1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
|
"""Small benchmark on the effect of chunksizes and compression on HDF5 files.
Francesc Alted
2007-11-25
"""
import math
import tempfile
import subprocess
from time import perf_counter as clock
from pathlib import Path
import numpy as np
import tables as tb
# Size of dataset
# N, M = 512, 2**16 # 256 MB
# N, M = 512, 2**18 # 1 GB
# N, M = 512, 2**19 # 2 GB
N, M = 2000, 1_000_000 # 15 GB
# N, M = 4000, 1000000 # 30 GB
datom = tb.Float64Atom() # elements are double precision
def quantize(data, least_significant_digit):
"""Quantize data to improve compression.
data is quantized using around(scale*data)/scale, where scale is
2**bits, and bits is determined from the least_significant_digit.
For example, if least_significant_digit=1, bits will be 4.
"""
precision = 10**-least_significant_digit
exp = math.log(precision, 10)
if exp < 0:
exp = math.floor(exp)
else:
exp = math.ceil(exp)
bits = math.ceil(math.log(10**-exp, 2))
scale = 2**bits
return np.around(scale * data) / scale
def get_db_size(filename):
sout = subprocess.Popen(
"ls -sh %s" % filename, shell=True, stdout=subprocess.PIPE
).stdout
line = sout[0]
return line.split()[0]
def bench(chunkshape, filters):
np.random.seed(1) # to have reproductible results
filename = tempfile.NamedTemporaryFile(suffix=".h5").name
print("Doing test on the file system represented by:", filename)
f = tb.open_file(filename, "w")
e = f.create_earray(
f.root,
"earray",
datom,
shape=(0, M),
filters=filters,
chunkshape=chunkshape,
)
# Fill the array
t1 = clock()
for i in range(N):
# e.append([np.random.rand(M)]) # use this for less compressibility
e.append([quantize(np.random.rand(M), 6)])
# os.system("sync")
print(f"Creation time: {clock() - t1:.3f}", end=" ")
filesize = get_db_size(filename)
filesize_bytes = Path(filename).stat().st_size
print("\t\tFile size: %d -- (%s)" % (filesize_bytes, filesize))
# Read in sequential mode:
e = f.root.earray
t1 = clock()
# Flush everything to disk and flush caches
# os.system("sync; echo 1 > /proc/sys/vm/drop_caches")
for row in e:
_ = row
print(f"Sequential read time: {clock() - t1:.3f}", end=" ")
# f.close()
# return
# Read in random mode:
i_index = np.random.randint(0, N, 128)
j_index = np.random.randint(0, M, 256)
# Flush everything to disk and flush caches
# os.system("sync; echo 1 > /proc/sys/vm/drop_caches")
# Protection against too large chunksizes
# 4 MB
if 0 and filters.complevel and chunkshape[0] * chunkshape[1] * 8 > 2**22:
f.close()
return
t1 = clock()
for i in i_index:
for j in j_index:
_ = e[i, j]
print(f"\tRandom read time: {clock() - t1:.3f}")
f.close()
# Benchmark with different chunksizes and filters
# for complevel in (0, 1, 3, 6, 9):
for complib in (None, "zlib", "lzo", "blosc"):
# for complib in ('blosc',):
if complib:
filters = tb.Filters(complevel=5, complib=complib)
else:
filters = tb.Filters(complevel=0)
print("8<--" * 20, "\nFilters:", filters, "\n" + "-" * 80)
# for ecs in (11, 14, 17, 20, 21, 22):
for ecs in range(10, 24):
# for ecs in (19,):
chunksize = 2**ecs
chunk1 = 1
chunk2 = chunksize / datom.itemsize
if chunk2 > M:
chunk1 = chunk2 / M
chunk2 = M
chunkshape = (chunk1, chunk2)
cs_str = str(chunksize / 1024) + " KB"
print("***** Chunksize:", cs_str, "/ Chunkshape:", chunkshape, "*****")
bench(chunkshape, filters)
|
