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.

Log message for cudnn disable on AMD. (#10418)

comfyanonymous2c2aa409b01f
diff --git a/comfy/model_management.py b/comfy/model_management.py
index 7467391cd..a2c318ec3 100644
--- a/comfy/model_management.py
+++ b/comfy/model_management.py
@@ -333,6 +333,7 @@ SUPPORT_FP8_OPS = args.supports_fp8_compute
try:
if is_amd():
torch.backends.cudnn.enabled = False # Seems to improve things a lot on AMD
+ logging.info("Set: torch.backends.cudnn.enabled = False for better AMD performance.")
try:
rocm_version = tuple(map(int, str(torch.version.hip).split(".")[:2]))
except: