def __init__(self, cache_size, **kwargs):
self.cache_size = cache_size
self.hirs_ratio = 0.01
process_kwargs(self, kwargs, acceptable_kws=['hirs_ratio'])
self.hirs_limit = max(1,
int((self.cache_size * self.hirs_ratio) + 0.5))
self.lirs_limit = self.cache_size - self.hirs_limit
self.hirs_count = 0
self.lirs_count = 0
self.demoted = 0
self.nonresident = 0
# s stack, semi-split to find nonresident HIRs quickly
self.lirs = DequeDict()
self.hirs = DequeDict()
# q, the resident HIR stack
self.q = DequeDict()
self.time = 0
self.visual = Visualizinator(labels=['Q size'])
self.pollution = Pollutionator(cache_size)
def __init__(self, cache_size, **kwargs):
# Randomness and Time
np.random.seed(123)
self.time = 0
# Cache
self.cache_size = cache_size
#two stacks for the cache(s and q) to mark demoted items faster
self.s = DequeDict()
self.q = DequeDict()
#lfu heap
self.lfu = HeapDict()
# Histories
self.history_size = cache_size // 2
self.lru_hist = DequeDict()
self.lfu_hist = DequeDict()
# Decision Weights Initilized
self.initial_weight = 0.5
# Learning Rate
self.learning_rate = self.Cacheus_Learning_Rate(cache_size, **kwargs)
process_kwargs(self,
kwargs,
acceptable_kws=['initial_weight', 'history_size'])
# Decision Weights
self.W = np.array([self.initial_weight, 1 - self.initial_weight],
dtype=np.float32)
# Variables
hirsRatio = 0.01
self.q_limit = max(1, int((hirsRatio * self.cache_size) + 0.5))
self.s_limit = self.cache_size - self.q_limit
self.q_size = 0
self.s_size = 0
self.dem_count = 0
self.nor_count = 0
self.q_sizes = []
# Visualize
self.visual = Visualizinator(
labels=['W_lru-Cacheus', 'W_lfu-Cacheus', 'q_size'])
# Pollution
self.pollution = Pollutionator(cache_size)
def __init__(self, period_length, **kwargs):
self.learning_rate = np.sqrt((2.0 * np.log(2)) / period_length)
process_kwargs(self, kwargs, acceptable_kws=['learning_rate'])
self.learning_rate_reset = min(max(self.learning_rate, 0.001), 1)
self.learning_rate_curr = self.learning_rate
self.learning_rate_prev = 0.0
self.learning_rates = []
self.period_len = period_length
self.hitrate = 0
self.hitrate_prev = 0.0
self.hitrate_diff_prev = 0.0
self.hitrate_zero_count = 0
self.hitrate_nega_count = 0
def __init__(self, cache_size, **kwargs):
# Randomness and Time
np.random.seed(123)
self.time = 0
# Cache
self.cache_size = cache_size
self.lru = DequeDict()
self.lfu = HeapDict()
# Histories
self.history_size = cache_size // 2
self.lru_hist = DequeDict()
self.lfu_hist = DequeDict()
# Decision Weights Initilized
self.initial_weight = 0.5
# Fixed Learning Rate
self.learning_rate = 0.45
# Fixed Discount Rate
self.discount_rate = 0.005**(1 / self.cache_size)
# Apply values in kwargs, before any acceptable_kws
# members are prerequisites
process_kwargs(
self,
kwargs,
acceptable_kws=['learning_rate', 'initial_weight', 'history_size'])
# Decision Weights
self.W = np.array([self.initial_weight, 1 - self.initial_weight],
dtype=np.float32)
# Visualize
self.visual = Visualizinator(
labels=['W_lru-LeCaR', 'W_lfu-LeCaR', 'hit-rate'],
windowed_labels=['hit-rate'],
window_size=cache_size)
# Pollution
self.pollution = Pollutionator(cache_size)
def __init__(self, cache_size, **kwargs):
np.random.seed(123)
self.time = 0
self.cache_size = cache_size
kwargs_arc = {}
kwargs_lfu = {}
if 'arc' in kwargs:
kwargs_arc = kwargs['arc']
if 'lfu' in kwargs:
kwargs_lfu = kwargs['lfu']
self.arc = self.ARC(cache_size, **kwargs_arc)
self.lfu = self.LFU(cache_size, **kwargs_lfu)
self.history_size = cache_size
self.history = DequeDict()
self.initial_weight = 0.5
self.learning_rate = self.ARCALeCaR_Learning_Rate(cache_size, **kwargs)
process_kwargs(self,
kwargs,
acceptable_kws=['initial_weight', 'history_size'])
self.W = np.array([self.initial_weight, 1 - self.initial_weight],
dtype=np.float32)
self.visual = Visualizinator(labels=['W_arc', 'W_lfu', 'hit-rate'],
windowed_labels=['hit-rate'],
window_size=cache_size,
**kwargs)
self.pollution = Pollutionator(cache_size, **kwargs)
def __init__(self, cache_size, **kwargs):
self.cache_size = cache_size
self.hirs_ratio = 0.01
process_kwargs(self, kwargs, acceptable_kws=['hirs_ratio'])
self.hirs_limit = max(2, int((self.cache_size * self.hirs_ratio)))
self.lirs_limit = self.cache_size - self.hirs_limit
self.hirs_count = 0
self.lirs_count = 0
self.nonresident = 0
# s stack, semi-split to find nonresident HIRs quickly
self.s = DequeDict()
self.nr_hirs = DequeDict()
# q, the resident HIR stack
self.q = DequeDict()
self.time = 0
self.last_oblock = None
self.pollution = Pollutionator(cache_size)