# Ubuntu 22.04 fresh install with conda/numba does not see GPU **URL:** \ **Category:** Numba\ **Created:** [March 1, 2023, 12:37am UTC](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806 "2023-03-01T00:37:13Z")\ **Posts on this page:** 20\ **Page:** 1
**Author:** ![kristof-333](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/kristof-333/32/789_2.png) [@kristof-333](https://numba.discourse.group/u/kristof-333)\ **Post date:** [March 1, 2023, 12:37am UTC](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806/1 "2023-03-01T00:37:13Z")
My apology for rehashing perhaps benign topic: Not being able to run numba on Ubuntu. It is a fresh install of Ubuntu with fresh install of conda/numba. I also have Nvidia CUDA installed and can successfully build and run different examples. It is very likely my situation is due to my inexperience since this is the first time I am doing this. The data **Time Stamp** Report started (local time) : 2023-02-28 16:16:55.961018 UTC start time : 2023-03-01 00:16:55.961021 Running time (s) : 2.635238 **Hardware Information** Machine : x86\_64 CPU Name : skylake CPU Count : 16 Number of accessible CPUs : 16 List of accessible CPUs cores : 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 CFS Restrictions (CPUs worth of runtime) : None CPU Features : 64bit adx aes avx avx2 bmi bmi2 clflushopt cmov cx16 cx8 f16c fma fsgsbase fxsr invpcid lzcnt mmx movbe pclmul popcnt prfchw rdrnd rdseed sahf sgx sse sse2 sse3 sse4.1 sse4.2 ssse3 xsave xsavec xsaveopt xsaves Memory Total (MB) : 64238 Memory Available (MB) : 56686 **OS Information** Platform Name : Linux-5.19.0-32-generic-x86\_64-with-glibc2.35 Platform Release : 5.19.0-32-generic OS Name : Linux OS Version : #33~22.04.1-Ubuntu SMP PREEMPT\_DYNAMIC Mon Jan 30 17:03:34 UTC 2 OS Specific Version : ? Libc Version : glibc 2.35 **Python Information** Python Compiler : GCC 11.2.0 Python Implementation : CPython Python Version : 3.9.13 Python Locale : en\_CA.UTF-8 **Numba Toolchain Versions** Numba Version : 0.55.1 llvmlite Version : 0.38.0 **LLVM Information** LLVM Version : 11.1.0 **CUDA Information** CUDA Device Initialized : False CUDA Driver Version : ? CUDA Runtime Version : ? CUDA NVIDIA Bindings Available : ? CUDA NVIDIA Bindings In Use : ? CUDA Detect Output: None CUDA Libraries Test Output: None **SVML Information** SVML State, config.USING\_SVML : False SVML Library Loaded : False llvmlite Using SVML Patched LLVM : True SVML Operational : False **Threading Layer Information** TBB Threading Layer Available : True ±-\>TBB imported successfully. OpenMP Threading Layer Available : True ±-\>Vendor: GNU Workqueue Threading Layer Available : True ±-\>Workqueue imported successfully. **Numba Environment Variable Information** NUMBA\_CUDA\_USE\_NVIDIA\_BINDINGS : 1 **Conda Information** Conda Build : 3.22.0 Conda Env : 23.1.0 Conda Platform : linux-64 Conda Python Version : 3.9.13.final.0 Conda Root Writable : True **Installed Packages** \_ipyw\_jlab\_nb\_ext\_conf 0.1.0 py39h06a4308\_1 \_libgcc\_mutex 0.1 main \_openmp\_mutex 5.1 1\_gnu alabaster 0.7.12 pyhd3eb1b0\_0 anaconda 2022.10 py39\_0 anaconda-client 1.11.0 py39h06a4308\_0 anaconda-navigator 2.4.0 py39h06a4308\_0 anaconda-project 0.11.1 py39h06a4308\_0 anyio 3.5.0 py39h06a4308\_0 appdirs 1.4.4 pyhd3eb1b0\_0 argon2-cffi 21.3.0 pyhd3eb1b0\_0 argon2-cffi-bindings 21.2.0 py39h7f8727e\_0 arrow 1.2.2 pyhd3eb1b0\_0 astroid 2.11.7 py39h06a4308\_0 astropy 5.1 py39h7deecbd\_0 atomicwrites 1.4.0 py\_0 attrs 21.4.0 pyhd3eb1b0\_0 automat 20.2.0 py\_0 autopep8 1.6.0 pyhd3eb1b0\_1 babel 2.9.1 pyhd3eb1b0\_0 backcall 0.2.0 pyhd3eb1b0\_0 backports 1.1 pyhd3eb1b0\_0 backports.functools\_lru\_cache 1.6.4 pyhd3eb1b0\_0 backports.tempfile 1.0 pyhd3eb1b0\_1 backports.weakref 1.0.post1 py\_1 bcrypt 3.2.0 py39h5eee18b\_1 beautifulsoup4 4.11.1 py39h06a4308\_0 binaryornot 0.4.4 pyhd3eb1b0\_1 bitarray 2.5.1 py39h5eee18b\_0 bkcharts 0.2 py39h06a4308\_1 black 22.6.0 py39h06a4308\_0 blas 1.0 mkl bleach 4.1.0 pyhd3eb1b0\_0 blosc 1.21.0 h4ff587b\_1 bokeh 2.4.3 py39h06a4308\_0 boto3 1.24.28 py39h06a4308\_0 botocore 1.27.28 py39h06a4308\_0 bottleneck 1.3.5 py39h7deecbd\_0 brotli 1.0.9 h5eee18b\_7 brotli-bin 1.0.9 h5eee18b\_7 brotlipy 0.7.0 py39h27cfd23\_1003 brunsli 0.1 h2531618\_0 bzip2 1.0.8 h7b6447c\_0 c-ares 1.18.1 h7f8727e\_0 ca-certificates 2022.07.19 h06a4308\_0 certifi 2022.9.14 py39h06a4308\_0 