-
Notifications
You must be signed in to change notification settings - Fork 0
/
datasets.py
231 lines (189 loc) · 13.9 KB
/
datasets.py
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
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
########################################################################################################################
# Author: Lorenzo Ferretti
# Year: 2018
#
# This files contains the class datasets containing the different datasets name, types and function used to retrieve
# the data from the exploration
########################################################################################################################
import sqlite3
import numpy as np
class Datasets:
def __init__(self, name):
self.benchmark_name = name
self.benchmarks_dict_type = {"ChenIDCt": "db", "adpcm_decode": "db", "adpcm_encode": "db",
"Reflection_coefficients": "db",
"Autocorrelation": "db", "test": None}
self.benchmarks_dictionary_data = {"ChenIDCt": self.get_chenidct_data,
"adpcm_decode": self.get_decode_data,
"adpcm_encode": self.get_encode_data,
"Reflection_coefficients": self.get_reflection_data,
"Autocorrelation": self.get_autocorr_data,
"Autocorrelation_extended": None,
"adpcm_decode_ck": None}
# Definition of adpcm_decode_ck extended_experiments
self.adpcm_decode_ck_unrolling = {'mac_loop': [0, 2, 5, 10], # 4
'update_loop': [0, 2, 5, 10], # 4
'main_loop': [0, 5, 10, 25, 50]} # 5
self.adpcm_decode_ck_bundling = [("compressed", "result"), ((0, 0), (0, 1))] # 2
self.adpcm_decode_ck_bundling_config = {'bundling': [0, 1]}
self.adpcm_decode_ck_inlining = {'upzero': [0, 1], # 2
'quantl': [0, 1]} # 2
self.adpcm_decode_ck_clocks = {'clock': [5, 10, 15, 20]} # 4
self.adpcm_decode_ck = {'unrolling': self.adpcm_decode_ck_unrolling,
'inlining': self.adpcm_decode_ck_inlining,
# 'pipelining': self.autcorrelation_extended_pipeline,
'bundling': self.adpcm_decode_ck_bundling_config,
# 'partitioning': self.autcorrelation_extended_partitioning,
'clock': self.adpcm_decode_ck_clocks}
self.adpcm_decode_ck_directives_ordered = [
('unrolling-main_loop', self.adpcm_decode_ck['unrolling']['main_loop']),
('unrolling-update_loop', self.adpcm_decode_ck['unrolling']['update_loop']),
('unrolling-mac_loop', self.adpcm_decode_ck['unrolling']['mac_loop']),
('clock-clock', self.adpcm_decode_ck['clock']['clock']),
('inlining-upzero', self.adpcm_decode_ck['inlining']['upzero']),
('inlining-quantl', self.adpcm_decode_ck['inlining']['quantl']),
('bundling-sets', self.adpcm_decode_ck['bundling']['bundling']),
]
# Definition of autocorrelation_extended_experiments
self.autcorrelation_extended_unrolling = {'max_loop': [0, 2, 4, 8, 16, 32, 40, 80, 160], # 9
'gsm_mult_loop': [0, 2, 4, 8, 16, 32, 40, 80, 160], # 9
'init_zero_loop': [0, 3, 9], # 3
'compute_loop': [0, 2, 4, 8, 19, 38, 76, 152], # 8
'left_shift_loop': [0, 3, 9], # 3
'rescaling_loop': [0, 2, 4, 8, 16, 32, 40, 80, 160]} # 9
self.autcorrelation_extended_inlining = {'gsm_norm': [0, 1]} # 2
# self.autcorrelation_extended_pipeline = {'gsm_norm': [0, 1] # 2
# 'max_loop': [0, 1], # 2
# 'gsm_mult_loop': [0, 1], # 2
# 'init_zero_loop': [0, 1], # 2
# 'compute_loop': [0, 1], # 2
# 'left_shift_loop': [0, 1], # 2
# 'rescaling_loop': [0, 1]} # 2
# This actually need to be defined manually
self.autcorrelation_extended_bundling = [("s", "L_ACF"), ((0, 0), (0, 1))] # 2
# self.autcorrelation_extended_bundling_sets = [(0, 0), (0, 1)] # 2
self.autcorrelation_extended_bundling_config = {'bundling': [0, 1]}
# self.autcorrelation_extended_partitioning = {'s': [0, 2, 4, 8]} # 5
#'L_ACF': [0, 3, 9], # 3
self.autcorrelation_extended_clocks = {'clock': [5, 10, 15, 20, 25, 30, 35, 40]} # 8
