torch.cuda.empty_cache() defaults to cuda:0 device unless explicitly set otherwise first. Updating torch_gc() to use the device set by --device-id if specified to avoid OOM edge cases on multi-GPU systems.

This commit is contained in:
Matthew McGoogan 2022-11-26 23:25:16 +00:00
parent b5050ad207
commit c67c40f983

View File

@ -44,8 +44,18 @@ def get_optimal_device():
def torch_gc():
if torch.cuda.is_available():
torch.cuda.empty_cache()
torch.cuda.ipc_collect()
from modules import shared
device_id = shared.cmd_opts.device_id
if device_id is not None:
cuda_device = f"cuda:{device_id}"
else:
cuda_device = "cuda"
with torch.cuda.device(cuda_device):
torch.cuda.empty_cache()
torch.cuda.ipc_collect()
def enable_tf32():