Skip to content

I have 2GPUs, each 24GB, yet I get "Cuda out of memory" Error #27

Description

@mmdrahmani

Hi again

I have 2GPUs, each 24GB. I am using lit-llama-7B model. I was trying to set using two GPUs, by setting the number of devices here, fabric = L.Fabric(accelerator=accelerator, devices=2), but the model uses only one GPU (cuda:0), and the other one is not used at all. However, I get this error:
CUDA out of memory. Tried to allocate 192.00 MiB (GPU 0; 23.68 GiB total capacity; 23.03 GiB already allocated; 142.56 MiB free; 23.03 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
Am I doing something wrong? How can I make the model use both GPUs?

Thank you so much for your support.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions