Hardware requirements for deep learning, Tensorflow, Spark

I am upgrading my laptop. Considering Dell Precision or Alienware and need advice with GPU i.e Quadro (eg P600) and Geforce (1060, 1070, etc) options with budget <$3000.

This thread covers most pros and cons of Quadros vs GeForce for deep learning.

In short, Quadros are good but you’re spending a more money for not a lot more performance. You’re better off with a GeForce device if you’re trying to get the best value for your money. I have a 1070 on my laptop and a 1080Ti on my desktop and I couldn’t be happier. We’ve got a bunch of Titan Xp’s in our work machines but I rarely ever use it because the 1070 and 1080 are pretty capable devices for prototyping and training reasonably sized models. If you plan on training extremely large models on large datasets, I would recommend looking for a desktop alternative. Don’t forget that the price you’re going to pay for having a good GPU in your laptop is horrendous battery life and pretty high temperatures.

The Alienware is also an extremely bulky device. It’s slightly better cooled than some of it’s counterparts but it’s still a heavy piece of kit if you travel a lot.