If you are thinking about buying one... or two... GPUs for your deep learning or GenAI computer, you must consider options like Ada Lovelace, 30-series, 40-series, 50-series Ampere, Blackwell, and GeForce. These brands are rapidly evolving. In this video, I discuss this ever-changing landscape as of 2025. There are many changes this year. NVLink is no more, unless you are dealing with the server class. The latest generation PCIe bus handles the cross GPU data transfer. 0:00 Assumptions 1:17 Geforce or RTX Pro? 1:41 RTX Pro Has More Memory 2:19 RTX Pro for Multiple GPUs 2:48 Cooling GPUs 3:43 GeForce RTX 4:42 MAX Performance 5:24 What GPU States Are Important 5:49 Buying Suggestions 7:09 What about NVLink? 7:43 Budget, Midline and Pro #gpu #cuda #lovelace #ada #blackwell #ampere #Python #Tensorflow #Keras #nvidia #3080 #3090 #4080 #4090 #5090 #pytorch
More information here: https://sites.wustl.edu/jeffheaton/t81-558/. These are the videos related to my "Applications of Deep Neural Networks" class. Update videos are added to these core videos as the class progresses. We are currently on the Spring 2018 semester..

Shorts