• Stars
    star
    270
  • Rank 151,287 (Top 3 %)
  • Language Cuda
  • Created over 2 years ago
  • Updated over 2 years ago

Reviews

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

Repository Details

Grab GPU whenever available

Grab the GPUs to run your own code!

Download (downward compatibility)

CUDA 10.1:

wget https://github.com/godweiyang/GrabGPU/releases/download/v1.0.0/gg_cu101

CUDA 11.0:

wget https://github.com/godweiyang/GrabGPU/releases/download/v1.0.0/gg_cu110

CUDA 11.2:

wget https://github.com/godweiyang/GrabGPU/releases/download/v1.0.0/gg_cu112

Compile the source code

nvcc gg.cu -o gg

Run default script

Usage:

./gg <GPU Memory (GB)> <Occupied Time (h)> <GPU ID>

Example:
Occupy 16 GB GPU memory for 24 hours using GPU 0, 1, 2, 3 to run default script.

./gg 16 24 0,1,2,3

Run your own script

Usage:

./gg <GPU Memory (GB)> <Occupied Time (h)> <GPU ID> <OPTIONAL: Script Path>

Example:
Occupy 16 GB GPU memory using GPU 0, 1, 2, 3 to run your own run.sh. Note that the occupied time here is useless.

./gg 16 24 0,1,2,3 run.sh