summaryrefslogtreecommitdiff
path: root/configs/common/Simulation.py
blob: 5d4398c80b7709296460e1920246dd039a6329fb (plain)
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
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
# Copyright (c) 2012-2013 ARM Limited
# All rights reserved
# 
# The license below extends only to copyright in the software and shall
# not be construed as granting a license to any other intellectual
# property including but not limited to intellectual property relating
# to a hardware implementation of the functionality of the software
# licensed hereunder.  You may use the software subject to the license
# terms below provided that you ensure that this notice is replicated
# unmodified and in its entirety in all distributions of the software,
# modified or unmodified, in source code or in binary form.
#
# Copyright (c) 2006-2008 The Regents of The University of Michigan
# Copyright (c) 2010 Advanced Micro Devices, Inc.
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are
# met: redistributions of source code must retain the above copyright
# notice, this list of conditions and the following disclaimer;
# redistributions in binary form must reproduce the above copyright
# notice, this list of conditions and the following disclaimer in the
# documentation and/or other materials provided with the distribution;
# neither the name of the copyright holders nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#
# Authors: Lisa Hsu

import sys
from os import getcwd
from os.path import join as joinpath

import CpuConfig
import MemConfig

import m5
from m5.defines import buildEnv
from m5.objects import *
from m5.util import *

addToPath('../common')

def getCPUClass(cpu_type):
    """Returns the required cpu class and the mode of operation."""
    cls = CpuConfig.get(cpu_type)
    return cls, cls.memory_mode()

def setCPUClass(options):
    """Returns two cpu classes and the initial mode of operation.

       Restoring from a checkpoint or fast forwarding through a benchmark
       can be done using one type of cpu, and then the actual
       simulation can be carried out using another type. This function
       returns these two types of cpus and the initial mode of operation
       depending on the options provided.
    """

    TmpClass, test_mem_mode = getCPUClass(options.cpu_type)
    CPUClass = None
    if TmpClass.require_caches() and \
            not options.caches and not options.ruby:
        fatal("%s must be used with caches" % options.cpu_type)

    if options.checkpoint_restore != None:
        if options.restore_with_cpu != options.cpu_type:
            CPUClass = TmpClass
            TmpClass, test_mem_mode = getCPUClass(options.restore_with_cpu)
    elif options.fast_forward:
        CPUClass = TmpClass
        TmpClass = AtomicSimpleCPU
        test_mem_mode = 'atomic'

    return (TmpClass, test_mem_mode, CPUClass)

def setMemClass(options):
    """Returns a memory controller class."""

    return MemConfig.get(options.mem_type)

def setWorkCountOptions(system, options):
    if options.work_item_id != None:
        system.work_item_id = options.work_item_id
    if options.num_work_ids != None:
        system.num_work_ids = options.num_work_ids
    if options.work_begin_cpu_id_exit != None:
        system.work_begin_cpu_id_exit = options.work_begin_cpu_id_exit
    if options.work_end_exit_count != None:
        system.work_end_exit_count = options.work_end_exit_count
    if options.work_end_checkpoint_count != None:
        system.work_end_ckpt_count = options.work_end_checkpoint_count
    if options.work_begin_exit_count != None:
        system.work_begin_exit_count = options.work_begin_exit_count
    if options.work_begin_checkpoint_count != None:
        system.work_begin_ckpt_count = options.work_begin_checkpoint_count
    if options.work_cpus_checkpoint_count != None:
        system.work_cpus_ckpt_count = options.work_cpus_checkpoint_count

def findCptDir(options, cptdir, testsys):
    """Figures out the directory from which the checkpointed state is read.

