nodes/Windecay-ComfyUI_Dynamic-RAMCache
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

feat: 添加RAM缓存极限清理节点

Browse Source 
- 新增RAMCacheExtremeCleanup节点，提供一次性强清理功能
- 清理包含虚拟内存占用，并自动恢复之前的模式与阈值
This commit is contained in:
windecay
2026-02-04 23:44:49 +08:00
parent f0a516f748
commit fb05365923
4 changed files with 105 additions and 25 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+15 -3
View File
@@ -9,14 +9,15 @@ A custom node for ComfyUI that dynamically manages RAM usage by intelligently co
- **Dual Cache Management Modes**:
- CLASSIC (No Eviction): Standard cache behavior with no automatic purging
- RAM_PRESSURE (Auto Purge): Smart cache purging when system memory falls below the threshold
- **RAM Cache Extreme Cleanup (🧹)**: Performs a one-time aggressive purge and automatically restores the previous mode and threshold.
- **Configurable Memory Threshold**: Fine-tune the minimum amount of free RAM to maintain
- **Seamless Cache Migration**: Preserves essential cache data when switching between modes
- **Real-time Memory Monitoring**: Continuously checks system RAM availability
- **Seamless Cache Migration**: Preserves aerospace cache data when switching between modes
- **Real-time Memory Monitoring**: Continuously checks system RAM and virtual memory availability
## Installation
1. Clone or download this repository
2. Place the `ComfyUI_Dynamic_RAMCache` folder in your ComfyUI `custom_nodes` directory
2. Place the `ComfyUI_Dynamic-RAMCache` folder in your ComfyUI `custom_nodes` directory
3. Restart ComfyUI
## Usage
@@ -35,7 +36,18 @@ A custom node for ComfyUI that dynamically manages RAM usage by intelligently co
- Default: 2.0 GB
- Determines the minimum amount of free RAM to maintain (in gigabytes)
### Extreme Cleanup Parameters
- **purge_threshold**:
- Range: 0.1 - 256.0 GB
- Default: 256.0 GB
- Temporary headroom used for the one-shot purge
- **restore behavior**:
- Automatically restores the mode and threshold that were active before cleanup
3. Connect the node anywhere in your workflow - it will continuously monitor and manage cache based on your settings
4. Optionally add the "RAMCacheExtremeCleanup" node at the end of a workflow to trigger a one-shot cleanup and restore the previous state
## How It Works
README_zh.md
+11 -4
View File
@@ -7,16 +7,17 @@
## 功能特点
- **动态缓存模式切换**：支持在CLASSIC（无内存回收）和RAM_PRESSURE（自动内存清理）模式之间切换
- **智能内存管理**：自动监控和清理RAM缓存，当内存不足时释放不必要的缓存数据
- **智能内存管理**：自动监控和清理 RAM 缓存，当内存不足时释放不必要的缓存数据
- **自定义清理阈值**：允许用户设置最小空闲内存保持量，灵活适应不同硬件配置
- **无缝数据迁移**：在切换缓存模式时保留现有缓存数据，避免重复计算
- **直观的节点接口**：简单易用的参数设置，适用于不同水平的用户
- **极限清理节点**：一次性强清理（包含虚拟内存占用）并自动恢复清理前的模式与阈值
## 安装方法
1. 确保已安装ComfyUI（支持2025.10.31及更新版本）
2. 克隆或下载本仓库
3. 将`ComfyUI_Dynamic_RAMCache`文件夹复制到ComfyUI的`custom_nodes`目录下
3. 将`ComfyUI_Dynamic-RAMCache`文件夹复制到ComfyUI的`custom_nodes`目录下
4. 重启ComfyUI
## 使用方法
@@ -26,7 +27,8 @@
- **mode**：选择缓存模式（CLASSIC或RAM_PRESSURE）
- **cleanup_threshold**：设置最小空闲内存保持量（GB）
3. 可以选择性地连接任意输入到`any_input`端口（节点会透传此输入）
4. 运行工作流
4. 可选：在工作流末尾添加`RAMCacheExtremeCleanup`节点，执行一次性清理并恢复之前状态
5. 运行工作流
## 参数说明
@@ -41,6 +43,11 @@
- 默认值：2.0 GB
- 说明：系统尝试维持的最小空闲内存量。当可用内存低于此值时，RAM_PRESSURE模式会触发缓存清理
### 极限清理参数
- **purge_threshold**：一次性清理时使用的临时阈值（默认 256.0 GB）
- **恢复行为**：自动恢复清理前的模式与阈值
### 输出
- **output_passthrough**：如果连接了输入，则透传输入值；否则返回None
@@ -98,4 +105,4 @@
## 贡献
欢迎提交问题报告和拉取请求！
欢迎提交问题报告和拉取请求！
__init__.py
+6 -4
View File
@@ -1,11 +1,13 @@
from .nodes import DynamicRAMCacheControl
from .nodes import DynamicRAMCacheControl, RAMCacheExtremeCleanup
NODE_CLASS_MAPPINGS = {
"DynamicRAMCacheControl": DynamicRAMCacheControl
"DynamicRAMCacheControl": DynamicRAMCacheControl,
"RAMCacheExtremeCleanup": RAMCacheExtremeCleanup
}
NODE_DISPLAY_NAME_MAPPINGS = {
"DynamicRAMCacheControl": "🔥 Dynamic RAM Cache Control"
"DynamicRAMCacheControl": "🔥 Dynamic RAM Cache Control",
"RAMCacheExtremeCleanup": "🧹 RAM Cache Extreme Cleanup"
}
__all__ = ["NODE_CLASS_MAPPINGS", "NODE_DISPLAY_NAME_MAPPINGS"]
__all__ = ["NODE_CLASS_MAPPINGS", "NODE_DISPLAY_NAME_MAPPINGS"]
nodes.py
+73 -14
View File
@@ -77,11 +77,7 @@ class DynamicRAMCacheControl:
target_mode_ram = "RAM_PRESSURE" in mode
executor = None
for obj in gc.get_objects():
if obj.__class__.__name__ == 'PromptExecutor':
executor = obj
break
executor = self._find_executor()
if executor is None:
logging.warning("[DynamicRAMCache] PromptExecutor not found.")
