Posts with tag nvidia

on ubuntu

2024-06-28
installnvidiasoftwares-and-tools

on ubuntuRef: https://blog.csdn.net/qq_30468723/article/details/107531062Ref: https://blog.csdn.net/qq_42887760/article/details/126903100Ref: https://blog.csdn.net/xiaosongshine/article/details/115720887Arch kde ref: https://www.skyone.host/2024/archlinux-plasma-faq清华源装 555.run(不含 cuda,不是 developer 搜索的是 google 搜 nvidia drivers download)装 kde。原因是,为了装驱动需进入文本界面关闭 gui 服务,但 gdm 不能直接关不然文本界面都没了,至少得装个 lightdm,那不如直接先装 sddm 的 kde 吧Kde 装好了 reboot进文本界面 systemctl status sddm 正常,gdm 没开sudo /etc/init.d/sddm stop 直接进入了一个底层的终端,没法动,重启那就 sudo telinit 3sudo service sddm stop ok!!./NVIDIA.run 说找不到 cc ,重启sudo apt install gcc g++ make 还真没装sudo ./NVIDIA.runok 现在它在加了两个 blacklist, /usr/lib 下和 /etc 重启下要删掉重启需要进 tty,如果你卡住了就 ctrl + alt + F2 之类的进 tty,运行 /.run这次要求关 Secure mode,那就重启并关闭并重启在 ./run 说找不到 pkg-config,但是能继续安装重启后进入桌面,但是桌面缩放错误,什么东西都特别大,完全操作不了~试试 https://www.skyone.host/2024/archlinux-plasma-faq 中的 nvidia-drm,重启缩放还是一样答辩sudo dpkg-reconfigure gdm3 ,虽然说 gdm3 服务 inactive 但是还是切换成功重启发现桌面正常,甚至有 nvidia-smi,那好吧直接在 gdm 下开发看

nvidia-driver.1

2024-01-15
nvidiasoftwares-and-tools

test on: 16.04Linux ubuntu16-1080Ti 4.15.0-112-generic #113~16.04.1-Ubuntu SMP Fri Jul 10 04:37:08 UTC 2020 x86_64 x86_64 x86_64 GNU/Linuxnvidia-smi and nvidia-detector看起来天生自带的。Now, nvidia-detector cmd shows none.when there's no nvidia-smi我 apt install nvidia-384 然后报错了,直接搞没啦!junjie@ubuntu16-1080Ti:~$ nvidia-smi Failed to initialize NVML: Driver/library version mismatchWell you should reboot! as in https://stackoverflow.com/questions/43022843/nvidia-nvml-driver-library-version-mismatch如果不行,执行下面的东西,你看得懂吧?https://askubuntu.com/questions/902636/nvidia-smi-command-not-found-ubuntu-16-04sudo apt purge nvidia-* sudo add-apt-repository ppa:graphics-drivers/ppa sudo apt update sudo apt install nvidia-38

old-nvidia-smi-or-called-driver

2024-01-15
nvidiasoftwares-and-tools

来自 yuanzhi 老学长的建议直接搜 cuda-toolkit 装。虽然 nvidia 官方貌似只提供 16.04 的 toolkit.run,然而似乎在 18.04 上也可以直接 run.那么这个时候你要搜索老版本 ubuntu 上关闭图形界面以后装 nvidia 驱动的方式。为什么要禁用图形界面 + reboot 呢,老夫也不知道。而且 reboot 以后是一个分辨率很低的可视化界面,这时候你按 F1 进入命令行界面。run 了以后是一个 TUI 打钩界面,其中甚至包含 nvidia-smi。一般有 smi 的情况下 smi 不打钩,但是如果没有 smi 就打钩。23.9.19 以上方法(用 cuda.run 装 nvidia-smi)报错了,要先装 nvidia-smi 再装 cuda.纯 ssh 尝试这次是要把 nvidia-driver 384 换成 nvidia-driver 470. 因为 384 不支持 cuda 9.2.先 sudo apt purge nvidia-*. 它删除了 1G 多的东西,其中看起来还有 cuda-toolkit 相关,只不过后缀有 384 所以我肯定就不要了。先下载一个驱动.run:搜索: https://www.nvidia.com/en-us/geforce/drivers/搜索结果: https://www.nvidia.com/en-us/geforce/drivers/results/205995/然后跑这个 run。跑得时候因为在 16.04 上,它会直接告诉你: You seem to be using an X server, stop it.于是我就按照 https://askubuntu.com/questions/149206/how-to-install-nvidia-run : 990 top | grep Xorg 991 kill -9 Xorg 992 kill -9 1114 993 sudo kill -9 1114 994 ls 995 ./470.run 996 sudo ./470.run 997 ps -ef | grep Xorg 998 ps -ef | grep X 999 sudo service lightdm stop 1000 sudo lightdm stop 1001 ls 1002 sudo killall Xorg然后可以跑了,跑的时候他说要不要 kernal xxx,我本来想退出的选了 No,然而它继续往下跑了,他说要不要装兼容 32 位的我说不要。接下来sudo reboot其他啥都没干居然好了

