4 ms·

"TL;DR advice I work with data sets > 250GB: GTX Titan eBay I have no money: 3GB GTX 580 I do Kaggle: GTX 980 or GTX 580 I am a researcher: 1-4x GTX 980 I

by etiam 12y ago

"TL;DR advice I work with data sets > 250GB: GTX Titan eBay I have no money: 3GB GTX 580 I do Kaggle: GTX 980 or GTX 580 I am a researcher: 1-4x GTX 980 I am a researcher with data sets > 250GB: 1-4x GTX Titan eBay I never used deep learning before: 3GB GTX 580 eBay I started deep learning and I am serious about it: Start with one GTX 580 and buy more GTX 580s as you feel the need for them; buy new Volta GPUs in 2016 Q2/Q3 I want to build a GPU cluster: This is really complicated, I will write some advice about this soon, but you can get some ideas here"

ajtulloch 12y ago
I don't understand the claim that you should get a large RAM GPU if you have a large dataset. It's the model and your training procedure that dictates the amount of GPU ram consumed (e.g. fft convolution kernels like fbcunn use a decent amount of RAM compared to CuDNN's fused gemm, larger batch sizes mean more ram for storing activations, etc).
- nshepperd 12y ago
  With a larger dataset a larger model is more useful though, since you are less likely to suffer from overfitting.