    There are two different ways in which the directories holding checkpoints
    can be named --
    1. cpt.<benchmark name>.<instruction count when the checkpoint was taken>
    2. cpt.<some number, usually the tick value when the checkpoint was taken>

    This function parses through the options to figure out which one of the
    above should be used for selecting the checkpoint, and then figures out
    the appropriate directory.
    """

    from os.path import isdir, exists
    from os import listdir
    import re

    if not isdir(cptdir):
        fatal("checkpoint dir %s does not exist!", cptdir)

    cpt_starttick = 0
    if options.at_instruction or options.simpoint:
        inst = options.checkpoint_restore
        if options.simpoint:
            # assume workload 0 has the simpoint
            if testsys.cpu[0].workload[0].simpoint == 0:
                fatal('Unable to find simpoint')
            inst += int(testsys.cpu[0].workload[0].simpoint)

        checkpoint_dir = joinpath(cptdir, "cpt.%s.%s" % (options.bench, inst))
        if not exists(checkpoint_dir):
            fatal("Unable to find checkpoint directory %s", checkpoint_dir)
    else:
        dirs = listdir(cptdir)
        expr = re.compile('cpt\.([0-9]*)')
        cpts = []
        for dir in dirs:
            match = expr.match(dir)
            if match:
                cpts.append(match.group(1))

        cpts.sort(lambda a,b: cmp(long(a), long(b)))

        cpt_num = options.checkpoint_restore
        if cpt_num > len(cpts):
            fatal('Checkpoint %d not found', cpt_num)

        cpt_starttick = int(cpts[cpt_num - 1])
        checkpoint_dir = joinpath(cptdir, "cpt.%s" % cpts[cpt_num - 1])

    return cpt_starttick, checkpoint_dir

def scriptCheckpoints(options, maxtick, cptdir):
    if options.at_instruction or options.simpoint:
        checkpoint_inst = int(options.take_checkpoints)

        # maintain correct offset if we restored from some instruction
        if options.checkpoint_restore != None:
            checkpoint_inst += options.checkpoint_restore

        print "Creating checkpoint at inst:%d" % (checkpoint_inst)
        exit_event = m5.simulate()
        exit_cause = exit_event.getCause()
        print "exit cause = %s" % exit_cause

        # skip checkpoint instructions should they exist
        while exit_cause == "checkpoint":
            exit_event = m5.simulate()
            exit_cause = exit_event.getCause()

        if exit_cause == "a thread reached the max instruction count":
            m5.checkpoint(joinpath(cptdir, "cpt.%s.%d" % \
                    (options.bench, checkpoint_inst)))
            print "Checkpoint written."

    else:
        when, period = options.take_checkpoints.split(",", 1)
        when = int(when)
        period = int(period)
        num_checkpoints = 0

        exit_event = m5.simulate(when - m5.curTick())
        exit_cause = exit_event.getCause()
        while exit_cause == "checkpoint":
            exit_event = m5.simulate(when - m5.curTick())
            exit_cause = exit_event.getCause()

        if exit_cause == "simulate() limit reached":
            m5.checkpoint(joinpath(cptdir, "cpt.%d"))
            num_checkpoints += 1

        sim_ticks = when
        max_checkpoints = options.max_checkpoints

        while num_checkpoints < max_checkpoints and \
                exit_cause == "simulate() limit reached":
            if (sim_ticks + period) > maxtick:
                exit_event = m5.simulate(maxtick - sim_ticks)
                exit_cause = exit_event.getCause()
                break
            else:
                exit_event = m5.simulate(period)
                exit_cause = exit_event.getCause()
                sim_ticks += period
                while exit_event.getCause() == "checkpoint":
                    exit_event = m5.simulate(sim_ticks - m5.curTick())
                if exit_event.getCause() == "simulate() limit reached":
                    m5.checkpoint(joinpath(cptdir, "cpt.%d"))
                    num_checkpoints += 1

    return exit_event

def benchCheckpoints(options, maxtick, cptdir):
    exit_event = m5.simulate(maxtick - m5.curTick())
    exit_cause = exit_event.getCause()

