{{ message }}
Repository navigation
Expand file tree
/
Copy pathmx_graalpython_benchmark.py
More file actions
1080 lines (914 loc) · 45.4 KB
/
Copy pathmx_graalpython_benchmark.py
File metadata and controls
1080 lines (914 loc) · 45.4 KB
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
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
# Copyright (c) 2018, 2025, Oracle and/or its affiliates.
# Copyright (c) 2013, Regents of the University of California
#
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without modification, are
# permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this list of
# conditions and the following disclaimer.
# 2. 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.
#
# 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 HOLDER 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.
from __future__ import print_function
import itertools
import functools
import statistics
import sys
import os
import re
import shlex
import subprocess
from abc import ABC, abstractproperty
from contextlib import contextmanager
from os.path import join
from datetime import datetime
from pathlib import Path
import mx
import mx_benchmark
from mx_benchmark import StdOutRule, java_vm_registry, OutputCapturingVm, GuestVm, VmBenchmarkSuite, AveragingBenchmarkMixin
from mx_graalpython_bench_param import HARNESS_PATH
# ----------------------------------------------------------------------------------------------------------------------
#
# the graalpython suite
#
# ----------------------------------------------------------------------------------------------------------------------
SUITE = mx.suite("graalpython")
DIR = Path(__file__).parent.resolve()
# ----------------------------------------------------------------------------------------------------------------------
#
# constants
#
# ----------------------------------------------------------------------------------------------------------------------
VM_NAME_GRAALPYTHON = "graalpython"
VM_NAME_GRAALPYTHON_SVM = "graalpython-svm"
GROUP_GRAAL = "Graal"
SUBGROUP_GRAAL_PYTHON = "graalpython"
PYTHON_VM_REGISTRY_NAME = "Python"
CONFIGURATION_DEFAULT = "default"
CONFIGURATION_INTERPRETER = "interpreter"
CONFIGURATION_NATIVE_INTERPRETER = "native-interpreter"
CONFIGURATION_DEFAULT_MULTI = "default-multi"
CONFIGURATION_INTERPRETER_MULTI = "interpreter-multi"
CONFIGURATION_NATIVE_INTERPRETER_MULTI = "native-interpreter-multi"
CONFIGURATION_DEFAULT_MULTI_TIER = "default-multi-tier"
CONFIGURATION_NATIVE = "native"
CONFIGURATION_NATIVE_MULTI = "native-multi"
CONFIGURATION_NATIVE_MULTI_TIER = "native-multi-tier"
CONFIGURATION_SANDBOXED = "sandboxed"
CONFIGURATION_SANDBOXED_MULTI = "sandboxed-multi"
PYTHON_JAVA_EMBEDDING_VM_REGISTRY_NAME = "PythonJavaDriver"
CONFIGURATION_JAVA_EMBEDDING_MULTI = "java-driver-multi-default"
CONFIGURATION_JAVA_EMBEDDING_MULTI_SHARED = "java-driver-multi-shared"
CONFIGURATION_JAVA_EMBEDDING_INTERPRETER_MULTI = "java-driver-interpreter-multi"
CONFIGURATION_JAVA_EMBEDDING_INTERPRETER_MULTI_SHARED = "java-driver-interpreter-multi-shared"
DEFAULT_ITERATIONS = 10
BENCH_BGV = 'benchmarks-bgv'
# ----------------------------------------------------------------------------------------------------------------------
#
# utils
#
# ----------------------------------------------------------------------------------------------------------------------
def is_sandboxed_configuration(conf):
return conf in (CONFIGURATION_SANDBOXED, CONFIGURATION_SANDBOXED_MULTI)
@contextmanager
def environ(env):
def _handle_var(key_value):
(k, v) = key_value
if v is None:
del os.environ[k]
else:
os.environ[k] = str(v)
if env:
prev_env = {v: os.getenv(v) for v in env}
list(map(_handle_var, list(env.items())))
else:
prev_env = None
try:
yield
finally:
if prev_env:
list(map(_handle_var, list(prev_env.items())))
# ----------------------------------------------------------------------------------------------------------------------
#
# the vm definitions
#
# ----------------------------------------------------------------------------------------------------------------------
class AbstractPythonVm(OutputCapturingVm, ABC):
def __init__(self, name, config_name, options=None):
super().__init__()
self._name = name
self._config_name = config_name