record-nvidia-smi-fix-230918

2024-01-15
nvidiasoftwares-and-tools

9949 sudo ln -sfT /etc/alternatives/cuda/cuda-11.1 /usr/local/cuda 9950 ls 9951 ls -l 9952 nv 9953 nvcc -V 9954 tmux a -t auto_train 9955 cd GarmentImitation/manifests/workflows 9956 ls 9957 vim train.yaml 9958 cat train.yaml 9959 tmux a -t auto_train 9960 cd GarmentImitation/ 9961 make manipulation.train_tshirt_short START_EPISODE_SHORT=$(cat $HOME/.unifolding/config/EPISODE_IDX) MODEL_CKPT_PATH_SHORT=$(cat $HOME/.unifolding/config/WORKING_DIR ) 9962 python 9963 nvcc -V 9964 ls /usr/local 9965 j local 9966 ls 9967 ls -l 9968 ls 9969 ls -l 9970 nvcc -V 9971 ls 9972 vim ~/.zshrc 9973 ls 9974 history 9975 sudo adduser junjie 9976 sudo visudo 9977 ls 9978 cd /usr/local 9979 ls 9980 ll 9981 ls 9982 lsls 9983 ls 9984 rm cuda 9985 sudo rm cuda 9986 sudo rm cuda-11 9987 ln -s /usr/local/cuda-11.3 /usr/local/cuda 9988 sudo ln -s /usr/local/cuda-11.3 /usr/local/cuda 9989 ll 9990 nvcc -V 9991 vim ~/.zshrc 9992 pip uninstall open3d 9993 pip install open3d 9994 nvcc -V 9995 vim ~/.zshrc 9996 source ~/.zshrc 9997 nvcc -V 9998 conda activate speedfolding 9999 python 10000 cat /var/log/dpkg.log | grep nvidia 10001 history | grep apt 10002 conda activate speedfolding 10003 python 10004 nvidia-smi 10005 sudo apt-get --purge remove cuda-11-1 10006 nvidia-smi 10007 sudo apt-get autoremove 10008 nvcc -V 10009 sudo apt-get autoclean 10010 cd /usr/local 10011 ll 10012 sudo rm -rf cuda-11.1 10013 nvidia-smi 10014 sudo apt uninstall cuda 10015 cd Downloads/ 10016 cd ~/Downloads 10017 ls 10018 sudo apt-get remove nvidia-* 10019 sudo apt-get install nvidia-driver-535 10020 nvidia-smi 10021 cd ~/Downloads 10022 ls 10023 ubuntu-drivers devices 10024 nvidia-smi 10025 sudo apt remove nvidia-* 10026 sudo apt remove nvidia* 10027 sudo apt-get remove nvidia* 10028 nvidia-smi 10029 nvcc -V 10030 conda activate speedfolding 10031 python 10032 pycharm.sh 10033 tmux a -t auto_train 10034 tmux new -s auto_train 10035 cd GarmentImitation/manifests/workflows 10036 ls 10037 conda activate speedfolding 10038 make daemon.workflow_keeper 10039 ls 10040 nvidia-smi 10041 conda activate speedfolding 10042 cd GarmentImitation/ 10043 conda activate speedfolding 10044 git pull 10045 git stash Makefile 10046 git diff Makefile 10047 git diff 10048 git stash -h 10049 git stash push -m makefile Makefile 10050 git pull 10051 git stash list 10052 git stash pop 10053 vim Makefile 10054 git stash pop 10055 vim Makefile 10056 git add Makefile 10057 git stash pop 10058 git stash drop 0 10059 git stash list 10060 vim Makefile 10061 git add Ma 10062 git add Makefile 10063 make daemon.workflow_keeper 10064 tmux new -s auto_train 10065 tmux a 10066 cd .. 10067 make daemon.workflow_keeper 10068 cd GarmentImitation 10069 l 10070 sls 10071 ls 10072 cd manifests/workflows 10073 ls 10074 vim train.yaml 10075 curl http://127.0.0.1:8082/v1/scheduled_workflows 10076 vim train.yaml 10077 curl http://127.0.0.1:8082/v1/scheduled_workflows 10078 pip install workfow-keeper==0.0.3 10079 pip install -i https://pypy.org/simple workfow-keeper==0.0.3 10080 pip install -i https://pypy.org/simple workflow-keeper==0.0.3 10081 pip install -i https://pypy.org/simple workflow-keeper 10082 pip install workflow-keeper 10083 pip install packaging==21.3 10084 pip uninstall black 10085 python -m workflow_keeper serve --api_port=8083 10086 python -m workflow_keeper serve --api.port=8083 10087 curl http://127.0.0.1:8082/v1/scheduled_workflows 10088 curl http://127.0.0.1:8082/v1/scheduled_workflows | jq 10089 apt-get install jq 10090 sudo apt-get install jq 10091 curl http://127.0.0.1:8082/v1/scheduled_workflows | jq 10092 make robot.console.right 10093 make robot.console.l 10094 make robot.console.left 10095 make manipulation.run_tshirt_short 10096 ls 10097 cd GarmentImitation 10098 ls 10099 cd log 10100 ls 10101 cd experiment_real 10102 ls 10103 chmod -R 777 tshirt_short_action14_real_corl_v2 10104 conda activate speedfolding 10105 chmod -R 777 tshirt_short_action14_real_corl_v2 10106 ls 10107 cd GarmentImitation 10108 ls 10109 cd log 10110 ls 10111 cd log experiment_real 10112 cd experiment_real 10113 chmod -R 777 tshirt_short_action14_real_corl_v2 10114 sudo chmod -R 777 tshirt_short_action14_real_corl_v2 10115 vim ~/.unifolding/config/WORKING_DIR 10116 make manipulation.test_tshirt_short 10117 conda activate speedfolding 10118 make robot.console.left 10119 sudo chmod -R 777 tshirt_short_action14_real_corl_v2 10120 make robot.console.left 10121 make manipulation.run_tshirt_short 10122 umask 002 10123 make manipulation.test_tshirt_short 10124 make robot.console.left 10125 vim ~/.zshrc 10126 cd GarmentImitation/manifests/workflows 10127 vim train.yaml 10128 vim data_collection_virtual.yaml 10129 make handey

No more posts to load.