    num_checkpoints = 0
    max_checkpoints = options.max_checkpoints

    while exit_cause == "checkpoint":
        m5.checkpoint(joinpath(cptdir, "cpt.%d"))
        num_checkpoints += 1
        if num_checkpoints == max_checkpoints:
            exit_cause = "maximum %d checkpoints dropped" % max_checkpoints
            break

        exit_event = m5.simulate(maxtick - m5.curTick())
        exit_cause = exit_event.getCause()

    return exit_event

def repeatSwitch(testsys, repeat_switch_cpu_list, maxtick, switch_freq):
    print "starting switch loop"
    while True:
        exit_event = m5.simulate(switch_freq)
        exit_cause = exit_event.getCause()

        if exit_cause != "simulate() limit reached":
            return exit_event

        m5.switchCpus(testsys, repeat_switch_cpu_list)

        tmp_cpu_list = []
        for old_cpu, new_cpu in repeat_switch_cpu_list:
            tmp_cpu_list.append((new_cpu, old_cpu))
        repeat_switch_cpu_list = tmp_cpu_list

        if (maxtick - m5.curTick()) <= switch_freq:
            exit_event = m5.simulate(maxtick - m5.curTick())
            return exit_event

def run(options, root, testsys, cpu_class):
    if options.checkpoint_dir:
        cptdir = options.checkpoint_dir
    elif m5.options.outdir:
        cptdir = m5.options.outdir
    else:
        cptdir = getcwd()

    if options.fast_forward and options.checkpoint_restore != None:
        fatal("Can't specify both --fast-forward and --checkpoint-restore")

    if options.standard_switch and not options.caches:
        fatal("Must specify --caches when using --standard-switch")

    if options.standard_switch and options.repeat_switch:
        fatal("Can't specify both --standard-switch and --repeat-switch")

    if options.repeat_switch and options.take_checkpoints:
        fatal("Can't specify both --repeat-switch and --take-checkpoints")

    np = options.num_cpus
    switch_cpus = None

    if options.prog_interval:
        for i in xrange(np):
            testsys.cpu[i].progress_interval = options.prog_interval

    if options.maxinsts:
        for i in xrange(np):
            testsys.cpu[i].max_insts_any_thread = options.maxinsts

    if cpu_class:
        switch_cpus = [cpu_class(switched_out=True, cpu_id=(i))
                       for i in xrange(np)]

        for i in xrange(np):
            if options.fast_forward:
                testsys.cpu[i].max_insts_any_thread = int(options.fast_forward)
            switch_cpus[i].system =  testsys
            switch_cpus[i].workload = testsys.cpu[i].workload
            switch_cpus[i].clk_domain = testsys.cpu[i].clk_domain
            # simulation period
            if options.maxinsts:
                switch_cpus[i].max_insts_any_thread = options.maxinsts
            # Add checker cpu if selected
            if options.checker:
                switch_cpus[i].addCheckerCpu()

        testsys.switch_cpus = switch_cpus
        switch_cpu_list = [(testsys.cpu[i], switch_cpus[i]) for i in xrange(np)]

    if options.repeat_switch:
        switch_class = getCPUClass(options.cpu_type)[0]
        if switch_class.require_caches() and \
                not options.caches:
            print "%s: Must be used with caches" % str(switch_class)
            sys.exit(1)
        if not switch_class.support_take_over():
            print "%s: CPU switching not supported" % str(switch_class)
            sys.exit(1)

        repeat_switch_cpus = [switch_class(switched_out=True, \
                                               cpu_id=(i)) for i in xrange(np)]

        for i in xrange(np):
            repeat_switch_cpus[i].system = testsys
            repeat_switch_cpus[i].workload = testsys.cpu[i].workload
            repeat_switch_cpus[i].clk_domain = testsys.cpu[i].clk_domain

            if options.maxinsts:
                repeat_switch_cpus[i].max_insts_any_thread = options.maxinsts

            if options.checker:
                repeat_switch_cpus[i].addCheckerCpu()