self.autcorrelation_extended = {'unrolling': self.autcorrelation_extended_unrolling,
'inlining': self.autcorrelation_extended_inlining,
# 'pipelining': self.autcorrelation_extended_pipeline,
'bundling': self.autcorrelation_extended_bundling_config,
# 'partitioning': self.autcorrelation_extended_partitioning,
'clock': self.autcorrelation_extended_clocks}
self.directive_dependences = [('unrolling-max_loop', 'pipelinining-max_loop'),
('unrolling-gsm_mult_loop', 'pipelinining-gsm_mult_loop'),
('unrolling-left_shift_loop', 'pipelinining-left_shift_loop'),
('unrolling-rescaling_loop', 'pipelinining-rescaling_loop'),
('unrolling-compute_loop', 'pipelinining-compute_loop'),
('unrolling-init_zero_loop', 'pipelinining-init_zero_loop')]
self.autcorrelation_extended_directives_ordered = [
('unrolling-max_loop', self.autcorrelation_extended['unrolling']['max_loop']),
('unrolling-rescaling_loop', self.autcorrelation_extended['unrolling']['rescaling_loop']),
('unrolling-gsm_mult_loop', self.autcorrelation_extended['unrolling']['gsm_mult_loop']),
('clock-clock', self.autcorrelation_extended['clock']['clock']),
('unrolling-compute_loop', self.autcorrelation_extended['unrolling']['compute_loop']),
# ('partitioning-s', self.autcorrelation_extended['partitioning']['s']),
('unrolling-init_zero_loop', self.autcorrelation_extended['unrolling']['init_zero_loop']),
('unrolling-left_shift_loop', self.autcorrelation_extended['unrolling']['left_shift_loop']),
# ('pipelining-left_shift_loop', self.autcorrelation_extended['pipelining']['left_shift_loop']),
# ('pipelining-gsm_mult_loop', self.autcorrelation_extended['pipelining']['gsm_mult_loop']),
# ('pipelining-gsm_norm', self.autcorrelation_extended['pipelining']['gsm_norm']),
('inlining-gsm_norm', self.autcorrelation_extended['inlining']['gsm_norm']),
('bundling-sets', self.autcorrelation_extended['bundling']['bundling']),
# ('pipelining-rescaling_loop', self.autcorrelation_extended['pipelining']['rescaling_loop']),
# ('pipelining-compute_loop', self.autcorrelation_extended['pipelining']['compute_loop']),
# ('pipelining-max_loop', self.autcorrelation_extended['pipelining']['max_loop']),
# ('pipelining-init_zero_loop', self.autcorrelation_extended['pipelining']['init_zero_loop']),
#('bundling-s', self.autcorrelation_extended['bundling']['s'])
]
# END OF BENCHMARK DEFINITION
self.benchmark_directives = {"ChenIDCt": ['column_loops', 'row_loops', 'bundle_b', 'accuracy_loops'],
"Autocorrelation_extended": (
self.autcorrelation_extended, self.autcorrelation_extended_directives_ordered),
"adpcm_decode": ['main_loop_unrolling', 'mac_loop_unrolling',
'update_loop_unrolling', 'encode_inline',
'help_function_unrolling', 'bundle_b'],
"adpcm_encode": ['main_loop_unrolling', 'mac_loop_unrolling',
'update_loop_unrolling', 'encode_inline',
'help_function_unrolling', 'bundle_b'],
"Reflection_coefficients": ['loop1', 'loop2', 'loop3', 'loop4',
'inline_abs', 'inline_norm', 'inline_div',
'inline_add', 'inline_mult_r', 'bundle_b'],
"Autocorrelation": ['max_loop', 'gsm_resc_loop', 'init_loop', 'compute_loop',
'shift_loop', 'bundle_b']}
function_to_call = self.benchmarks_dictionary_data[name]
if function_to_call is not None:
self.benchmark_synthesis_results, self.benchmark_configurations, \
self.benchmark_feature_sets, self.benchmark_directives = function_to_call()
else:
self.benchmark_synthesis_results, self.benchmark_configurations, \
self.benchmark_feature_sets, self.benchmark_directives = None, None, None, None
def get_chenidct_data(self):
directives_str = ",".join(self.benchmark_directives[self.benchmark_name])
