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.

Better memory estimation for ROCm that support mem efficient attention.

comfyanonymous8773ccf74d18
diff --git a/comfy/model_management.py b/comfy/model_management.py
index f3d90c668..212ce9af2 100644
--- a/comfy/model_management.py
+++ b/comfy/model_management.py
@@ -909,6 +909,8 @@ def pytorch_attention_flash_attention():
return True
if is_ascend_npu():
return True
+ if is_amd():
+ return True #if you have pytorch attention enabled on AMD it probably supports at least mem efficient attention
return False
def mac_version():