        testsys.repeat_switch_cpus = repeat_switch_cpus

        if cpu_class:
            repeat_switch_cpu_list = [(switch_cpus[i], repeat_switch_cpus[i])
                                      for i in xrange(np)]
        else:
            repeat_switch_cpu_list = [(testsys.cpu[i], repeat_switch_cpus[i])
                                      for i in xrange(np)]

    if options.standard_switch:
        switch_cpus = [TimingSimpleCPU(switched_out=True, cpu_id=(i))
                       for i in xrange(np)]
        switch_cpus_1 = [DerivO3CPU(switched_out=True, cpu_id=(i))
                        for i in xrange(np)]

        for i in xrange(np):
            switch_cpus[i].system =  testsys
            switch_cpus_1[i].system =  testsys
            switch_cpus[i].workload = testsys.cpu[i].workload
            switch_cpus_1[i].workload = testsys.cpu[i].workload
            switch_cpus[i].clk_domain = testsys.cpu[i].clk_domain
            switch_cpus_1[i].clk_domain = testsys.cpu[i].clk_domain

            # if restoring, make atomic cpu simulate only a few instructions
            if options.checkpoint_restore != None:
                testsys.cpu[i].max_insts_any_thread = 1
            # Fast forward to specified location if we are not restoring
            elif options.fast_forward:
                testsys.cpu[i].max_insts_any_thread = int(options.fast_forward)
            # Fast forward to a simpoint (warning: time consuming)
            elif options.simpoint:
                if testsys.cpu[i].workload[0].simpoint == 0:
                    fatal('simpoint not found')
                testsys.cpu[i].max_insts_any_thread = \
                    testsys.cpu[i].workload[0].simpoint
            # No distance specified, just switch
            else:
                testsys.cpu[i].max_insts_any_thread = 1

            # warmup period
            if options.warmup_insts:
                switch_cpus[i].max_insts_any_thread =  options.warmup_insts

            # simulation period
            if options.maxinsts:
                switch_cpus_1[i].max_insts_any_thread = options.maxinsts

            # attach the checker cpu if selected
            if options.checker:
                switch_cpus[i].addCheckerCpu()
                switch_cpus_1[i].addCheckerCpu()

        testsys.switch_cpus = switch_cpus
        testsys.switch_cpus_1 = switch_cpus_1
        switch_cpu_list = [(testsys.cpu[i], switch_cpus[i]) for i in xrange(np)]
        switch_cpu_list1 = [(switch_cpus[i], switch_cpus_1[i]) for i in xrange(np)]

    # set the checkpoint in the cpu before m5.instantiate is called
    if options.take_checkpoints != None and \
           (options.simpoint or options.at_instruction):
        offset = int(options.take_checkpoints)
        # Set an instruction break point
        if options.simpoint:
            for i in xrange(np):
                if testsys.cpu[i].workload[0].simpoint == 0:
                    fatal('no simpoint for testsys.cpu[%d].workload[0]', i)
                checkpoint_inst = int(testsys.cpu[i].workload[0].simpoint) + offset
                testsys.cpu[i].max_insts_any_thread = checkpoint_inst
                # used for output below
                options.take_checkpoints = checkpoint_inst
        else:
            options.take_checkpoints = offset
            # Set all test cpus with the right number of instructions
            # for the upcoming simulation
            for i in xrange(np):
                testsys.cpu[i].max_insts_any_thread = offset

    checkpoint_dir = None
    if options.checkpoint_restore:
        cpt_starttick, checkpoint_dir = findCptDir(options, cptdir, testsys)
    m5.instantiate(checkpoint_dir)