self._options = options
@property
def options(self):
return self._options
def name(self):
return self._name
def config_name(self):
return self._config_name
@abstractproperty
def interpreter(self):
return None
def post_process_command_line_args(self, args):
return args
def run_vm(self, args, out=None, err=None, cwd=None, nonZeroIsFatal=False, env=None):
cmd = [self.interpreter] + args
cmd = mx.apply_command_mapper_hooks(cmd, self.command_mapper_hooks)
mx.logv(shlex.join(cmd))
return mx.run(
cmd,
out=out,
err=err,
cwd=cwd,
nonZeroIsFatal=nonZeroIsFatal,
env=env,
)
class AbstractPythonIterationsControlVm(AbstractPythonVm):
def __init__(self, name, config_name, options=None, iterations=None):
super().__init__(name, config_name, options=options)
try:
self._iterations = int(iterations)
except:
self._iterations = None
def override_iterations(self, requested_iterations):
return self._iterations if self._iterations is not None else requested_iterations
def _override_iterations_args(self, args):
_args = []
i = 0
while i < len(args):
arg = args[i]
_args.append(arg)
if arg == '-i':
_args.append(str(self.override_iterations(int(args[i + 1]))))
i += 1
i += 1
return _args
def run_vm(self, args, *splat, **kwargs):
args = self._override_iterations_args(args)
return super().run_vm(args, *splat, **kwargs)
class CPythonVm(AbstractPythonIterationsControlVm):
def __init__(self, config_name, options=None, virtualenv=None, iterations=0):
super().__init__("cpython", config_name, options=options, iterations=iterations)
self._virtualenv = virtualenv
def override_iterations(self, requested_iterations):
# CPython has no JIT right now, just a quickening interpreter
return min(requested_iterations, 3)
@property
def interpreter(self):
candidates_pre = [
self._virtualenv,
mx.get_env("VIRTUAL_ENV"),
mx.get_env("PYTHON3_HOME"),
]
candidates_suf = [
join("bin", "python3"),
join("bin", "python"),
"python3",
"python",
sys.executable,
]
for p, s in itertools.product(candidates_pre, candidates_suf):
if os.path.exists(exe := os.path.join(p or "", s)):
mx.log(f"CPython VM {exe=}")
return exe
assert False, "sys.executable should really exist"
def run_vm(self, args, *splat, **kwargs):
for idx, arg in enumerate(args):
if "--vm.Xmx" in arg:
mx.warn(f"Ignoring {arg}, cannot restrict memory on CPython.")
args = args[:idx] + args[idx + 1 :]
break
return super().run_vm(args, *splat, **kwargs)
class PyPyVm(AbstractPythonIterationsControlVm):
def __init__(self, config_name, options=None, virtualenv=None, iterations=0):
super().__init__("pypy", config_name, options=options, iterations=iterations)
def override_iterations(self, requested_iterations):
# PyPy warms up much faster, half should be enough
return int(requested_iterations / 2)
@property
def interpreter(self):
if home := mx.get_env("PYPY_HOME"):
exe = join(home, "bin", "pypy3")
else:
try:
exe = subprocess.check_output("which pypy3", shell=True).decode().strip()
except OSError:
mx.abort("PYPY_HOME is not set!")
mx.log(f"PyPy {exe=}")
return exe
def run_vm(self, args, *splat, env=None, **kwargs):
env = env or os.environ.copy()
xmxArg = re.compile("--vm.Xmx([0-9]+)([kKgGmM])")
pypyGcMax = "8GB"
for idx, arg in enumerate(args):
if m := xmxArg.search(arg):
args = args[:idx] + args[idx + 1 :]
pypyGcMax = f"{m.group(1)}{m.group(2).upper()}B"
mx.log(f"Setting PYPY_GC_MAX={pypyGcMax} via {arg}")
break
else:
mx.log(
f"Setting PYPY_GC_MAX={pypyGcMax}, use --vm.Xmx argument to override it"
)
env["PYPY_GC_MAX"] = pypyGcMax
return super().run_vm(args, *splat, env=env, **kwargs)
class GraalPythonVm(AbstractPythonIterationsControlVm):
def __init__(self, config_name, options=None, virtualenv=None, iterations=None, extra_polyglot_args=None):
super().__init__(VM_NAME_GRAALPYTHON, config_name, options=options, iterations=iterations)
self._extra_polyglot_args = extra_polyglot_args or []
@property
@functools.lru_cache
def launcher_type(self):
if mx.dependency("GRAALPY_NATIVE_STANDALONE", fatalIfMissing=False):
return "native"
else:
return "jvm"
@property
@functools.lru_cache
def interpreter(self):
from mx_graalpython import graalpy_standalone
launcher = graalpy_standalone(self.launcher_type, build=False)
mx.log(f"Using {launcher} based on enabled/excluded GraalPy standalone build targets.")
