Spaces:
Running
on
Zero
Running
on
Zero
Update device_manager.py
Browse files- device_manager.py +33 -12
device_manager.py
CHANGED
@@ -25,25 +25,46 @@ class DeviceManager:
|
|
25 |
|
26 |
def check_zero_gpu_availability(self):
|
27 |
try:
|
28 |
-
|
29 |
-
|
30 |
-
|
31 |
-
|
32 |
-
|
|
|
|
|
|
|
|
|
|
|
33 |
except Exception as e:
|
34 |
logger.warning(f"Error checking ZeroGPU availability: {e}")
|
35 |
-
|
36 |
|
37 |
def get_optimal_device(self):
|
38 |
if self._current_device is None:
|
39 |
if self.check_zero_gpu_availability():
|
40 |
try:
|
41 |
-
|
42 |
-
|
43 |
-
|
|
|
|
|
|
|
|
|
|
|
44 |
self._current_device = torch.device('cpu')
|
45 |
-
logger.info("
|
46 |
else:
|
47 |
self._current_device = torch.device('cpu')
|
48 |
-
logger.info("Using CPU")
|
49 |
-
return self._current_device
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
25 |
|
26 |
def check_zero_gpu_availability(self):
|
27 |
try:
|
28 |
+
# 檢查 Hugging Face Space 環境變數
|
29 |
+
if not os.environ.get('SPACE_ID'):
|
30 |
+
return False
|
31 |
+
|
32 |
+
# 檢查是否在 Spaces 環境中並且啟用了 ZeroGPU
|
33 |
+
if os.environ.get('ZERO_GPU_AVAILABLE') == '1':
|
34 |
+
return True
|
35 |
+
|
36 |
+
return False
|
37 |
+
|
38 |
except Exception as e:
|
39 |
logger.warning(f"Error checking ZeroGPU availability: {e}")
|
40 |
+
return False
|
41 |
|
42 |
def get_optimal_device(self):
|
43 |
if self._current_device is None:
|
44 |
if self.check_zero_gpu_availability():
|
45 |
try:
|
46 |
+
# 確保 CUDA 可用
|
47 |
+
if torch.cuda.is_available():
|
48 |
+
self._current_device = torch.device('cuda')
|
49 |
+
logger.info("Using ZeroGPU")
|
50 |
+
else:
|
51 |
+
raise RuntimeError("CUDA not available")
|
52 |
+
except Exception as e:
|
53 |
+
logger.warning(f"Failed to initialize ZeroGPU: {e}")
|
54 |
self._current_device = torch.device('cpu')
|
55 |
+
logger.info("Fallback to CPU due to GPU initialization failure")
|
56 |
else:
|
57 |
self._current_device = torch.device('cpu')
|
58 |
+
logger.info("Using CPU (ZeroGPU not available)")
|
59 |
+
return self._current_device
|
60 |
+
|
61 |
+
def move_to_device(self, tensor_or_model):
|
62 |
+
device = self.get_optimal_device()
|
63 |
+
try:
|
64 |
+
if hasattr(tensor_or_model, 'to'):
|
65 |
+
return tensor_or_model.to(device)
|
66 |
+
except Exception:
|
67 |
+
self._current_device = torch.device('cpu')
|
68 |
+
if hasattr(tensor_or_model, 'to'):
|
69 |
+
return tensor_or_model.to('cpu')
|
70 |
+
return tensor_or_model
|