6

I've been running this notebook with the Runtime Type as "high-RAM" "GPU." I was getting the following error:

CUDA out of memory. Tried to allocate 64.00 MiB (GPU 0; 15.90 GiB total capacity; 14.81 GiB already allocated; 31.75 MiB free; 14.94 GiB reserved in total by PyTorch)

So I upgraded from Pro to Pro+, because that's supposed to give me more memory, but I'm still getting the same error.

ans
  • 378
  • 1
  • 5
  • 18
Emmy
  • 71
  • 1
  • 3

1 Answers1

4

I don't think a better GPU was promised with Colab Pro+.

enter image description here

Kevin K
  • 96
  • 1
  • 1
  • 8
  • 2
    Dang, you're certainly right. I didn't understand the difference between the RAM and the GPU's memory, I just thought increasing my RAM would fix my memory errors. Thank you. – Emmy Aug 19 '21 at 18:37
  • This puzzles me because their pitch for becoming a paid Colab Pro user includes "more memory" than free users. I get the same out of memory error whether I'm free or Pro. – cbmtrx Jun 12 '22 at 20:26
  • @cbmtrx are y on pro or pro plus? – Kevin K Jun 12 '22 at 20:29
  • @KevinK I went from free to Pro this am; haven't been able to run a single program since I paid. – cbmtrx Jun 12 '22 at 20:35
  • @cbmtrx that is all. Did u check ur colab settings upon initialization? – Kevin K Jun 12 '22 at 20:35
  • @KevinK I'm using a public Colab that I had been successfully using all of yesterday as a free. Suddenly today I can't use it without it failing. Doesn't seem like I would need to change how I was using it. – cbmtrx Jun 12 '22 at 20:37