return launcher
def post_process_command_line_args(self, args):
if os.environ.get('BYTECODE_DSL_INTERPRETER', '').lower() == 'true' and not self.is_bytecode_dsl_config():
print("Found environment variable BYTECODE_DSL_INTERPRETER, but the guest vm config is not Bytecode DSL config.")
print("Did you want to use, e.g., `mx benchmark ... -- --host-vm-config=default-bc-dsl`?")
sys.exit(1)
return self.get_extra_polyglot_args() + args
def is_bytecode_dsl_config(self):
return '--vm.Dpython.EnableBytecodeDSLInterpreter=true' in self.get_extra_polyglot_args()
def extract_vm_info(self, args=None):
out_version = subprocess.check_output([self.interpreter, '--version'], universal_newlines=True)
# The benchmark data goes back a ways, we modify the reported dims for
# continuity with the historical queries
graalvm_version_match = re.search(r"\(([^\)]+ ((?:\d+\.?)+)).*\)", out_version)
if not graalvm_version_match:
mx.log(f"Using {out_version} as platform version string input")
graalvm_version_match = [out_version, out_version, out_version]
dims = {
'guest-vm': self.name(),
'guest-vm-config': self.config_name(),
'host-vm': 'graalvm-' + ('ee' if 'Oracle GraalVM' in out_version else 'ce'),
'host-vm-config': self.launcher_type,
"platform.graalvm-edition": 'EE' if 'Oracle GraalVM' in out_version else 'CE',
"platform.graalvm-version": graalvm_version_match[2],
"platform.graalvm-version-string": graalvm_version_match[1],
}
if dims['guest-vm-config'].endswith('-3-compiler-threads'):
dims['guest-vm-config'] = dims['guest-vm-config'].replace('-3-compiler-threads', '')
dims['host-vm-config'] += '-3-compiler-threads'
self._dims = dims
def run(self, *args, **kwargs):
code, out, dims = super().run(*args, **kwargs)
dims.update(self._dims)
is_bytecode_dsl_config = self.is_bytecode_dsl_config()
if "using bytecode DSL interpreter:" not in out:
# Let's be lenient unless in CI
print(f"BENCHMARK WARNING: could not verify whether running on bytecode DSL or not")
elif code == 0 and not f"using bytecode DSL interpreter: {is_bytecode_dsl_config}" in out:
print(f"ERROR: host VM config does not match what the the harness reported. "
f"Expected Bytecode DSL interpreter = {is_bytecode_dsl_config}. Harness output:\n{out}", file=sys.stderr)
return 1, out, dims
return code, out, dims
def get_extra_polyglot_args(self):
return ["--experimental-options", "-snapshot-startup", "--python.MaxNativeMemory=%s" % (2**34), *self._extra_polyglot_args]
class GraalPythonJavaDriverVm(GuestVm):
def __init__(self, config_name=CONFIGURATION_DEFAULT, cp_suffix=None, distributions=None, cp_prefix=None,
host_vm=None, extra_vm_args=None, extra_polyglot_args=None):
super().__init__(host_vm=host_vm)
self._config_name = config_name
self._distributions = distributions or ['GRAALPYTHON_BENCH']
self._cp_suffix = cp_suffix
self._cp_prefix = cp_prefix
self._extra_vm_args = extra_vm_args
self._extra_polyglot_args = extra_polyglot_args if isinstance(extra_polyglot_args, list) else []
def name(self):
return VM_NAME_GRAALPYTHON
def config_name(self):
return self._config_name
def hosting_registry(self):
return java_vm_registry
def get_classpath(self):
cp = []
if self._cp_prefix:
cp.append(self._cp_prefix)
if self._cp_suffix:
cp.append(self._cp_suffix)
return cp
def with_host_vm(self, host_vm):
return self.__class__(config_name=self._config_name, distributions=self._distributions,
cp_suffix=self._cp_suffix, cp_prefix=self._cp_prefix, host_vm=host_vm,
extra_vm_args=self._extra_vm_args, extra_polyglot_args=self._extra_polyglot_args)
def launcher_class(self):
return 'com.oracle.graal.python.benchmarks.JavaBenchmarkDriver'
def run(self, cwd, args):
extra_polyglot_args = self.get_extra_polyglot_args()
host_vm = self.host_vm()
cp = self.get_classpath()