conn = sqlite3.connect('./datasets/ChenIDct.'+self.benchmarks_dict_type[self.benchmark_name])
synthesis_result = conn.execute('select latencies, ffs from '+self.benchmark_name).fetchall()
configurations = conn.execute(
'select '+directives_str+' from '+self.benchmark_name).fetchall()
# Define DS
config_reordered, feature_sets_orderd, directives_ordered = self._define_desing_space(configurations)
return synthesis_result, config_reordered, feature_sets_orderd, directives_ordered
def get_encode_data(self):
directives_str = ",".join(self.benchmark_directives[self.benchmark_name])
conn = sqlite3.connect('./datasets/'+self.benchmark_name+'.'+self.benchmarks_dict_type[self.benchmark_name])
synthesis_result = conn.execute('select latencies, ffs from '+self.benchmark_name).fetchall()
configurations = conn.execute(
'select '+directives_str+' from '+self.benchmark_name).fetchall()
# Define DS
config_reordered, feature_sets_orderd, directives_ordered = self._define_desing_space(configurations)
return synthesis_result, config_reordered, feature_sets_orderd, directives_ordered
def get_decode_data(self):
directives_str = ",".join(self.benchmark_directives[self.benchmark_name])
conn = sqlite3.connect('./datasets/'+self.benchmark_name+'.'+self.benchmarks_dict_type[self.benchmark_name])
synthesis_result = conn.execute('select latencies, ffs from '+self.benchmark_name).fetchall()
configurations = conn.execute(
'select '+directives_str+' from '+self.benchmark_name).fetchall()
# Define DS
config_reordered, feature_sets_orderd, directives_ordered = self._define_desing_space(configurations)
return synthesis_result, config_reordered, feature_sets_orderd, directives_ordered
def get_autocorr_data(self):
directives_str = ",".join(self.benchmark_directives[self.benchmark_name])
conn = sqlite3.connect('./datasets/'+self.benchmark_name+'.'+self.benchmarks_dict_type[self.benchmark_name])
synthesis_result = conn.execute('select latencies, ffs from '+self.benchmark_name).fetchall()
configurations = conn.execute(
'select '+directives_str+' from '+self.benchmark_name).fetchall()
# Define DS
config_reordered, feature_sets_orderd, directives_ordered = self._define_desing_space(configurations)
return synthesis_result, config_reordered, feature_sets_orderd, directives_ordered
def get_reflection_data(self):
directives_str = ",".join(self.benchmark_directives[self.benchmark_name])
conn = sqlite3.connect('./datasets/'+self.benchmark_name+'.'+self.benchmarks_dict_type[self.benchmark_name])
synthesis_result = conn.execute('select intervals, ffs from '+self.benchmark_name).fetchall()
configurations = conn.execute(
'select '+directives_str+' from '+self.benchmark_name).fetchall()
# Define DS
config_reordered, feature_sets_orderd, directives_ordered = self._define_desing_space(configurations)
return synthesis_result, config_reordered, feature_sets_orderd, directives_ordered
def _define_desing_space(self, configurations):
feature_sets = []
config_list = [list(c) for c in configurations]
matrix = np.array(config_list)
matrix_shape = matrix.shape
for i in xrange(matrix_shape[1]):
column = matrix[:, i]
unique_vector = np.unique(column)
tmp = unique_vector.tolist()
tmp.sort()
feature_sets.append(tmp)
ordered_sets = [i[0] for i in sorted(enumerate(feature_sets), reverse=True, key=lambda x: len(x[1]))]
config_reordered = []
tmp = []
for r in configurations:
for i in ordered_sets:
tmp.append(r[i])
config_reordered.append(tmp)
tmp = []
feature_sets_orderd = [feature_sets[i] for i in ordered_sets]
directives_ordered = [self.benchmark_directives[self.benchmark_name][i] for i in ordered_sets]
return config_reordered, feature_sets_orderd, directives_ordered