1 # Copyright (c) 2003-2004 The Regents of The University of Michigan
4 # Redistribution and use in source and binary forms, with or without
5 # modification, are permitted provided that the following conditions are
6 # met: redistributions of source code must retain the above copyright
7 # notice, this list of conditions and the following disclaimer;
8 # redistributions in binary form must reproduce the above copyright
9 # notice, this list of conditions and the following disclaimer in the
10 # documentation and/or other materials provided with the distribution;
11 # neither the name of the copyright holders nor the names of its
12 # contributors may be used to endorse or promote products derived from
13 # this software without specific prior written permission.
15 # THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
16 # "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
17 # LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
18 # A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
19 # OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
20 # SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
21 # LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
22 # DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
23 # THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
24 # (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
25 # OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
27 # Authors: Nathan Binkert
32 def __init__(self
, options
):
33 self
.name
= options
.db
34 self
.host
= options
.host
35 self
.user
= options
.user
36 self
.passwd
= options
.passwd
42 self
.mydb
= MySQLdb
.connect(db
='mysql', host
=self
.host
, user
=self
.user
,
44 self
.cursor
= self
.mydb
.cursor()
48 self
.mydb
= MySQLdb
.connect(db
=self
.name
, host
=self
.host
,
49 user
=self
.user
, passwd
=self
.passwd
)
50 self
.cursor
= self
.mydb
.cursor()
53 if self
.mydb
is not None:
58 self
.cursor
.execute(sql
)
61 self
.query('DROP DATABASE IF EXISTS %s' % self
.name
)
64 self
.query('CREATE DATABASE %s' % self
.name
)
68 # Each run (or simulation) gets its own entry in the runs table to
69 # group stats by where they were generated
72 # 'id' is a unique identifier for each run to be used in other
74 # 'name' is the user designated name for the data generated. It is
75 # configured in the simulator.
76 # 'user' identifies the user that generated the data for the given
78 # 'project' another name to identify runs for a specific goal
79 # 'date' is a timestamp for when the data was generated. It can be
80 # used to easily expire data that was generated in the past.
81 # 'expire' is a timestamp for when the data should be removed from
82 # the database so we don't have years worth of junk.
85 # 'run' is indexed so you can find out details of a run if the run
86 # was retreived from the data table.
87 # 'name' is indexed so that two all run names are forced to be unique
91 rn_id SMALLINT UNSIGNED NOT NULL AUTO_INCREMENT,
92 rn_name VARCHAR(200) NOT NULL,
93 rn_sample VARCHAR(32) NOT NULL,
94 rn_user VARCHAR(32) NOT NULL,
95 rn_project VARCHAR(100) NOT NULL,
96 rn_date TIMESTAMP NOT NULL,
97 rn_expire TIMESTAMP NOT NULL,
99 UNIQUE (rn_name,rn_sample)
103 # We keep the bin names separate so that the data table doesn't get
104 # huge since bin names are frequently repeated.
107 # 'id' is the unique bin identifer.
108 # 'name' is the string name for the bin.
111 # 'bin' is indexed to get the name of a bin when data is retrieved
112 # via the data table.
113 # 'name' is indexed to get the bin id for a named bin when you want
114 # to search the data table based on a specific bin.
118 bn_id SMALLINT UNSIGNED NOT NULL AUTO_INCREMENT,
119 bn_name VARCHAR(255) NOT NULL,
125 # The stat table gives us all of the data for a particular stat.
128 # 'stat' is a unique identifier for each stat to be used in other
129 # tables for references.
130 # 'name' is simply the simulator derived name for a given
132 # 'descr' is the description of the statistic and what it tells
134 # 'type' defines what the stat tells you. Types are:
135 # SCALAR: A simple scalar statistic that holds one value
136 # VECTOR: An array of statistic values. Such a something that
137 # is generated per-thread. Vectors exist to give averages,
138 # pdfs, cdfs, means, standard deviations, etc across the
140 # DIST: Is a distribution of data. When the statistic value is
141 # sampled, its value is counted in a particular bucket.
142 # Useful for keeping track of utilization of a resource.