jhm = mx.dependency("mx:JMH_1_21")
cp_deps = mx.classpath_entries(filter(None, [
mx.distribution('GRAALPYTHON_BENCH', fatalIfMissing=True),
mx.distribution('TRUFFLE_RUNTIME', fatalIfMissing=True),
mx.distribution('TRUFFLE_ENTERPRISE', fatalIfMissing=False),
jhm,
mx.dependency("sdk:LAUNCHER_COMMON")
] + jhm.deps))
cp += [x.classpath_repr() for x in cp_deps]
java_args = ['-cp', ':'.join(cp)] + [self.launcher_class()]
out = mx.TeeOutputCapture(mx.OutputCapture())
code = host_vm.run_java(java_args + extra_polyglot_args + args, cwd=cwd, out=out, err=out)
out = out.underlying.data
dims = host_vm.dimensions(cwd, args, code, out)
return code, out, dims
def get_extra_polyglot_args(self):
return ["--experimental-options", "--python.MaxNativeMemory=%s" % (2**34), *self._extra_polyglot_args]
# ----------------------------------------------------------------------------------------------------------------------
#
# the benchmark definition
#
# ----------------------------------------------------------------------------------------------------------------------
python_vm_registry = mx_benchmark.VmRegistry(PYTHON_VM_REGISTRY_NAME, known_host_registries=[java_vm_registry])
python_java_embedding_vm_registry = mx_benchmark.VmRegistry(PYTHON_JAVA_EMBEDDING_VM_REGISTRY_NAME, known_host_registries=[java_vm_registry])
class PythonBaseBenchmarkSuite(VmBenchmarkSuite, AveragingBenchmarkMixin):
def __init__(self, name, benchmarks):
super().__init__()
self._checkup = 'GRAALPYTHON_BENCHMARKS_CHECKUP' in os.environ
self._name = name
self._benchmarks = benchmarks
self._graph_dump_time = None
self._benchmarks_graphs = {}
def has_graph_tags(self, benchmark):
has_tag = False
self._benchmarks_graphs[benchmark] = set()
benchmark_path = join(self._bench_path, "{}.py".format(benchmark))
with open(benchmark_path, 'r') as fp:
text = fp.read()
for l in text.split('\n'):
if '# igv:' in l:
s = l.replace('# igv:', '').strip()
if s:
self._benchmarks_graphs[benchmark].add(s)
has_tag = True
return benchmark if has_tag else None
def has_graph_option(self, bmSuiteArgs):
if "--graph" in bmSuiteArgs:
import time
self._graph_dump_time = time.time()
os.mkdir(join(BENCH_BGV, '%d' % self._graph_dump_time))
bmSuiteArgs.remove('--graph')
return True
return False
def add_graph_dump_option(self, bmSuiteArgs, benchmark):
if self._graph_dump_time and self._benchmarks_graphs.get(benchmark):
dump_opt = '-Dgraal.Dump='
dump_path = join(BENCH_BGV, '%d' % self._graph_dump_time, 'run')
dump_path_opt = '-Dgraal.DumpPath=' + dump_path
if dump_opt not in bmSuiteArgs:
return [dump_opt, dump_path_opt]
return [dump_path_opt]
return []
def post_run_graph(self, benchmark, host_vm_config, guest_vm_config):
graph_tags = self._benchmarks_graphs.get(benchmark)
if self._graph_dump_time and graph_tags:
repo_hash = SUITE.vc.tip(SUITE.dir, 'master').strip()[:7]
name_format = '{h}.{g_vm}.{h_vm}.{s}.{b}.{t}.bgv'
dump_path = join(BENCH_BGV, '%d' % self._graph_dump_time, 'run')
selected_bgv = {}
num_tags = len(graph_tags)
for f in sorted(os.listdir(dump_path), reverse=True):
selected = False
if num_tags == 0 or not f.endswith('.bgv'):
os.remove(join(dump_path, f))
continue
for t in graph_tags:
if t in selected_bgv:
continue
if t in f:
graph_tags.remove(t)
selected = True
num_tags -= 1
selected_bgv[t] = f
break
if not selected:
os.remove(join(dump_path, f))
bench_bgv_dir = join(BENCH_BGV, '%d' % self._graph_dump_time)
for t in selected_bgv:
old_name = selected_bgv[t]
new_name = name_format.format(
h=repo_hash,
g_vm=guest_vm_config,
h_vm=host_vm_config,
s=self._name,
b=benchmark,
t=t)
print('rename: %s to %s' % (old_name, new_name))
os.rename(join(dump_path, old_name), join(bench_bgv_dir, new_name))
if graph_tags:
print('The following tags did not correspond to any bgv file:')
