Skip to content

[bug fix] fix multi-gpu model loading bug #1188

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
billcao2000 wants to merge 3 commits into from
Closed

[bug fix] fix multi-gpu model loading bug #1188

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
593b3cc
fix multi-gpu model loading bug
Jan 10, 2026
c6aab9e
optimize code structure
Jan 10, 2026
0bbf574
add npu support; update load_state_dict_from_safetensors
Jan 12, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 4 additions & 0 deletions diffsynth/core/loader/file.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,6 +15,10 @@ def load_state_dict(file_path, torch_dtype=None, device="cpu"):


def load_state_dict_from_safetensors(file_path, torch_dtype=None, device="cpu"):
import torch.distributed as dist
from diffsynth.core.device.npu_compatible_device import get_device_name, IS_NPU_AVAILABLE, IS_CUDA_AVAILABLE
if dist.is_available() and dist.is_initialized() and (IS_CUDA_AVAILABLE or IS_NPU_AVAILABLE):
device = get_device_name()
state_dict = {}
with safe_open(file_path, framework="pt", device=str(device)) as f:
for k in f.keys():
Expand Down
4 changes: 4 additions & 0 deletions diffsynth/core/vram/disk_map.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -28,6 +28,10 @@ def get_slice(self, name):
class DiskMap:

def __init__(self, path, device, torch_dtype=None, state_dict_converter=None, buffer_size=10**9):
import torch.distributed as dist
from diffsynth.core.device.npu_compatible_device import get_device_name, IS_NPU_AVAILABLE, IS_CUDA_AVAILABLE
if dist.is_available() and dist.is_initialized() and (IS_CUDA_AVAILABLE or IS_NPU_AVAILABLE):
device = get_device_name()
self.path = path if isinstance(path, list) else [path]
self.device = device
self.torch_dtype = torch_dtype
Expand Down