cffi 1.15.1 py39h74dc2b5\_0 cfitsio 3.470 h5893167\_7 chardet 4.0.0 py39h06a4308\_1003 charls 2.2.0 h2531618\_0 charset-normalizer 2.0.4 pyhd3eb1b0\_0 click 8.0.4 py39h06a4308\_0 cloudpickle 2.0.0 pyhd3eb1b0\_0 clyent 1.2.2 py39h06a4308\_1 colorama 0.4.5 py39h06a4308\_0 colorcet 3.0.0 py39h06a4308\_0 conda 23.1.0 py39h06a4308\_0 conda-build 3.22.0 py39h06a4308\_0 conda-content-trust 0.1.3 py39h06a4308\_0 conda-env 2.6.0 1 conda-pack 0.6.0 pyhd3eb1b0\_0 conda-package-handling 1.9.0 py39h5eee18b\_0 conda-repo-cli 1.0.20 py39h06a4308\_0 conda-token 0.4.0 pyhd3eb1b0\_0 conda-verify 3.4.2 py\_1 constantly 15.1.0 pyh2b92418\_0 cookiecutter 1.7.3 pyhd3eb1b0\_0 cryptography 37.0.1 py39h9ce1e76\_0 cssselect 1.1.0 pyhd3eb1b0\_0 cudatoolkit 11.3.1 h2bc3f7f\_2 curl 7.84.0 h5eee18b\_0 cycler 0.11.0 pyhd3eb1b0\_0 cython 0.29.32 py39h6a678d5\_0 cytoolz 0.11.0 py39h27cfd23\_0 daal4py 2021.6.0 py39h79cecc1\_1 dal 2021.6.0 hdb19cb5\_916 dask 2022.7.0 py39h06a4308\_0 dask-core 2022.7.0 py39h06a4308\_0 dataclasses 0.8 pyh6d0b6a4\_7 datashader 0.14.1 py39h06a4308\_0 datashape 0.5.4 py39h06a4308\_1 dbus 1.13.18 hb2f20db\_0 debugpy 1.5.1 py39h295c915\_0 decorator 5.1.1 pyhd3eb1b0\_0 defusedxml 0.7.1 pyhd3eb1b0\_0 diff-match-patch 20200713 pyhd3eb1b0\_0 dill 0.3.4 pyhd3eb1b0\_0 distributed 2022.7.0 py39h06a4308\_0 docutils 0.18.1 py39h06a4308\_3 entrypoints 0.4 py39h06a4308\_0 et\_xmlfile 1.1.0 py39h06a4308\_0 expat 2.4.9 h6a678d5\_0 fftw 3.3.9 h27cfd23\_1 filelock 3.6.0 pyhd3eb1b0\_0 flake8 4.0.1 pyhd3eb1b0\_1 flask 1.1.2 pyhd3eb1b0\_0 fontconfig 2.13.1 h6c09931\_0 fonttools 4.25.0 pyhd3eb1b0\_0 freetype 2.11.0 h70c0345\_0 fsspec 2022.7.1 py39h06a4308\_0 future 0.18.2 py39h06a4308\_1 gensim 4.1.2 py39h295c915\_0 giflib 5.2.1 h7b6447c\_0 glib 2.69.1 h4ff587b\_1 glob2 0.7 pyhd3eb1b0\_0 gmp 6.2.1 h295c915\_3 gmpy2 2.1.2 py39heeb90bb\_0 greenlet 1.1.1 py39h295c915\_0 gst-plugins-base 1.14.0 h8213a91\_2 gstreamer 1.14.0 h28cd5cc\_2 h5py 3.7.0 py39h737f45e\_0 hdf5 1.10.6 h3ffc7dd\_1 heapdict 1.0.1 pyhd3eb1b0\_0 holoviews 1.15.0 py39h06a4308\_0 hvplot 0.8.0 py39h06a4308\_0 hyperlink 21.0.0 pyhd3eb1b0\_0 icu 58.2 he6710b0\_3 idna 3.3 pyhd3eb1b0\_0 imagecodecs 2021.8.26 py39hf0132c2\_1 imageio 2.19.3 py39h06a4308\_0 imagesize 1.4.1 py39h06a4308\_0 importlib-metadata 4.11.3 py39h06a4308\_0 importlib\_metadata 4.11.3 hd3eb1b0\_0 incremental 21.3.0 pyhd3eb1b0\_0 inflection 0.5.1 py39h06a4308\_0 iniconfig 1.1.1 pyhd3eb1b0\_0 intake 0.6.5 pyhd3eb1b0\_0 intel-openmp 2021.4.0 h06a4308\_3561 intervaltree 3.1.0 pyhd3eb1b0\_0 ipykernel 6.15.2 py39h06a4308\_0 ipython 7.31.1 py39h06a4308\_1 ipython\_genutils 0.2.0 pyhd3eb1b0\_1 ipywidgets 7.6.5 pyhd3eb1b0\_1 isort 5.9.3 pyhd3eb1b0\_0 itemadapter 0.3.0 pyhd3eb1b0\_0 itemloaders 1.0.4 pyhd3eb1b0\_1 itsdangerous 2.0.1 pyhd3eb1b0\_0 jdcal 1.4.1 pyhd3eb1b0\_0 jedi 0.18.1 py39h06a4308\_1 jeepney 0.7.1 pyhd3eb1b0\_0 jellyfish 0.9.0 py39h7f8727e\_0 jinja2 2.11.3 pyhd3eb1b0\_0 jinja2-time 0.2.0 pyhd3eb1b0\_3 jmespath 0.10.0 pyhd3eb1b0\_0 joblib 1.1.0 pyhd3eb1b0\_0 jpeg 9e h7f8727e\_0 jq 1.6 h27cfd23\_1000 json5 0.9.6 pyhd3eb1b0\_0 jsonschema 4.16.0 py39h06a4308\_0 jupyter 1.0.0 py39h06a4308\_8 jupyter\_client 7.3.4 py39h06a4308\_0 jupyter\_console 6.4.3 pyhd3eb1b0\_0 jupyter\_core 4.11.1 py39h06a4308\_0 jupyter\_server 1.18.1 py39h06a4308\_0 jupyterlab 3.4.4 py39h06a4308\_0 jupyterlab\_pygments 0.1.2 py\_0 jupyterlab\_server 2.10.3 pyhd3eb1b0\_1 jupyterlab\_widgets 1.0.0 pyhd3eb1b0\_1 jxrlib 1.1 h7b6447c\_2 keyring 23.4.0 py39h06a4308\_0 kiwisolver 1.4.2 py39h295c915\_0 krb5 1.19.2 hac12032\_0 lazy-object-proxy 1.6.0 py39h27cfd23\_0 lcms2 2.12 h3be6417\_0 ld\_impl\_linux-64 2.38 h1181459\_1 lerc 3.0 h295c915\_0 libaec 1.0.4 he6710b0\_1 libarchive 3.6.1 hab531cd\_0 libbrotlicommon 1.0.9 h5eee18b\_7 libbrotlidec 1.0.9 h5eee18b\_7 libbrotlienc 1.0.9 h5eee18b\_7 libclang 10.0.1 default\_hb85057a\_2 libcurl 7.84.0 h91b91d3\_0 libdeflate 1.8 h7f8727e\_5 libedit 3.1.20210910 h7f8727e\_0 libev 4.33 h7f8727e\_1 libevent 2.1.12 h8f2d780\_0 libffi 3.3 he6710b0\_2 libgcc-ng 11.2.0 h1234567\_1 libgfortran-ng 11.2.0 h00389a5\_1 libgfortran5 11.2.0 h1234567\_1 libgomp 11.2.0 h1234567\_1 libidn2 2.3.2 h7f8727e\_0 liblief 0.11.5 h295c915\_1 libllvm10 10.0.1 hbcb73fb\_5 libllvm11 11.1.0 h9e868ea\_5 libnghttp2 1.46.0 hce63b2e\_0 libpng 1.6.37 hbc83047\_0 libpq 12.9 h16c4e8d\_3 libsodium 1.0.18 h7b6447c\_0 