comfyanonymous/ComfyUI
1
0
Fork 0
mirror of https://github.com/comfyanonymous/ComfyUI.git synced 2026-03-11 08:10:06 +00:00
Code Issues Packages Projects Releases Wiki Activity

Widen OOM_EXCEPTION to AcceleratorError form (#12835)

Browse Source 
Pytorch only filters for OOMs in its own allocators however there are
paths that can OOM on allocators made outside the pytorch allocators.
These manifest as an AllocatorError as pytorch does not have universal
error translation to its OOM type on exception. Handle it. A log I have
for this also shows a double report of the error async, so call the
async discarder to cleanup and make these OOMs look like OOMs.
This commit is contained in:
rattus
2026-03-09 21:41:02 -07:00
committed by GitHub
parent a912809c25
commit 535c16ce6e
7 changed files with 27 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
execution.py
View File

@@ -612,7 +612,7 @@ async def execute(server, dynprompt, caches, current_item, extra_data, executed,
logging.error(traceback.format_exc())
tips = ""
if isinstance(ex, comfy.model_management.OOM_EXCEPTION):
if comfy.model_management.is_oom(ex):
tips = "This error means you ran out of memory on your GPU.\n\nTIPS: If the workflow worked before you might have accidentally set the batch_size to a large number."
logging.info("Memory summary: {}".format(comfy.model_management.debug_memory_summary()))
logging.error("Got an OOM, unloading all loaded models.")