Save 4 hours! We auto-setup your workflow! Free!

Drop your workflow.json — we handle every dependency, custom node, and model. Just open the link and run.

Auto-Setup Workflow Json (Free) Now!
ComfyUI > Nodes > comfyui-minimax-h3-audio-T8 > MiniMax H3 Qwen Reference Prefix Cache / 参考前缀缓存 (Advanced)

ComfyUI Node: MiniMax H3 Qwen Reference Prefix Cache / 参考前缀缓存 (Advanced)

Class Name

MiniMaxH3QwenReferencePrefixCacheT8Advanced

Category
T8/MiniMax H3/Conditioning/Experimental
Author
T8mars (Account age: 1708days)
Extension
comfyui-minimax-h3-audio-T8
Latest Updated
2026-08-20
Github Stars
0.75K

How to Install comfyui-minimax-h3-audio-T8

Install this extension via the ComfyUI Manager by searching for comfyui-minimax-h3-audio-T8
  • 1. Click the Manager button in the main menu
  • 2. Select Custom Nodes Manager button
  • 3. Enter comfyui-minimax-h3-audio-T8 in the search bar
After installation, click the Restart button to restart ComfyUI. Then, manually refresh your browser to clear the cache and access the updated list of nodes.

Visit ComfyUI Online for ready-to-use ComfyUI environment

  • Free trial available
  • 16GB VRAM to 80GB VRAM GPU machines
  • 400+ preloaded models/nodes
  • Freedom to upload custom models/nodes
  • 200+ ready-to-run workflows
  • 100% private workspace with up to 200GB storage
  • Dedicated Support

Run ComfyUI Online

MiniMax H3 Qwen Reference Prefix Cache / 参考前缀缓存 (Advanced) Description

Enhances H3 Qwen3-VL encoder efficiency with visual-reference prefix CPU-memory cache for AI artists.

MiniMax H3 Qwen Reference Prefix Cache / 参考前缀缓存 (Advanced):

The MiniMaxH3QwenReferencePrefixCacheT8Advanced node is designed to enhance the efficiency of the H3 Qwen3-VL encoder by implementing a bounded CPU-memory cache specifically for the visual-reference prefix. This node allows you to opt-in for caching, which can significantly reduce the computational load by storing and reusing the exact visual-reference prefix, while still recomputing the prompt text to ensure accuracy. It is particularly beneficial for scenarios where repeated visual references are used, as it optimizes performance without altering the core files of ComfyUI or the input CLIP object. This advanced caching mechanism is crucial for maintaining high efficiency in processing visual data, making it an essential tool for AI artists working with complex visual inputs.

MiniMax H3 Qwen Reference Prefix Cache / 参考前缀缓存 (Advanced) Input Parameters:

clip

The clip parameter is the input CLIP object that the node processes. It serves as the primary data source for the caching operation, allowing the node to extract and cache the visual-reference prefix. This parameter is crucial as it directly influences the node's ability to perform its caching function effectively.

mode

The mode parameter determines the operational mode of the cache. It offers two options: report_only and memory_lru_exp. The default is report_only, which provides a report on cache usage without altering the cache state. The memory_lru_exp mode enables a least-recently-used (LRU) expiration policy, which helps manage memory usage by removing the least accessed entries. This parameter is essential for controlling how the cache behaves and manages its entries.

max_entries

The max_entries parameter specifies the maximum number of entries that the cache can hold. It ranges from 1 to 16, with a default value of 1. This parameter is critical for defining the cache's capacity, directly impacting its ability to store and retrieve visual-reference prefixes efficiently.

maximum_cache_mib

The maximum_cache_mib parameter sets the maximum size of the cache in mebibytes (MiB). It can be adjusted between 64.0 and 65536.0 MiB, with a default of 1024.0 MiB. This parameter is vital for managing the memory footprint of the cache, ensuring it operates within the available system resources.

cache_epoch

The cache_epoch parameter is used to create a fresh, empty cache without affecting the input CLIP. It is an integer value ranging from 0 to 2147483647, with a default of 0. Incrementing this value resets the cache, which can be useful for starting a new caching session or clearing outdated entries.

MiniMax H3 Qwen Reference Prefix Cache / 参考前缀缓存 (Advanced) Output Parameters:

output

The output parameter provides the processed data from the node, which includes the cached visual-reference prefix. This output is essential for subsequent processing steps, as it contains the optimized data ready for further use.

cache

The cache parameter outputs the current state of the cache, including its contents and configuration. This information is crucial for understanding the cache's performance and making informed decisions about its management.

report

The report parameter delivers a JSON-formatted report detailing the cache's operation, including hit/miss statistics and memory usage. This output is valuable for monitoring the cache's effectiveness and identifying areas for optimization.

MiniMax H3 Qwen Reference Prefix Cache / 参考前缀缓存 (Advanced) Usage Tips:

  • To optimize performance, use the memory_lru_exp mode when working with large datasets or when memory resources are limited, as it helps manage cache size effectively.
  • Regularly monitor the report output to understand cache usage patterns and adjust max_entries and maximum_cache_mib parameters accordingly for optimal performance.

MiniMax H3 Qwen Reference Prefix Cache / 参考前缀缓存 (Advanced) Common Errors and Solutions:

unsupported Qwen prefix cache mode

  • Explanation: This error occurs when an invalid mode is specified for the cache operation.
  • Solution: Ensure that the mode parameter is set to either report_only or memory_lru_exp.

cache_epoch must be between 0 and 2147483647

  • Explanation: This error indicates that the cache_epoch value is outside the acceptable range.
  • Solution: Adjust the cache_epoch parameter to a value within the specified range of 0 to 2147483647.

MiniMax H3 Qwen Reference Prefix Cache / 参考前缀缓存 (Advanced) Related Nodes

Go back to the extension to check out more related nodes.
comfyui-minimax-h3-audio-T8
RunComfy
Copyright 2025 RunComfy. All Rights Reserved.

RunComfy is the premier ComfyUI platform, offering ComfyUI online environment and services, along with ComfyUI workflows featuring stunning visuals. RunComfy also provides AI Models, enabling artists to harness the latest AI tools to create incredible art.

MiniMax H3 Qwen Reference Prefix Cache / 参考前缀缓存 (Advanced)