Install NVIDIA GPU-Accelerated Deep Learning Libraries on your Home Computer (CUDA / CuDNN) (Eps7)

[johnbarca]

Ha ha ha. I wish you had posted this last year. I did all of this last year, Ubuntu 18.04, CUDA 10, with a 2080 Titan sitting inside a XG Station Pro eGPU case, using thunderbolt connectors to an ASUS laptop. I also Fubar'd my box, big time. The fix that made it all run perfectly was to use LightDM instead of GDM. Having said that, this was 6 months ago, and it is now flawless, and the eGPU gives big advantages over an internal GPU.

[barmalini]

I have AMD Radeon graphics card in my system, and have been using PlaidML + Keras with some success for more than a year, so I would appreciate the truth spoken about it, instead of un-grounded "NO AMD and NO INTEL" statements. This is especially important, because those less popular frameworks need more attention, so they can attract more people and develop faster. Thank you.

[debayandas]

Thank you for this video. Really helpful. Had messed up my poor system so many times without even knowing what I did wrong. Currently using system76-pop OS which circumvents all of the issues with a single line terminal install, but this video will be a future reference for me.

[slouma]

Hey thanks for the quality video made me subscribe! I have one question tho, I've read that tensorflow 2 implements keras in it, so no need to pip install keras, or no need to implement it using import tf.keras, am I missing something here ?

[Diamond_Hanz]

How do u know that it's running correctly?

[minnesotanature]

You can now run Anaconda on Windows Linux Subsystem
Setting up Anaconda on Windows 10 linux subsystem
1. Docs are here, easy step by step :
2. Download from Windows Store your linux version - I downloaded Ubuntu 18.04:
Open a Linux bash window and run the following commands
# bash ** install
4. Re-open a Linux bash window so PATHS to conda work type :
# export Display=:0
5. Create your first environment with conda example:
# conda create -f --name mltest python=3.6 numpy tensorflow scipy matplotlib pandas scipy seaborn
# conda activate mltest
6. Now you can follow along on installing CUDA