libspatialindex 1.9.3 h2531618\_0 libssh2 1.10.0 h8f2d780\_0 libstdcxx-ng 11.2.0 h1234567\_1 libtiff 4.4.0 hecacb30\_0 libunistring 0.9.10 h27cfd23\_0 libuuid 1.0.3 h7f8727e\_2 libwebp 1.2.2 h55f646e\_0 libwebp-base 1.2.2 h7f8727e\_0 libxcb 1.15 h7f8727e\_0 libxkbcommon 1.0.1 hfa300c1\_0 libxml2 2.9.14 h74e7548\_0 libxslt 1.1.35 h4e12654\_0 libzopfli 1.0.3 he6710b0\_0 llvmlite 0.38.0 py39h4ff587b\_0 locket 1.0.0 py39h06a4308\_0 lxml 4.9.1 py39h1edc446\_0 lz4 3.1.3 py39h27cfd23\_0 lz4-c 1.9.3 h295c915\_1 lzo 2.10 h7b6447c\_2 markdown 3.3.4 py39h06a4308\_0 markupsafe 2.0.1 py39h27cfd23\_0 matplotlib 3.5.2 py39h06a4308\_0 matplotlib-base 3.5.2 py39hf590b9c\_0 matplotlib-inline 0.1.6 py39h06a4308\_0 mccabe 0.7.0 pyhd3eb1b0\_0 mistune 0.8.4 py39h27cfd23\_1000 mkl 2021.4.0 h06a4308\_640 mkl-service 2.4.0 py39h7f8727e\_0 mkl\_fft 1.3.1 py39hd3c417c\_0 mkl\_random 1.2.2 py39h51133e4\_0 mock 4.0.3 pyhd3eb1b0\_0 mpc 1.1.0 h10f8cd9\_1 mpfr 4.0.2 hb69a4c5\_1 mpi 1.0 mpich mpich 3.3.2 external\_0 mpmath 1.2.1 py39h06a4308\_0 msgpack-python 1.0.3 py39hd09550d\_0 multipledispatch 0.6.0 py39h06a4308\_0 munkres 1.1.4 py\_0 mypy\_extensions 0.4.3 py39h06a4308\_1 navigator-updater 0.3.0 py39h06a4308\_0 nbclassic 0.3.5 pyhd3eb1b0\_0 nbclient 0.5.13 py39h06a4308\_0 nbconvert 6.4.4 py39h06a4308\_0 nbformat 5.5.0 py39h06a4308\_0 ncurses 6.3 h5eee18b\_3 nest-asyncio 1.5.5 py39h06a4308\_0 networkx 2.8.4 py39h06a4308\_0 nltk 3.7 pyhd3eb1b0\_0 nose 1.3.7 pyhd3eb1b0\_1008 notebook 6.4.12 py39h06a4308\_0 nspr 4.33 h295c915\_0 nss 3.74 h0370c37\_0 numba 0.55.1 py39h51133e4\_0 numexpr 2.8.3 py39h807cd23\_0 numpy 1.21.5 py39h6c91a56\_3 numpy-base 1.21.5 py39ha15fc14\_3 numpydoc 1.4.0 py39h06a4308\_0 olefile 0.46 pyhd3eb1b0\_0 oniguruma 6.9.7.1 h27cfd23\_0 openjpeg 2.4.0 h3ad879b\_0 openpyxl 3.0.10 py39h5eee18b\_0 openssl 1.1.1q h7f8727e\_0 packaging 21.3 pyhd3eb1b0\_0 pandas 1.4.4 py39h6a678d5\_0 pandocfilters 1.5.0 pyhd3eb1b0\_0 panel 0.13.1 py39h06a4308\_0 param 1.12.0 pyhd3eb1b0\_0 parsel 1.6.0 py39h06a4308\_0 parso 0.8.3 pyhd3eb1b0\_0 partd 1.2.0 pyhd3eb1b0\_1 patch 2.7.6 h7b6447c\_1001 patchelf 0.13 h295c915\_0 pathlib 1.0.1 pyhd3eb1b0\_1 pathspec 0.9.0 py39h06a4308\_0 patsy 0.5.2 py39h06a4308\_1 pcre 8.45 h295c915\_0 pep8 1.7.1 py39h06a4308\_1 pexpect 4.8.0 pyhd3eb1b0\_3 pickleshare 0.7.5 pyhd3eb1b0\_1003 pillow 9.2.0 py39hace64e9\_1 pip 22.2.2 py39h06a4308\_0 pkginfo 1.8.2 pyhd3eb1b0\_0 platformdirs 2.5.2 py39h06a4308\_0 plotly 5.9.0 py39h06a4308\_0 pluggy 1.0.0 py39h06a4308\_1 ply 3.11 py39h06a4308\_0 poyo 0.5.0 pyhd3eb1b0\_0 prometheus\_client 0.14.1 py39h06a4308\_0 prompt-toolkit 3.0.20 pyhd3eb1b0\_0 prompt\_toolkit 3.0.20 hd3eb1b0\_0 protego 0.1.16 py\_0 psutil 5.9.0 py39h5eee18b\_0 ptyprocess 0.7.0 pyhd3eb1b0\_2 py 1.11.0 pyhd3eb1b0\_0 py-lief 0.11.5 py39h295c915\_1 pyasn1 0.4.8 pyhd3eb1b0\_0 pyasn1-modules 0.2.8 py\_0 pycodestyle 2.8.0 pyhd3eb1b0\_0 pycosat 0.6.3 py39h27cfd23\_0 pycparser 2.21 pyhd3eb1b0\_0 pyct 0.4.8 py39h06a4308\_1 pycurl 7.45.1 py39h8f2d780\_0 pydispatcher 2.0.5 py39h06a4308\_2 pydocstyle 6.1.1 pyhd3eb1b0\_0 pyerfa 2.0.0 py39h27cfd23\_0 pyflakes 2.4.0 pyhd3eb1b0\_0 pygments 2.11.2 pyhd3eb1b0\_0 pyhamcrest 2.0.2 pyhd3eb1b0\_2 pyjwt 2.4.0 py39h06a4308\_0 pylint 2.14.5 py39h06a4308\_0 pyls-spyder 0.4.0 pyhd3eb1b0\_0 pyodbc 4.0.34 py39h6a678d5\_0 pyopenssl 22.0.0 pyhd3eb1b0\_0 pyparsing 3.0.9 py39h06a4308\_0 pyqt 5.15.7 py39h6a678d5\_1 pyqt5-sip 12.11.0 py39h6a678d5\_1 pyqtwebengine 5.15.7 py39h6a678d5\_1 pyrsistent 0.18.0 py39heee7806\_0 pysocks 1.7.1 py39h06a4308\_0 pytables 3.6.1 py39h77479fe\_1 pytest 7.1.2 py39h06a4308\_0 python 3.9.13 haa1d7c7\_1 python-dateutil 2.8.2 pyhd3eb1b0\_0 python-fastjsonschema 2.16.2 py39h06a4308\_0 python-libarchive-c 2.9 pyhd3eb1b0\_1 python-lsp-black 1.2.1 py39h06a4308\_0 python-lsp-jsonrpc 1.0.0 pyhd3eb1b0\_0 python-lsp-server 1.5.0 py39h06a4308\_0 python-slugify 5.0.2 pyhd3eb1b0\_0 python-snappy 0.6.0 py39h2531618\_3 pytz 2022.1 py39h06a4308\_0 pyviz\_comms 2.0.2 pyhd3eb1b0\_0 pywavelets 1.3.0 py39h7f8727e\_0 pyxdg 0.27 pyhd3eb1b0\_0 pyyaml 6.0 py39h7f8727e\_1 … zope.interface 5.4.0 py39h7f8727e\_0 zstd 1.5.2 ha4553b6\_0 No errors reported. **Warning log** Warning (cuda): CUDA device intialisation problem. Message:Error at driver init: Call to cuInit results in CUDA\_ERROR\_NO\_DEVICE (100) Exception class: \ Warning (no file): /sys/fs/cgroup/cpuacct/cpu.cfs\_quota\_us