print('\n'.join(graph_tags))
raise FileNotFoundError
def benchmarkList(self, bmSuiteArgs):
if self.has_graph_option(bmSuiteArgs):
return [b for b in self._benchmarks.keys() if self.has_graph_tags(b)]
return list(self._benchmarks.keys())
def benchmarks(self):
raise FutureWarning('the benchmarks method has been deprecated for VmBenchmarkSuite instances, '
'use the benchmarkList method instead')
def successPatterns(self):
return [
re.compile(r"^### iteration=(?P<iteration>[0-9]+), name=(?P<benchmark>[a-zA-Z0-9.\-_]+), duration=(?P<time>[0-9]+(\.[0-9]+)?$)", re.MULTILINE), # pylint: disable=line-too-long
re.compile(r"^@@@ name=(?P<benchmark>[a-zA-Z0-9.\-_]+), duration=(?P<time>[0-9]+(\.[0-9]+)?$)", re.MULTILINE), # pylint: disable=line-too-long
]
def failurePatterns(self):
return [
# lookahead pattern for when truffle compilation details are enabled in the log
re.compile(r"^(?!(\[truffle\])).*Exception")
]
def group(self):
return GROUP_GRAAL
def name(self):
return self._name
def subgroup(self):
return SUBGROUP_GRAAL_PYTHON
@staticmethod
def with_branch_and_commit_dict(d):
"""
We run our benchmark from the graalpython directories, but with other
suites as primary suites in the CI, so we potentially want to update
branch and commit info.
"""
if mx.primary_suite().dir != os.getcwd():
if any(os.path.isdir(d) and d.startswith("mx.graalpython") for d in os.listdir()):
vc = SUITE.vc
if vc is None:
return d
# We want to report the commit from graalpython repo, but the branch from the current repo. The
# reason is that apptests benchmarks may be run from a PR and then the branch detection misbehaves in
# the gates, always reporting master for some reason.
branch = vc.active_branch(os.getcwd(), abortOnError=False) or "<unknown>"
info = vc.parent_info(SUITE.dir)
url = vc.default_pull(SUITE.dir, abortOnError=False) or "unknown"
d.update({
"branch": branch,
"commit.rev": vc.parent(SUITE.dir),
"commit.repo-url": url,
"commit.author": info["author"],
"commit.author-ts": info["author-ts"],
"commit.committer": info["committer"],
"commit.committer-ts": info["committer-ts"],
})
return d
def rules(self, output, benchmarks, bm_suite_args):
bench_name = self.get_bench_name(benchmarks)
arg = self.get_arg(self.runArgs(bm_suite_args), bench_name)
return [
# warmup curves
StdOutRule(
r"^### iteration=(?P<iteration>[0-9]+), name=(?P<benchmark>[a-zA-Z0-9._\-]+), duration=(?P<time>[0-9]+(\.[0-9]+)?$)", # pylint: disable=line-too-long
self.with_branch_and_commit_dict({
"benchmark": '{}.{}'.format(self._name, bench_name),
"metric.name": "warmup",
"metric.iteration": ("<iteration>", int),
"metric.type": "numeric",
"metric.value": ("<time>", float),
"metric.unit": "s",
"metric.score-function": "id",
"metric.better": "lower",
"config.run-flags": "".join(arg),
})
),
# secondary metric(s)
StdOutRule(
r"### WARMUP detected at iteration: (?P<endOfWarmup>[0-9]+$)",
self.with_branch_and_commit_dict({
"benchmark": '{}.{}'.format(self._name, bench_name),
"metric.name": "end-of-warmup",
"metric.iteration": 0,
"metric.type": "numeric",
"metric.value": ("<endOfWarmup>", int),
"metric.unit": "s",
"metric.score-function": "id",
"metric.better": "lower",
"config.run-flags": "".join(arg),
})
),
# no warmups
StdOutRule(
r"^@@@ name=(?P<benchmark>[a-zA-Z0-9._\-]+), duration=(?P<time>[0-9]+(\.[0-9]+)?$)", # pylint: disable=line-too-long
self.with_branch_and_commit_dict({
"benchmark": '{}.{}'.format(self._name, bench_name),
"metric.name": "time",
"metric.iteration": 0,
"metric.type": "numeric",
"metric.value": ("<time>", float),
"metric.unit": "s",
"metric.score-function": "id",
"metric.better": "lower",
"config.run-flags": "".join(arg),
})
),
]
def runAndReturnStdOut(self, benchmarks, bmSuiteArgs):
