Comfy-Org/ComfyUI

Create caching option that organizes cache based on loading times of cached values

Open

#8367 opened on Jun 1, 2025

View on GitHub
 (0 comments) (0 reactions) (0 assignees)Python (112,952 stars) (13,218 forks)batch import
Featurehelp wanted

Description

Feature Idea

Create a new caching algorithm/class that

  • scores each cached output on the basis of (1) time to load cold and (2) relative size of cached output compared with available system memory
  • arranges cached values in min- or max-heap such that the item being popped from cache is always the highest scored item

The effect is that we are more likely to pop items from the cache that take a small amount of time to load relative to their size.

Then expose that option via command line option, to exist alongside current options (lru, classic, no caching)

Existing Solutions

No response

Other

No response

Contributor guide

Tech stack
python
Domain
backendperformance
Issue type
feature
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
3
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
over 1 week
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
fresh
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
clear
Prerequisites
Pythondata structures (heap)caching conceptscommand line interface basics
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
40
Research direction
Examine the existing caching implementations in ComfyUI (likely in `comfy/caching.py` or similar). Understand how the current LRU and 'classic' cache options are structured and integrated via command line arguments. Design a new class that scores cached outputs by cold load time and size relative to system memory, and uses a heap to evict the highest scored item. Implement the class and add it as a new caching option in the CLI. Test with a variety of workloads to ensure performance benefits. No maintainer guidance has been provided yet, so it's advisable to open a discussion or seek feedback on the design before implementation.