wangbo/ComfyUI
1
0
Fork 0
mirror of https://github.com/comfyanonymous/ComfyUI.git synced 2026-02-14 23:42:35 +08:00
Code Issues Actions 14 Packages Projects Releases Wiki Activity

Add env TORCH_AMD_CUDNN_ENABLED

Browse Source
This commit is contained in:
Christopher Anderson 2025-11-05 18:35:00 +11:00
parent c4a6b389de
commit 58db8864a6
1 changed files with 10 additions and 30 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

40
comfy/model_management.py
View File

@ -26,6 +26,7 @@ import importlib
import platform import platform
import weakref import weakref
import gc import gc
import os
class VRAMState(Enum): class VRAMState(Enum):
DISABLED = 0 #No vram present: no need to move models to vram DISABLED = 0 #No vram present: no need to move models to vram
@ -338,8 +339,11 @@ try:
if is_amd(): if is_amd():
arch = torch.cuda.get_device_properties(get_torch_device()).gcnArchName arch = torch.cuda.get_device_properties(get_torch_device()).gcnArchName
if not (any((a in arch) for a in AMD_RDNA2_AND_OLDER_ARCH)): if not (any((a in arch) for a in AMD_RDNA2_AND_OLDER_ARCH)):
torch.backends.cudnn.enabled = False # Seems to improve things a lot on AMD torch.backends.cudnn.enabled = os.environ.get("TORCH_AMD_CUDNN_ENABLED", "0").strip().lower() not in {
logging.info("Set: torch.backends.cudnn.enabled = False for better AMD performance.") "0", "off", "false", "disable", "disabled", "no"}
if not torch.backends.cudnn.enabled:
logging.info(
"ComfyUI has set torch.backends.cudnn.enabled to False for better AMD performance. Set environment var TORCH_AMD_CUDDNN_ENABLED=1 to enable it again.")
try: try:
rocm_version = tuple(map(int, str(torch.version.hip).split(".")[:2])) rocm_version = tuple(map(int, str(torch.version.hip).split(".")[:2]))
@ -1082,20 +1086,8 @@ def cast_to_device(tensor, device, dtype, copy=False):
non_blocking = device_supports_non_blocking(device) non_blocking = device_supports_non_blocking(device)
return cast_to(tensor, dtype=dtype, device=device, non_blocking=non_blocking, copy=copy) return cast_to(tensor, dtype=dtype, device=device, non_blocking=non_blocking, copy=copy)
PINNED_MEMORY = {}
TOTAL_PINNED_MEMORY = 0
if PerformanceFeature.PinnedMem in args.fast:
if WINDOWS:
MAX_PINNED_MEMORY = get_total_memory(torch.device("cpu")) * 0.45 # Windows limit is apparently 50%
else:
MAX_PINNED_MEMORY = get_total_memory(torch.device("cpu")) * 0.95
else:
MAX_PINNED_MEMORY = -1
def pin_memory(tensor): def pin_memory(tensor):
global TOTAL_PINNED_MEMORY if PerformanceFeature.PinnedMem not in args.fast:
if MAX_PINNED_MEMORY <= 0:
return False return False
if not is_nvidia(): if not is_nvidia():
@ -1104,21 +1096,13 @@ def pin_memory(tensor):
if not is_device_cpu(tensor.device): if not is_device_cpu(tensor.device):
return False return False
size = tensor.numel() * tensor.element_size() if torch.cuda.cudart().cudaHostRegister(tensor.data_ptr(), tensor.numel() * tensor.element_size(), 1) == 0:
if (TOTAL_PINNED_MEMORY + size) > MAX_PINNED_MEMORY:
return False
ptr = tensor.data_ptr()
if torch.cuda.cudart().cudaHostRegister(ptr, size, 1) == 0:
PINNED_MEMORY[ptr] = size
TOTAL_PINNED_MEMORY += size
return True return True
return False return False
def unpin_memory(tensor): def unpin_memory(tensor):
global TOTAL_PINNED_MEMORY if PerformanceFeature.PinnedMem not in args.fast:
if MAX_PINNED_MEMORY <= 0:
return False return False
if not is_nvidia(): if not is_nvidia():
@ -1127,11 +1111,7 @@ def unpin_memory(tensor):
if not is_device_cpu(tensor.device): if not is_device_cpu(tensor.device):
return False return False
ptr = tensor.data_ptr() if torch.cuda.cudart().cudaHostUnregister(tensor.data_ptr()) == 0:
if torch.cuda.cudart().cudaHostUnregister(ptr) == 0:
TOTAL_PINNED_MEMORY -= PINNED_MEMORY.pop(ptr)
if len(PINNED_MEMORY) == 0:
TOTAL_PINNED_MEMORY = 0
return True return True
return False return False