Skip to content

The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world.

Revert "Disable cuda malloc by default."

comfyanonymousf1d6cef71c70
diff --git a/comfy/cli_args.py b/comfy/cli_args.py
index a19d98e21..2397de3d6 100644
--- a/comfy/cli_args.py
+++ b/comfy/cli_args.py
@@ -51,8 +51,8 @@ parser.add_argument("--auto-launch", action="store_true", help="Automatically la
parser.add_argument("--disable-auto-launch", action="store_true", help="Disable auto launching the browser.")
parser.add_argument("--cuda-device", type=int, default=None, metavar="DEVICE_ID", help="Set the id of the cuda device this instance will use.")
cm_group = parser.add_mutually_exclusive_group()
-cm_group.add_argument("--cuda-malloc", action="store_true", help="Enable cudaMallocAsync.")
-cm_group.add_argument("--disable-cuda-malloc", action="store_true", help="Disable cudaMallocAsync (The current default).")
+cm_group.add_argument("--cuda-malloc", action="store_true", help="Enable cudaMallocAsync (enabled by default for torch 2.0 and up).")
+cm_group.add_argument("--disable-cuda-malloc", action="store_true", help="Disable cudaMallocAsync.")
fp_group = parser.add_mutually_exclusive_group()
diff --git a/cuda_malloc.py b/cuda_malloc.py
index 04d42f917..eb2857c5f 100644
--- a/cuda_malloc.py
+++ b/cuda_malloc.py
@@ -2,7 +2,6 @@ import os
import importlib.util
from comfy.cli_args import args
import subprocess
-import logging
#Can't use pytorch to get the GPU names because the cuda malloc has to be set before the first import.
def get_gpu_names():
@@ -64,7 +63,7 @@ def cuda_malloc_supported():
return True
-if args.cuda_malloc:
+if not args.cuda_malloc:
try:
version = ""
torch_spec = importlib.util.find_spec("torch")
@@ -75,11 +74,8 @@ if args.cuda_malloc:
module = importlib.util.module_from_spec(spec)
spec.loader.exec_module(module)
version = module.__version__
- supported = False
if int(version[0]) >= 2: #enable by default for torch version 2.0 and up
- supported = cuda_malloc_supported()
- if not supported:
- logging.warning("WARNING: cuda malloc enabled but not supported.")
+ args.cuda_malloc = cuda_malloc_supported()
except:
pass