    # Handle the max tick settings now that tick frequency was resolved
    # during system instantiation
    # NOTE: the maxtick variable here is in absolute ticks, so it must
    # include any simulated ticks before a checkpoint
    explicit_maxticks = 0
    maxtick_from_abs = m5.MaxTick
    maxtick_from_rel = m5.MaxTick
    maxtick_from_maxtime = m5.MaxTick
    if options.abs_max_tick:
        maxtick_from_abs = options.abs_max_tick
        explicit_maxticks += 1
    if options.rel_max_tick:
        maxtick_from_rel = options.rel_max_tick
        if options.checkpoint_restore:
            # NOTE: this may need to be updated if checkpoints ever store
            # the ticks per simulated second
            maxtick_from_rel += cpt_starttick
            if options.at_instruction or options.simpoint:
                warn("Relative max tick specified with --at-instruction or" \
                     " --simpoint\n      These options don't specify the " \
                     "checkpoint start tick, so assuming\n      you mean " \
                     "absolute max tick")
        explicit_maxticks += 1
    if options.maxtime:
        maxtick_from_maxtime = m5.ticks.fromSeconds(options.maxtime)
        explicit_maxticks += 1
    if explicit_maxticks > 1:
        warn("Specified multiple of --abs-max-tick, --rel-max-tick, --maxtime."\
             " Using least")
    maxtick = min([maxtick_from_abs, maxtick_from_rel, maxtick_from_maxtime])

    if options.checkpoint_restore != None and maxtick < cpt_starttick:
        fatal("Bad maxtick (%d) specified: " \
              "Checkpoint starts starts from tick: %d", maxtick, cpt_starttick)

    if options.standard_switch or cpu_class:
        if options.standard_switch:
            print "Switch at instruction count:%s" % \
                    str(testsys.cpu[0].max_insts_any_thread)
            exit_event = m5.simulate()
        elif cpu_class and options.fast_forward:
            print "Switch at instruction count:%s" % \
                    str(testsys.cpu[0].max_insts_any_thread)
            exit_event = m5.simulate()
        else:
            print "Switch at curTick count:%s" % str(10000)
            exit_event = m5.simulate(10000)
        print "Switched CPUS @ tick %s" % (m5.curTick())

        m5.switchCpus(testsys, switch_cpu_list)

        if options.standard_switch:
            print "Switch at instruction count:%d" % \
                    (testsys.switch_cpus[0].max_insts_any_thread)

            #warmup instruction count may have already been set
            if options.warmup_insts:
                exit_event = m5.simulate()
            else:
                exit_event = m5.simulate(options.standard_switch)
            print "Switching CPUS @ tick %s" % (m5.curTick())
            print "Simulation ends instruction count:%d" % \
                    (testsys.switch_cpus_1[0].max_insts_any_thread)
            m5.switchCpus(testsys, switch_cpu_list1)

    # If we're taking and restoring checkpoints, use checkpoint_dir
    # option only for finding the checkpoints to restore from.  This
    # lets us test checkpointing by restoring from one set of
    # checkpoints, generating a second set, and then comparing them.
    if options.take_checkpoints and options.checkpoint_restore:
        if m5.options.outdir:
            cptdir = m5.options.outdir
        else:
            cptdir = getcwd()

    if options.take_checkpoints != None :
        # Checkpoints being taken via the command line at <when> and at
        # subsequent periods of <period>.  Checkpoint instructions
        # received from the benchmark running are ignored and skipped in
        # favor of command line checkpoint instructions.
        exit_event = scriptCheckpoints(options, maxtick, cptdir)
    else:
        if options.fast_forward:
            m5.stats.reset()
        print "**** REAL SIMULATION ****"

        # If checkpoints are being taken, then the checkpoint instruction
        # will occur in the benchmark code it self.
        if options.repeat_switch and maxtick > options.repeat_switch:
            exit_event = repeatSwitch(testsys, repeat_switch_cpu_list,
                                      maxtick, options.repeat_switch)
        else:
            exit_event = benchCheckpoints(options, maxtick, cptdir)

    print 'Exiting @ tick %i because %s' % (m5.curTick(), exit_event.getCause())
    if options.checkpoint_at_end:
        m5.checkpoint(joinpath(cptdir, "cpt.%d"))

    if not m5.options.interactive:
        sys.exit(exit_event.getCode())