Warning (no file): /sys/fs/cgroup/cpuacct/cpu.cfs\_period\_us * * * GPU or driver and not identified * * * and from nvidia-smi Tue Feb 28 16:04:57 2023 ±----------------------------------------------------------------------------+ | NVIDIA-SMI 525.78.01 Driver Version: 525.78.01 CUDA Version: 12.0 | |-------------------------------±---------------------±---------------------+ | 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 … Off | 00000000:01:00.0 On | N/A | | 0% 43C P0 72W / 290W | 980MiB / 8192MiB | 2% Default | | | | N/A | ±------------------------------±---------------------±---------------------+ ±----------------------------------------------------------------------------+ | Processes: | | GPU GI CI PID Type Process name GPU Memory | | ID ID Usage | |=============================================================================| | 0 N/A N/A 2200 G /usr/lib/xorg/Xorg 450MiB | | 0 N/A N/A 2661 G …ome-remote-desktop-daemon 3MiB | | 0 N/A N/A 2702 G /usr/bin/gnome-shell 74MiB | | 0 N/A N/A 3831 G …918332966337200593,131072 267MiB | | 0 N/A N/A 5659 G …/usr/bin/telegram-desktop 3MiB | | 0 N/A N/A 27825 G …veSuggestionsOnlyOnDemand 42MiB | | 0 N/A N/A 41855 G …b/thunderbird/thunderbird 135MiB | ±----------------------------------------------------------------------------+ ## a simple python program from numba import cuda # Press the green button in the gutter to run the script. if **name** == ‘ **main** ’: print(cuda.gpus) results in error numba.cuda.cudadrv.error.CudaSupportError: Error at driver init: Call to cuInit results in CUDA\_ERROR\_NO\_DEVICE (100) Can anybody offer any suggestions on where to look for an issue or perhaps resolve this problem? ---
**Author:** ![gmarkall](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/gmarkall/32/42_2.png) [@gmarkall](https://numba.discourse.group/u/gmarkall)\ **Post date:** [March 1, 2023, 1:54pm UTC](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806/2 "2023-03-01T13:54:13Z")
You have the environment variable `NUMBA_CUDA_USE_NVIDIA_BINDINGS` set but I can’t see evidence that the NVIDIA cuda-python package is installed. Can you unset that environment variable and test again please? ---
**Author:** ![kristof-333](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/kristof-333/32/789_2.png) [@kristof-333](https://numba.discourse.group/u/kristof-333)\ **Post date:** [March 2, 2023, 1:20am UTC](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806/3 "2023-03-02T01:20:27Z")
Thank you very much … I am in a process of another clear install. I successfully managed to mess up the system while changing nvidia drivers. I will let you know how the behaviour changed once everything is installed. ---
**Author:** ![kristof-333](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/kristof-333/32/789_2.png) [@kristof-333](https://numba.discourse.group/u/kristof-333)\ **Post date:** [March 2, 2023, 4:59pm UTC](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806/4 "2023-03-02T16:59:22Z")
Update: fresh install. I made sure NUMBA\_CUDA\_USE\_NVIDIA\_BINDINGS is unset Python 3.9.13 (main, Aug 25 2022, 23:26:10) [GCC 11.2.0] :: Anaconda, Inc. on linux numba -s **Time Stamp** Report started (local time) : 2023-03-02 08:30:18.438154 UTC start time : 2023-03-02 16:30:18.438157 Running time (s) : 2.492168 **Hardware Information** Machine : x86\_64 CPU Name : skylake CPU Count : 16 Number of accessible CPUs : 16 List of accessible CPUs cores : 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 CFS Restrictions (CPUs worth of runtime) : None CPU Features : 64bit adx aes avx avx2 bmi bmi2 clflushopt cmov cx16 cx8 f16c fma fsgsbase fxsr invpcid lzcnt mmx movbe pclmul popcnt prfchw rdrnd rdseed rtm sahf sgx sse sse2 sse3 sse4.1 sse4.2 ssse3 xsave xsavec xsaveopt xsaves Memory Total (MB) : 64238 Memory Available (MB) : 59022 **OS Information** Platform Name : Linux-5.19.0-35-generic-x86\_64-with-glibc2.35 Platform Release : 5.19.0-35-generic OS Name : Linux OS Version : #36~22.04.1-Ubuntu SMP PREEMPT\_DYNAMIC Fri Feb 17 15:17:25 UTC 2 OS Specific Version : ? Libc Version : glibc 2.35 **Python Information** Python Compiler : GCC 11.2.0 Python Implementation : CPython Python Version : 3.9.13 Python Locale : en\_CA.UTF-8 **Numba Toolchain Versions** Numba Version : 0.55.1 llvmlite Version : 0.38.0 **LLVM Information** LLVM Version : 11.1.0 **CUDA Information** CUDA