Commits on Aug 24, 2024

1. fix: unload from RAM as intended; force empty cache …

   Prior to this commit the following erroneous behavior was in play:
   
   - `unload_all_models_vram()` did not effectively remove all models from VRAM
   - `unload_all_models_ram()` did not remove any models from RAM
   - With AMD specifically, though its possible more generally as well, the torch cache was not cleared as this required an argument (`True`) sent to `_comfy_soft_empty_cache(...)`
   
   I have also added some additional logging to help further identify what is and is not helpful.

   

   tazlin committed Aug 24, 2024
   Configuration menu

   • View commit details
   • Copy full SHA for a6ec17d
   • Browse repository at this point

   Copy the full SHA

   a6ec17d View commit details

   Browse the repository at this point in the history