Transformers documentation
Accelerator selection
Accelerator selection
You can control which accelerators (CUDA, XPU, MPS, HPU, etc.) PyTorch sees and in what order during distributed training. Prioritize faster devices or limit training to a subset of available hardware. It works with both DistributedDataParallel and DataParallel, and doesn’t require Accelerate or the DeepSpeed integration.
Order of accelerators
Use the hardware-specific environment variable to select accelerators and set their order. Set it on the command line per run, or add it to ~/.bashrc or another startup config file.
Avoid exporting environment variables because if you forget how an environment variable was set up, you may silently train on the wrong accelerators. Set the environment variable on the same command line as the training run.
For example, to select accelerators 0 and 2 out of four:
CUDA_VISIBLE_DEVICES=0,2 torchrun trainer-program.py ...PyTorch sees only GPUs 0 and 2, which are mapped to cuda:0 and cuda:1. To reverse the order (use GPU 2 as cuda:0 and GPU 0 as cuda:1):
CUDA_VISIBLE_DEVICES=2,0 torchrun trainer-program.py ...To run without any GPUs:
CUDA_VISIBLE_DEVICES= python trainer-program.py ...Control the order of CUDA devices with CUDA_DEVICE_ORDER.
Order by PCIe bus ID (matches
nvidia-smi):export CUDA_DEVICE_ORDER=PCI_BUS_IDOrder by compute capability (fastest first):
export CUDA_DEVICE_ORDER=FASTEST_FIRST