@@ -93,14 +89,8 @@ class DynamicRAMCacheControl:
old_ram_arg = executor.cache_args.get('ram', 0)
executor.cache_args['ram'] = cleanup_threshold
if not hasattr(executor, 'caches'):
logging.warning("[DynamicRAMCache] PromptExecutor has no 'caches' attribute.")
return
cache_set = executor.caches
if not hasattr(cache_set, 'outputs'):
logging.warning("[DynamicRAMCache] CacheSet has no 'outputs' attribute.")
cache_set = self._get_cache_set(executor)
if cache_set is None:
return
current_cache = cache_set.outputs
@@ -138,6 +128,24 @@ class DynamicRAMCacheControl:
except Exception:
pass
def _find_executor(self):
for obj in gc.get_objects():
if obj.__class__.__name__ == 'PromptExecutor':
return obj
return None
def _get_cache_set(self, executor):
if not hasattr(executor, 'caches'):
logging.warning("[DynamicRAMCache] PromptExecutor has no 'caches' attribute.")
return None
cache_set = executor.caches
if not hasattr(cache_set, 'outputs'):
logging.warning("[DynamicRAMCache] CacheSet has no 'outputs' attribute.")
return None
return cache_set
def _update_cache_set(self, cache_set, new_cache):
cache_set.outputs = new_cache
@@ -184,4 +192,55 @@ class DynamicRAMCacheControl:
new_cache.subcaches = old_cache.subcaches
new_cache.dynprompt = getattr(old_cache, 'dynprompt', None)
new_cache.cache_key_set = getattr(old_cache, 'cache_key_set', None)
new_cache.initialized = getattr(old_cache, 'initialized', False)
new_cache.initialized = getattr(old_cache, 'initialized', False)
class RAMCacheExtremeCleanup(DynamicRAMCacheControl):
def __init__(self):
pass
@classmethod
def INPUT_TYPES(s):
return {
"required": {
"purge_threshold": ("FLOAT", {"default": 256.0, "min": 0.1, "max": 256.0, "step": 0.1, "tooltip": "Minimum free RAM to maintain (GB)"}),
},
"optional": {
"any_input": (any_type, {}),
}
}
RETURN_TYPES = (any_type,)
RETURN_NAMES = ("output_passthrough",)
FUNCTION = "extreme_cleanup"
CATEGORY = "utils/dynamic_ramcache"
def extreme_cleanup(self, purge_threshold, any_input=None):
if caching is not None and execution is not None:
executor = self._find_executor()
if executor is None:
logging.warning("[DynamicRAMCache] PromptExecutor not found.")
else:
if not hasattr(executor, 'cache_args'):
executor.cache_args = {}
old_ram_arg = executor.cache_args.get('ram', 2.0)
cache_set = self._get_cache_set(executor)
if cache_set is not None:
RAMPressureCacheClass = getattr(caching, 'RAMPressureCache', None)
if RAMPressureCacheClass:
is_currently_ram = isinstance(cache_set.outputs, RAMPressureCacheClass)
old_mode = "RAM_PRESSURE (Auto Purge)" if is_currently_ram else "CLASSIC (No Eviction)"
else:
old_mode = "CLASSIC (No Eviction)"
self._execute_cache_logic("RAM_PRESSURE (Auto Purge)", purge_threshold)
self._execute_cache_logic(old_mode, old_ram_arg)
else:
logging.warning("[DynamicRAMCache] Plugin disabled: Missing internal modules.")
if any_input is not None:
return (any_input,)
else:
try:
from comfy_execution.graph import ExecutionBlocker
return (ExecutionBlocker(None),)
except ImportError:
return (None,)