143 # (e.g. fraction of time it is 25% used vs. 50% vs. 100%)
144 # VECTORDIST: Can be used when the distribution needs to be
145 # factored out into a per-thread distribution of data for
146 # example. It can still be summed across threads to find
147 # the total distribution.
148 # VECTOR2D: Can be used when you have a stat that is not only
149 # per-thread, but it is per-something else. Like
151 # FORMULA: This statistic is a formula, and its data must be
152 # looked up in the formula table, for indicating how to
153 # present its values.
154 # 'subdata' is potentially used by any of the vector types to
155 # give a specific name to all of the data elements within a
157 # 'print' indicates whether this stat should be printed ever.
158 # (Unnamed stats don't usually get printed)
159 # 'prereq' only print the stat if the prereq is not zero.
160 # 'prec' number of decimal places to print
161 # 'nozero' don't print zero values
162 # 'nonan' don't print NaN values
163 # 'total' for vector type stats, print the total.
164 # 'pdf' for vector type stats, print the pdf.
165 # 'cdf' for vector type stats, print the cdf.
167 # The Following are for dist type stats:
168 # 'min' is the minimum bucket value. Anything less is an underflow.
169 # 'max' is the maximum bucket value. Anything more is an overflow.
170 # 'bktsize' is the approximate number of entries in each bucket.
171 # 'size' is the number of buckets. equal to (min/max)/bktsize.
174 # 'stat' is indexed so that you can find out details about a stat
175 # if the stat id was retrieved from the data table.
176 # 'name' is indexed so that you can simply look up data about a
181 st_id SMALLINT UNSIGNED NOT NULL AUTO_INCREMENT,
182 st_name VARCHAR(255) NOT NULL,
183 st_descr TEXT NOT NULL,
184 st_type ENUM("SCALAR", "VECTOR", "DIST", "VECTORDIST",
185 "VECTOR2D", "FORMULA") NOT NULL,
186 st_print BOOL NOT NULL,
187 st_prereq SMALLINT UNSIGNED NOT NULL,
188 st_prec TINYINT NOT NULL,
189 st_nozero BOOL NOT NULL,
190 st_nonan BOOL NOT NULL,
191 st_total BOOL NOT NULL,
192 st_pdf BOOL NOT NULL,
193 st_cdf BOOL NOT NULL,
194 st_min DOUBLE NOT NULL,
195 st_max DOUBLE NOT NULL,
196 st_bktsize DOUBLE NOT NULL,
197 st_size SMALLINT UNSIGNED NOT NULL,
203 # This is the main table of data for stats.
206 # 'stat' refers to the stat field given in the stat table.
208 # 'x' referrs to the first dimension of a multi-dimensional stat. For
209 # a vector, x will start at 0 and increase for each vector
211 # For a distribution:
212 # -1: sum (for calculating standard deviation)
213 # -2: sum of squares (for calculating standard deviation)
214 # -3: total number of samples taken (for calculating
215 # standard deviation)
220 # 'y' is used by a VECTORDIST and the VECTOR2D to describe the second
222 # 'run' is the run that the data was generated from. Details up in
224 # 'tick' is a timestamp generated by the simulator.
225 # 'bin' is the name of the bin that the data was generated in, if
227 # 'data' is the actual stat value.
230 # 'stat' is indexed so that a user can find all of the data for a
231 # particular stat. It is not unique, because that specific stat
232 # can be found in many runs, bins, and samples, in addition to
233 # having entries for the mulidimensional cases.
234 # 'run' is indexed to allow a user to remove all of the data for a
235 # particular execution run. It can also be used to allow the
236 # user to print out all of the data for a given run.
240 dt_stat SMALLINT UNSIGNED NOT NULL,
241 dt_x SMALLINT NOT NULL,
242 dt_y SMALLINT NOT NULL,
243 dt_run SMALLINT UNSIGNED NOT NULL,
244 dt_tick BIGINT UNSIGNED NOT NULL,
245 dt_bin SMALLINT UNSIGNED NOT NULL,
246 dt_data DOUBLE NOT NULL,
249 UNIQUE (dt_stat,dt_x,dt_y,dt_run,dt_tick,dt_bin)
253 # Names and descriptions for multi-dimensional stats (vectors, etc.)