Device Initialized : False CUDA Driver Version : ? CUDA Runtime Version : ? CUDA NVIDIA Bindings Available : ? CUDA NVIDIA Bindings In Use : ? CUDA Detect Output: None CUDA Libraries Test Output: None **SVML Information** SVML State, config.USING\_SVML : False SVML Library Loaded : False llvmlite Using SVML Patched LLVM : True SVML Operational : False **Threading Layer Information** TBB Threading Layer Available : True ±-\>TBB imported successfully. OpenMP Threading Layer Available : True ±-\>Vendor: GNU Workqueue Threading Layer Available : True ±-\>Workqueue imported successfully. **Numba Environment Variable Information** NUMBA\_CUDA\_USE\_NVIDIA\_BINDINGS : 0 **Conda Information** Conda Build : 3.22.0 Conda Env : 23.1.0 Conda Platform : linux-64 Conda Python Version : 3.9.13.final.0 Conda Root Writable : True nvidia-smi Thu Mar 2 08:36:38 2023 ±----------------------------------------------------------------------------+ | NVIDIA-SMI 525.85.05 Driver Version: 525.85.05 CUDA Version: 12.0 | |-------------------------------±---------------------±---------------------+ | 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 … Off | 00000000:01:00.0 On | N/A | | 0% 37C P8 10W / 290W | 550MiB / 8192MiB | 0% Default | | | | N/A | ±------------------------------±---------------------±---------------------+ ±----------------------------------------------------------------------------+ | Processes: | | GPU GI CI PID Type Process name GPU Memory | | ID ID Usage | |=============================================================================| | 0 N/A N/A 2155 G /usr/lib/xorg/Xorg 369MiB | | 0 N/A N/A 2426 G …ome-remote-desktop-daemon 3MiB | | 0 N/A N/A 2463 G /usr/bin/gnome-shell 68MiB | | 0 N/A N/A 3647 G …894695049316347738,131072 96MiB | | 0 N/A N/A 4104 G …b/thunderbird/thunderbird 8MiB | ±----------------------------------------------------------------------------+ Following instructions from Nvidia → how-to-cuda-python i tried to install support for python conda install pyculib Collecting package metadata (current\_repodata.json): done Solving environment: failed with initial frozen solve. Retrying with flexible solve. Collecting package metadata (repodata.json): done Solving environment: failed with initial frozen solve. Retrying with flexible solve. PackagesNotFoundError: The following packages are not available from current channels: - pyculib Current channels: … So I am exactly where I was before. ---
**Author:** ![gmarkall](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/gmarkall/32/42_2.png) [@gmarkall](https://numba.discourse.group/u/gmarkall)\ **Post date:** [March 6, 2023, 10:38am UTC](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806/5 "2023-03-06T10:38:49Z")
> [@kristof-333](#): > > Following instructions from Nvidia → how-to-cuda-python > > i tried to install support for python > > conda install pyculib Where are these instructions from? pyculib has been deprecated for many years, so there shouldn’t be any mention of it in current instructions. Can you create and activate a new environment with: ```auto conda create -n numba-cuda python=3.9 numba=0.56.4 numpy=1.23 cudatoolkit=11.8 conda activate numba-cuda ``` Then run: ```auto python -c "from numba import cuda; cuda.cudadrv.libs.test()" ``` and ```auto python -c "from numba import cuda; cuda.detect()" ``` and paste the output of those commands here please? Also - how are you installing the NVIDIA drivers? ---
**Author:** ![kristof-333](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/kristof-333/32/789_2.png) [@kristof-333](https://numba.discourse.group/u/kristof-333)\ **Post date:** [March 6, 2023, 4:59pm UTC](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806/6 "2023-03-06T16:59:41Z")
> [@gmarkall](#): > > Where are these instructions from? pyculib has been deprecated for many years, so there shouldn’t be any mention of it in current instructions. The instructions are prominently displayed on Nvidia Cuda installation page. I will follow your recommendations but before I have another questions to clarify something what I perhaps do not understand: I installed Cuda and the only way I can run C programs using Cuda is with sudo. I develop Python using PyCharm Community and similarly I have to run as root to have them work correctly. ---
