However the version 7 does not work well with cuda and cudnn. Install CUDA, cuDNN and tensorflow Prerequisitesīefore starting installation of cuda and cudnn, you have to change g++ version from 7 to 6. Then you can proceed to installation of cuda and cudnn, which are necessary for GPU enabled machine learning. If the GPU usage and other information are show, the installation is successful. Although it can also be installed via terminal, I used “ Software & Updates”, which is an Ubuntu’s standard application, since I believed it would be more secure.īefore that, we add an apt repository for newer nvidia graphics drivers: $ sudo add-apt-repository ppa:graphics-drivers $ sudo apt-get update $ sudo apt-get install nvidia-driver-410 $ sudo rebootĬheck if the installation was successful after rebooting: $ nvidia-smi Since any nvidia graphics driver hasn’t been installed, we install the latest one manually. After downloading the proper iso file of Ubuntu from here, launch Startup Disk Creator, then select the USB device on which you are going to install Ubuntu.Īfter creating the partitions correctly, it looks like this.Ĭlick “install now”, and then the installation process begins. I created the USB installer on another Ubuntu machine and its preinstalled software “ Startup Disk Creator”. Although details on how to make an Ubuntu USB installer are outside the scope of this post, I can briefly show you one way I used. The Ubuntu 18 is going to be installed using an USB installer prepared in advance. Windows 10 already installed (with fingerprint scanner enabled).Dell (2019 model) with 4K touch display.Thus I decided to install Ubuntu on this new laptop along with Windows 10. In addition to that, I believe installing a proper nvidia graphics driver and APIs needed to do machine leaning with GPU enabled, such as CUDA and cuDNN, might be quite a bit easier.
If you want to do machine leaning, using Ubuntu is recommended rather than Windows 10, since much more APIs and documents on the web are available. However, in my country, I could not purchase it with Ubuntu installed, and it ships only with Windows 10. Since it has a high-end CPU i7–8750H and a GPU nvidia 1050-ti, it should be capable of doing machine learning, especially deep leaning, in Python using the GPU. Recently I bought a new laptop xps-15 to do machine learning. Dell XPS 15 (9570) Dual-Booting Windows 10 and Ubuntu 18
0 kommentar(er)
