ComfyUI/comfy/node_cache.py
efwfe 160246643f
Update node_cache.py
Do not filter the load device
2024-05-24 16:43:50 +08:00

73 lines
2.1 KiB
Python

from collections import OrderedDict, namedtuple
from functools import wraps
import os
import torch
MAX_CHECKPOINT_CACHE = int(os.environ.get("MAX_CHECKPOINT_CACHE", 1))
MAX_CONTROLNET_CACHE = int(os.environ.get("MAX_CONTROLNET_CACHE", 1))
MAX_VAE_CACHE = int(os.environ.get("MAX_VAE_CACHE", 1))
class Cache:
def __init__(self, max_cache_size=0):
self._max_size = max_cache_size
self._cache = OrderedDict()
@property
def cache(self):
return self._cache
@property
def size(self):
return self._max_size
def clear_cache(self):
while len(self._cache) > self.size:
k, _ = self._cache.popitem(last=False)
print(f"[==Cache==] Removed {k} from cache.")
def add_cache(self, k, item):
"""model patch change the key name of tensor, cache the names here"""
keys = list(item.keys())
values = list(item.values())
self._cache[k] = (keys, values)
self._cache.move_to_end(k)
def get_cache(self, k):
data = self._cache.get(k)
if data:
keys, values = data
return dict(zip(keys, values))
return None
caches_mapping = {
"checkpoints": Cache(max_cache_size=MAX_CHECKPOINT_CACHE),
"controlnet": Cache(max_cache_size=MAX_CONTROLNET_CACHE),
"vae": Cache(max_cache_size=MAX_VAE_CACHE),
}
def checkpoint_cache_decorator(func):
@wraps(func)
def wrapper(ckpt, *args, device=None, **kwargs):
model_dir, model_name = os.path.split(ckpt)
dir_name = os.path.basename(model_dir)
# get matched ckpt, vae, controlnet queue
cache_mapper = caches_mapping.get(dir_name, None)
if cache_mapper is not None:
if cache_mapper.get_cache(ckpt) is not None:
return cache_mapper.get_cache(ckpt)
sd = func(ckpt, *args, device=device, **kwargs)
if cache_mapper is not None:
cache_mapper.add_cache(ckpt, sd)
print(f"[==Cache==] Add {dir_name} {model_name} to cache")
cache_mapper.clear_cache()
return sd
return wrapper