**Author:** ![kristof-333](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/kristof-333/32/789_2.png) [@kristof-333](https://numba.discourse.group/u/kristof-333)\ **Post date:** [March 6, 2023, 5:02pm UTC](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806/7 "2023-03-06T17:02:08Z")
> [@gmarkall](#): > > Also - how are you installing the NVIDIA drivers? Since I run on Ubuntu 22.04 I selected Software Updater-\> Settings → Additional drivers ![image](https://global.discourse-cdn.com/free1/uploads/numba/original/1X/75310ce0690473538aa4dfc8a9e5d43bdf30114a.png) Thank I installed Cuda environment from Nviadia. ---
**Author:** ![gmarkall](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/gmarkall/32/42_2.png) [@gmarkall](https://numba.discourse.group/u/gmarkall)\ **Post date:** [March 6, 2023, 5:55pm UTC](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806/8 "2023-03-06T17:55:45Z")
> [@kristof-333](#): > > The instructions are prominently displayed on Nvidia Cuda installation page. Can you please provide the URL? > [@kristof-333](#): > > I installed Cuda and the only way I can run C programs using Cuda is with sudo. > I develop Python using PyCharm Community and similarly I have to run as root to have them work correctly. I think the [CUDA Setup and Installation](https://forums.developer.nvidia.com/c/accelerated-computing/cuda/cuda-setup-and-installation/8) NVIDIA Developer Forum is the best place to ask to resolve this. I think the Numba issue may persist at least until this issue is resolved. ---
**Author:** ![kristof-333](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/kristof-333/32/789_2.png) [@kristof-333](https://numba.discourse.group/u/kristof-333)\ **Post date:** [March 6, 2023, 6:20pm UTC](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806/9 "2023-03-06T18:20:12Z")
Thank you for reply. The URL to nvidia page https:–developer-dot-nvidia-dot-com/how-to-cuda-python ![image](https://global.discourse-cdn.com/free1/uploads/numba/original/1X/e1c600dd98a4ca5409279779802cb6fbe427c3c1.png) I did not get traction any traction on Nvidia forum. I found some responses from the moderator that some files have to be created in the runtime under root. I wanted to ask. In your experience, you can successfully develop in python and run your Cuda enabled applications using Pycharm or VSC without need for root account? ---
**Author:** ![kristof-333](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/kristof-333/32/789_2.png) [@kristof-333](https://numba.discourse.group/u/kristof-333)\ **Post date:** [March 6, 2023, 7:20pm UTC](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806/10 "2023-03-06T19:20:15Z")
So this is the result I am getting conda create -n numba-cuda python=3.9 numba=0.56.4 numpy=1.23 cudatoolkit=11.8 Collecting package metadata (current\_repodata.json): done Solving environment: failed with repodata from current\_repodata.json, will retry with next repodata source. Collecting package metadata (repodata.json): done Solving environment: failed PackagesNotFoundError: The following packages are not available from current channels: - cudatoolkit=11.8 Keep in mind I currently am on Cuda installation Cuda 12. Does it mean I need to downgrade my CUDA package? I am sorry, I am new to dealing with all the toolkits and not necessarily understand all the connections between different parts. ---
**Author:** ![gmarkall](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/gmarkall/32/42_2.png) [@gmarkall](https://numba.discourse.group/u/gmarkall)\ **Post date:** [March 6, 2023, 9:38pm UTC](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806/11 "2023-03-06T21:38:40Z")
> The URL to nvidia page Thanks for providing the URL - I’ll try to get this looked into. The Numba documentation may be a little more helpful for getting set up: [Overview — Numba 0.59.0dev0+179.ga4664180.dirty documentation](https://numba.readthedocs.io/en/latest/cuda/overview.html#requirements) > So this is the result I am getting I see - I hadn’t realised that cudatoolkit 11.8 was not available in the Anaconda channels. If you do ```auto conda create -n numba-cuda python=3.9 numba=0.56.4 numpy=1.23 conda-forge::cudatoolkit=11.8 ``` then that should install the toolkit from conda-forge. > Keep in mind I currently am on Cuda installation Cuda 12. > Does it mean I need to downgrade my CUDA package? There are two components, the driver and the toolkit. Your driver is version 12 (as is the toolkit you had installed, I think), and it’s OK to use an older toolkit with a newer driver. The CUDA 12 toolkit is not yet available in conda-forge, but using 11.8 is fine with a 12.0 driver. > I am sorry, I am new to dealing with all the toolkits and not necessarily understand all the connections between different parts. No problem, hopefully we can work through and solve the problem! ---
