich to suddenly claim he collaborated getting Davinci Resolve working with Cuda on Freebsd

how can someone who has never installed Davinci Resolve,
written any of the code or documentation collaborated getting it working
 
NapoleonWils0n said:
i havent tested the 570 driver because it crashed my machine
Click to expand...
Does this thread seem relevant for you?

GSP firmware support introduced 560 series and later are known to prevent suspend/resume from working (for anyone possible to do so, unlike me, with older versions of drivers), it was forcibly disabled at commit 9c0e0196bdc6.

But after that, the above mentioned thread was opened.

Unfortunately the above mentioned commit was too aggressive that it no longer allow enabling GSP firmware via tunable hw.nvidia.registry.EnableGpuFirmware set to 1.

So I've opened a review D50053 to make it gentle, landed at commit 637e9e68f1a2. Now the above mentioned tunable should work again.

And x11/nvidia-driver and friends are upgraded to 570.144 now.
 
T-Aoki : If I have understood well,if now I will install the 570.144 driver,I will be able to fix my problem,too ? I mean,this one :

Code: 
CUDA initialization: Unexpected error from cudaGetDeviceCount()

? thanks.
 
ZioMario said:
T-Aoki : If I have understood well,if now I will install the 570.144 driver,I will be able to fix my problem,too ? I mean,this one :

Code: 
CUDA initialization: Unexpected error from cudaGetDeviceCount()

? thanks.
Click to expand...
Sorry, not at all sure. I don't have proper environment to test CUDA and/or GSP.
The only thing the Release Highlights states for 570.144 is Fixed an issue that could cause render-offloaded applications using KDE Frameworks 6 to crash. Same as Linux version.

I've been using 570.144 without new issues until I was working on upgrading ports to it, but on GPU (physically) without GSP, Quadro P1000 and I don't have any CUDA codes to test. Simply making x11/linux-nvidia-libs to be in sync with x11/nvidia-driver, which is required.

And latest Beta branch of driver 575.51.02 can be built/installed/run using current forms of ports tree (main [aka latest] branch). Supports to allow it are already incorporated and what pruned were supports for old versions that do not affect in-tree legacy versions.
 
Is this expected ? (CUDA not recognized)

Code: 
marietto@marietto:/usr/home/marietto/Desktop $ nv-sglrun nvidia-smi

/usr/local/lib/libc6-shim/libc6.so: shim init
Sun May 11 04:33:26 2025      
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 570.144                Driver Version: 570.144        CUDA Version: N/A      |
|-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA GeForce GTX 1060 3GB    Off |   00000000:01:00.0  On |                  N/A |
| 55%   35C    P8              8W /  120W |     329MiB /   3072MiB |      5%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+
|   1  NVIDIA GeForce RTX 2080 Ti     Off |   00000000:02:00.0 Off |                  N/A |
| 31%   41C    P8             21W /  250W |       2MiB /  11264MiB |      0%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+
                                                                                        
+-----------------------------------------------------------------------------------------+
| Processes:                                                                              |
|  GPU   GI   CI              PID   Type   Process name                        GPU Memory |
|        ID   ID                                                               Usage      |
|=========================================================================================|
+-----------------------------------------------------------------------------------------+
 
Works on 15-current
It did not work when I used 570.124 nvidia-driver and 570.144 nvidia-drm
needs to be exaclty same version apperantly.
I had buildt the NVIDIA-Driver a few days before and the updated ports so Nvidia DRM became newer version.

$ nv-sglrun nvidia-smi
/usr/local/lib/libc6-shim/libc6.so: shim init

Mon May 12 22:59:53 2025
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 570.144 Driver Version: 570.144 CUDA Version: 12.8 |
|-----------------------------------------+------------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+========================+======================|
| 0 Quadro T1000 Off | 00000000:01:00.0 On | N/A |
| N/A 40C P8 5W / 50W | 618MiB / 4096MiB | 6% Default |
| | | N/A |
+-----------------------------------------+------------------------+----------------------+

+-----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
+-----------------------------------------------------------------------------------------+
$ nv-sglrun clinfo
/usr/local/lib/libc6-shim/libc6.so: shim init
Number of platforms 1
Platform Name NVIDIA CUDA
Platform Vendor NVIDIA Corporation
Platform Version OpenCL 3.0 CUDA 12.8.97
Platform Profile FULL_PROFILE
 
drsnx60 said:
Works on 15-current
It did not work when I used 570.124 nvidia-driver and 570.144 nvidia-drm
needs to be exaclty same version apperantly.
I had buildt the NVIDIA-Driver a few days before and the updated ports so Nvidia DRM became newer version.

$ nv-sglrun nvidia-smi
/usr/local/lib/libc6-shim/libc6.so: shim init

Mon May 12 22:59:53 2025
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 570.144 Driver Version: 570.144 CUDA Version: 12.8 |
|-----------------------------------------+------------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+========================+======================|
| 0 Quadro T1000 Off | 00000000:01:00.0 On | N/A |
| N/A 40C P8 5W / 50W | 618MiB / 4096MiB | 6% Default |
| | | N/A |
+-----------------------------------------+------------------------+----------------------+

+-----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
+-----------------------------------------------------------------------------------------+
$ nv-sglrun clinfo
/usr/local/lib/libc6-shim/libc6.so: shim init
Number of platforms 1
Platform Name NVIDIA CUDA
Platform Vendor NVIDIA Corporation
Platform Version OpenCL 3.0 CUDA 12.8.97
Platform Profile FULL_PROFILE
Click to expand...

Can u explain what worked and what didn't work for you ?
 
You must log in or register to reply here.
Back
Top