254 # are stored here instead of having their own entry in the statistics
255 # table. This allows all parts of a single stat to easily share a
259 # 'stat' is the unique stat identifier from the stat table.
260 # 'x' is the first dimension for multi-dimensional stats
261 # corresponding to the data table above.
262 # 'y' is the second dimension for multi-dimensional stats
263 # corresponding to the data table above.
264 # 'name' is the specific subname for the unique stat,x,y combination.
265 # 'descr' is the specific description for the uniqe stat,x,y
269 # 'stat' is indexed so you can get the subdata for a specific stat.
272 CREATE TABLE subdata(
273 sd_stat SMALLINT UNSIGNED NOT NULL,
274 sd_x SMALLINT NOT NULL,
275 sd_y SMALLINT NOT NULL,
276 sd_name VARCHAR(255) NOT NULL,
278 UNIQUE (sd_stat,sd_x,sd_y)
283 # The formula table is maintained separately from the data table
284 # because formula data, unlike other stat data cannot be represented
288 # 'stat' refers to the stat field generated in the stat table.
289 # 'formula' is the actual string representation of the formula
293 # 'stat' is indexed so that you can just look up a formula.
296 CREATE TABLE formulas(
297 fm_stat SMALLINT UNSIGNED NOT NULL,
298 fm_formula BLOB NOT NULL,
303 # Each stat used in each formula is kept in this table. This way, if
304 # you want to print out a particular formula, you can simply find out
305 # which stats you need by looking in this table. Additionally, when
306 # you remove a stat from the stats table and data table, you remove
307 # any references to the formula in this table. When a formula is no
308 # longer referred to, you remove its entry.
311 # 'stat' is the stat id from the stat table above.
312 # 'child' is the stat id of a stat that is used for this formula.
313 # There may be many children for any given 'stat' (formula)
316 # 'stat' is indexed so you can look up all of the children for a
318 # 'child' is indexed so that you can remove an entry when a stat is
322 CREATE TABLE formula_ref(
323 fr_stat SMALLINT UNSIGNED NOT NULL,
324 fr_run SMALLINT UNSIGNED NOT NULL,
325 UNIQUE (fr_stat,fr_run),
331 # 'event' is the unique event id from the event_desc table
332 # 'run' is simulation run id that this event took place in
333 # 'tick' is the tick when the event happened
336 # 'event' is indexed so you can look up all occurences of a
338 # 'run' is indexed so you can find all events in a run
339 # 'tick' is indexed because we want the unique thing anyway
340 # 'event,run,tick' is unique combination
343 ev_event SMALLINT UNSIGNED NOT NULL,
344 ev_run SMALLINT UNSIGNED NOT NULL,
345 ev_tick BIGINT UNSIGNED NOT NULL,
349 UNIQUE(ev_event,ev_run,ev_tick)
353 # 'id' is the unique description id
354 # 'name' is the name of the event that occurred
357 # 'id' is indexed because it is the primary key and is what you use
358 # to look up the descriptions
359 # 'name' is indexed so one can find the event based on name
362 CREATE TABLE event_names(
363 en_id SMALLINT UNSIGNED NOT NULL AUTO_INCREMENT,
364 en_name VARCHAR(255) NOT NULL,
373 LEFT JOIN runs ON dt_run=rn_id
374 WHERE rn_id IS NULL''')
379 LEFT JOIN runs ON fr_run=rn_id
380 WHERE rn_id IS NULL''')
385 LEFT JOIN formula_ref ON fm_stat=fr_stat
386 WHERE fr_stat IS NULL''')
391 LEFT JOIN data ON st_id=dt_stat
392 WHERE dt_stat IS NULL''')
397 LEFT JOIN data ON sd_stat=dt_stat
398 WHERE dt_stat IS NULL''')
403 LEFT JOIN data ON bn_id=dt_bin
404 WHERE dt_bin IS NULL''')
409 LEFT JOIN runs ON ev_run=rn_id
410 WHERE rn_id IS NULL''')
415 LEFT JOIN events ON en_id=ev_event
416 WHERE ev_event IS NULL''')