**Author:** ![gmarkall](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/gmarkall/32/42_2.png) [@gmarkall](https://numba.discourse.group/u/gmarkall)\ **Post date:** [March 7, 2023, 9:15am UTC](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806/12 "2023-03-07T09:15:28Z")
Sorry, I missed a couple of your questions: > I did not get traction any traction on Nvidia forum. I found some responses from the moderator that some files have to be created in the runtime under root. I found your post on the forums, I think - is it [Permissions for cuda drivers on Ubuntu 22.04 - CUDA Setup and Installation - NVIDIA Developer Forums](https://forums.developer.nvidia.com/t/permissions-for-cuda-drivers-on-ubuntu-22-04/244722) ? Some other posts I found suggest it might be a permissions issue related to the groups you’re in, but I don’t have an exact idea of what might need to change if that’s the case. To figure out the next steps, can you provide the output of the `groups` command for your user please? For example on my system, I have: ```auto $ groups gmarkall adm cdrom sudo dip plugdev lpadmin sambashare docker ``` > I wanted to ask. In your experience, you can successfully develop in python and run your Cuda enabled applications using Pycharm or VSC without need for root account? Yes, it’s normal to use an account without root access - your situation, where root is required, is the first instance of this problem that I’ve been aware of. ---
**Author:** ![kristof-333](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/kristof-333/32/789_2.png) [@kristof-333](https://numba.discourse.group/u/kristof-333)\ **Post date:** [March 7, 2023, 2:22pm UTC](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806/13 "2023-03-07T14:22:55Z")
> [@gmarkall](#): > > I found your post on the forums, I think - is it [Permissions for cuda drivers on Ubuntu 22.04 - CUDA Setup and Installation - NVIDIA Developer Forums]? Yes this is the post I wrote. In Nvidia documentation for installation there is a statement which suggests if things do not work it maybe permissions issue. But my permissions on the drivers are correct. ---
**Author:** ![kristof-333](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/kristof-333/32/789_2.png) [@kristof-333](https://numba.discourse.group/u/kristof-333)\ **Post date:** [March 7, 2023, 2:27pm UTC](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806/14 "2023-03-07T14:27:40Z")
> [@gmarkall](#): > > For example on my system, I have: > > ```auto > $ groups > gmarkall adm cdrom sudo dip plugdev lpadmin sambashare docker > > ``` My groups are kristof adm cdrom sudo dip video plugdev lpadmin lxd sambashare I also added myself to sudoers with no effect on Cuda. I created & added myself to video group, thinking perhaps this will help. When dealing with some communication on Raspberry Pi and using SPI & I2C I had to be in a specific group. But like any other attempts it had no effect. ---
**Author:** ![kristof-333](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/kristof-333/32/789_2.png) [@kristof-333](https://numba.discourse.group/u/kristof-333)\ **Post date:** [March 7, 2023, 2:39pm UTC](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806/15 "2023-03-07T14:39:56Z")
> [@gmarkall](#): > > There are two components, the driver and the toolkit. Your driver is version 12 (as is the toolkit you had installed, I think), and it’s OK to use an older toolkit with a newer driver. The CUDA 12 toolkit is not yet available in conda-forge, but using 11.8 is fine with a 12.0 driver. Thank you for confirmation. So I was understanding it correctly. There is no secret magic there. ---
