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.

--windows-standalone-build now enables --cuda-malloc

comfyanonymous55d0fca9fa7f
diff --git a/comfy/cli_args.py b/comfy/cli_args.py
index 9a388b03b..e7ce256b7 100644
--- a/comfy/cli_args.py
+++ b/comfy/cli_args.py
@@ -85,3 +85,4 @@ args = parser.parse_args()
if args.windows_standalone_build:
args.auto_launch = True
+ args.cuda_malloc = True #work around memory issue in nvidia drivers > 531