Transcript:
Hello, guys, hope you are well today. I’m going to show you how to install CUDA and CEO DNA library properly on Windows. It will help us to start our own deep learning and machine learning projects properly on Windows. So let’s get it started. Please do subscribe on my channel and click on the notification icon to gate. It is video updates, So the first thing that you need to do is basically we need to go to. Google and then search for NVIDIA CUDA toolkit and what it will does. It will take us to the Nvidia website where we can download the full version. So you can see the updated. Cuda version is cuda 10.2 but we are not interested in cuda, 10.2 rather we’re interested in can put a 10.1 because you can see. If you go to tensorflow, you can see that the latest version, then that they support is good A 10.1 They don’t support the 10.2 yet, so it is also the same case for Pi Torch. If you go to Pi Tours, Dot o-r-g and click on the gate is started. You can see that the latest version that they support in different versions in likeliness Max Windows, Whatever they support 10.1 they don’t support the 10.2 version. That’s why we need to go to. We need to go with 10.1 rather than 10.2 so forth from here, we need to click on the legacy releases and here you can see CUDA to kick 10.1 update 2 and it is on August 2019 Then click Windows because we are working on Windows. Then then we will download that, so it it is a and it is around 2.5 gigabytes and we will download that say it will take some time to download the whole thing, so just keep patience, so it is downloading right now. The CUDA one. So in the meantime, we can go to CU Dnn and create our own account because it it requires an account to download the CU Tnan Library, So you can click on the CU DNA library and then you can join or log in. If you already have an account, you can just log in on that and you can start down. It can disturb the download, but if you don’t have an account, you need to actually create an account for that so and I will show you how to create an account so so if you enter your email and password and a pseudonym doesn’t get any account with that, it will give me a take you directly to the sign up option and from here, you need to enter your first name, last name, gender organization and your occupation, something like that, and you need to fill up those information and then then you are good to go, And by this way, you can actually create your own account and click. Submit it will take some time. Then you need to agree the terms and conditions of Invidia, and from here you need to be careful that we need to choose the 10.1 version, not the 10.2 version as I discussed earlier. So from here, you need to choose Windows 10 because we are working on understand, so if you click on that, it will basically download the zip file of the CU Dnn Library, So it is downloading the CUDA and Co DNN library. So another important thing is that you need to install visual is to 2019 before you studying the cuda installation. You can watch my video on how to install visual is 2019 on my Youtube channel. The link is in the description below, so you need to make sure that you install 2019 visuals 2019 before starting the CUDA installation. So from here now we click on the CUDA installation file and just click. OK, it will take some time to install the CUDA [Music] [Music] so you can see that the installation of Nvidia CUDA is basically finished and you can just close this window and now we need to install the co DNN library and for that what we need to do is basically, we already have downloaded the co. Dnn from Nvidia. We need to extract that CU Dng file, and you can use any kind of Jeep Jeep software like 7g for our wengie any kind of software and you just need to unzip that directory and here you can see that the beam file there are Bin Folder include folder and La Li B Library folder, so you need to copy and paste these things to the installation directory of Gouda in order to install the CU DNA library and for this you need to first. You need to locate where CUDA is installed and you can get it. I think it is on the program. File C program file NVIDIA GP engine NVIDIA GPU Computing Computing Toolkit the toolkit folder and then CUDA, then 10.1 You need to go to that folder and just go to that folder 10.1 because we install 10.1 version and here you can also see that there is being include an L. I believe leaf directory there. So what we need to do is basically. We need to copy paste these files into that installation folder, so copy bin file and then paste it. If you’d ask for any permission, just click continue one years and then the and then copy the include folder files like CU D in an h-file. Just paste it and then the library fines. You need to exactly face the five that is and like the way it is in the G files, so it is already there in a organized way, So whatever the folder is in the zip file of the CEO DMG file. You just need to copy the same thing on the original installation folder. Then I think then it is installed and now you also need to make sure that you set up the environment variables properly and you can go to the environment variables and check that whether Kuda CU Das underscore pair path, Pa, Th Path is properly indicating to the directory that where the installation has been done. So you need to make sure the directory is right, then click. OK, and so then, if everything is OK, you are good to go and you can taste that. Whether the KUDA has been installed properly or not, you can go to command prompt and give this common to check that whether the versions and that you install are properly installed or not. You can. You can also type Nvidia. SMI, NVIDIA SMI. And you can see that it is showing CUDA version 10.1 and it is also showing your GPU and everything. So so that’s how you actually installs cuda and co DNN library of Nvidia on windows. So see you in the next video. Please don’t forget to subscribe our Youtube channel like our Facebook page. And if you like this video, please gives a thumbs up. Hope this serve, thank you. [applause] [Music] you?