**Author:** ![kristof-333](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/kristof-333/32/789_2.png) [@kristof-333](https://numba.discourse.group/u/kristof-333)\ **Post date:** [March 7, 2023, 2:56pm UTC](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806/16 "2023-03-07T14:56:46Z")
> [@gmarkall](#): > > Yes, it’s normal to use an account without root access - your situation, where root is required, is the first instance of this problem that I’ve been aware of. My installation is a vanilla installation of Ubuntu. Everything I installed to-date was done following online documentation. I found a post about issues with Cuda and 22.04: (askubuntu-dot-com/questions/1427289/cuda-install-issues-22-04-lts) but I am getting very sceptical about some of the things I read. A lot of them seem to be sketchy and I do not see a consistent pattern to any solutions. ---
**Author:** ![kristof-333](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/kristof-333/32/789_2.png) [@kristof-333](https://numba.discourse.group/u/kristof-333)\ **Post date:** [March 7, 2023, 3:15pm UTC](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806/17 "2023-03-07T15:15:06Z")
> [@gmarkall](#): > > I see - I hadn’t realised that cudatoolkit 11.8 was not available in the Anaconda channels. If you do > > ```auto > conda create -n numba-cuda python=3.9 numba=0.56.4 numpy=1.23 conda-forge::cudatoolkit=11.8 > > ``` > > then that should install the toolkit from conda-forge. Thank you for instructions. Successfully executed. Running my simplistic test program from numba import cuda if **name** == ‘ **main** ’: print(cuda.gpus) in PyCharm with numba-conda environment activated still suffers from the same issue numba.cuda.cudadrv.error.CudaSupportError: Error at driver init: Call to cuInit results in CUDA\_ERROR\_NO\_DEVICE (100) I built an executable using pyinstaller and run it with sudo … no problems (numba-cuda) kristof@kristof-i9:~/Development/ILLUMISONICS/CUDA-examples/cuda-gpus/dist/main$ sudo ./main \ Running without sudo (numba-cuda) kristof@kristof-i9:~/Development/ILLUMISONICS/CUDA-examples/cuda-gpus/dist/main$ ./main Traceback (most recent call last): File “numba/cuda/cudadrv/driver.py”, line 247, in ensure\_initialized File “numba/cuda/cudadrv/driver.py”, line 320, in safe\_cuda\_api\_call File “numba/cuda/cudadrv/driver.py”, line 388, in \_check\_ctypes\_error numba.cuda.cudadrv.driver.CudaAPIError: [100] Call to cuInit results in CUDA\_ERROR\_NO\_DEVICE During handling of the above exception, another exception occurred: Traceback (most recent call last): File “main.py”, line 5, in File “numba/cuda/cudadrv/devices.py”, line 43, in **str** File “numba/cuda/cudadrv/devices.py”, line 26, in **getattr** File “numba/cuda/cudadrv/driver.py”, line 417, in get\_device\_count File “numba/cuda/cudadrv/driver.py”, line 285, in **getattr** File “numba/cuda/cudadrv/driver.py”, line 251, in ensure\_initialized numba.cuda.cudadrv.error.CudaSupportError: Error at driver init: Call to cuInit results in CUDA\_ERROR\_NO\_DEVICE (100) ---
**Author:** ![kristof-333](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/kristof-333/32/789_2.png) [@kristof-333](https://numba.discourse.group/u/kristof-333)\ **Post date:** [March 7, 2023, 3:20pm UTC](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806/18 "2023-03-07T15:20:15Z")
Here is my PyCharm setup ![image](https://global.discourse-cdn.com/free1/uploads/numba/original/1X/1c27ee317587327dffaad62fa238704a39c6a4ed.png) ---
**Author:** ![kristof-333](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/kristof-333/32/789_2.png) [@kristof-333](https://numba.discourse.group/u/kristof-333)\ **Post date:** [March 10, 2023, 11:54pm UTC](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806/19 "2023-03-10T23:54:12Z")
@gmarkall I got it to work I reinstalled Ubuntu 22.04.2 and activated older kernel 5.15.0-46-generic selected nvidia-driver-515 good enough for cuda 11.7 installed conda with cuda 11.7 now everything works thank you for helping me along … I probably would not arrived at the solution if not you ---
**Author:** ![gmarkall](https://yyz2.discourse-cdn.com/free1/user_avatar/numba.discourse.group/gmarkall/32/42_2.png) [@gmarkall](https://numba.discourse.group/u/gmarkall)\ **Post date:** [March 15, 2023, 7:22am UTC](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806/20 "2023-03-15T07:22:02Z")
Many thanks for the update! Glad you got it to work… I’m not too sure how I helped, I was a bit stumped - do you know what the difference between your current and previous setups is? [Next page](https://numba.discourse.group/t/ubuntu-22-04-fresh-install-with-conda-numba-does-not-see-gpu/1806.md?page=2)