ret_code, out, dims = super().runAndReturnStdOut(benchmarks, bmSuiteArgs)
self.post_run_graph(benchmarks[0], dims['host-vm-config'], dims['guest-vm-config'])
if self._checkup:
self.checkup(out)
return ret_code, out, dims
def run(self, benchmarks, bm_suite_args):
if '--checkup' in bm_suite_args:
self._checkup = True
bm_suite_args.remove('--checkup')
results = super().run(benchmarks, bm_suite_args)
self.addAverageAcrossLatestResults(results)
return results
def defaultIterations(self, bm):
default_bench_args = self._benchmarks[bm]
if "-i" in default_bench_args:
bench_idx = default_bench_args.index("-i")
if bench_idx + 1 < len(default_bench_args):
return int(default_bench_args[bench_idx + 1])
return DEFAULT_ITERATIONS
def postprocess_run_args(self, run_args):
vm_options = []
remaining = []
i = 0
while i < len(run_args):
arg = run_args[i]
if not arg.startswith("-"):
remaining = run_args[i:]
break
if arg.startswith("-i"):
if len(run_args) >= i and run_args[i + 1] == "-1":
pass
elif self._checkup and len(run_args) >= i:
iterations = int(run_args[i + 1]) * 2
remaining = ["-i", str(iterations)] + run_args[i+2:]
break
else:
remaining = run_args[i:]
break
else:
vm_options.append(arg)
i += 1
if not (remaining and "-i" in remaining):
iterations = DEFAULT_ITERATIONS + self.getExtraIterationCount(DEFAULT_ITERATIONS)
if self._checkup:
iterations *= 2
remaining = ["-i", str(iterations)] + (remaining if remaining else [])
if self._checkup:
vm_options += ['--engine.TraceCompilation', '--vm.XX:+PrintGC']
remaining += ['--live-results']
return vm_options, remaining
def createCommandLineArgs(self, benchmarks, bmSuiteArgs):
return self.createVmCommandLineArgs(benchmarks, bmSuiteArgs)
def checkup(self, out):
lines = out.split('\n')
benchmark_name = None
current_iteration = -1
iterations_count = -1
iteration_times = []
gc_times = []
late_compilation = -1
for i in range(len(lines)):
line = lines[i]
# this marks the beginning of an output of a benchmark
benchmark_info = re.search("### (.*), \\d+ warmup iterations, (\\d+) bench iterations", line)
if benchmark_name is None and benchmark_info:
benchmark_name = benchmark_info.group(1)
iterations_count = int(benchmark_info.group(2))
mx.log("Checking benchmark %s with %d iterations" % (benchmark_name, iterations_count))
continue
if benchmark_name is None:
continue
# this marks the end of the processing of a single benchmark
warmup_match = re.search("### WARMUP detected at iteration: (\\d+)", line)
if warmup_match:
warmup = int(warmup_match.group(1))
for i in range(warmup, len(iteration_times)):
if gc_times[i] > iteration_times[i] / 10:
mx.warn("Benchmark checkup: %s: excessive GC pause of %.8f (on %d iteration)" % (benchmark_name, gc_times[i], i))
if warmup > iterations_count / 2:
mx.warn("Benchmark checkup: %s: warmup detected too late (on %d iteration)" % (benchmark_name, warmup))
if late_compilation > 0:
mx.warn("Benchmark checkup: %s: compilation detected too late (on %d iteration)" % (benchmark_name, late_compilation))
iteration_times = []
gc_times = []
current_iteration = -1
benchmark_name = None
late_compilation = False
continue
# following is done only when we are inside benchmark output:
iteration_info = re.search("### iteration=(\\d+), name=.*, duration=([0-9.]*)", line)
if iteration_info:
current_iteration += 1
iteration_times += [float(iteration_info.group(2))]
gc_times += [0.0]
continue
if current_iteration == -1:
continue
gc_log = re.search("\\[GC .* ([0-9,]*) secs]", line)
if gc_log:
gc_times[len(gc_times) - 1] += float(gc_log.group(1).replace(',', '.'))
if current_iteration >= iterations_count / 2 and "[engine] opt done" in line:
late_compilation = current_iteration
class PythonBenchmarkSuite(PythonBaseBenchmarkSuite):
def __init__(self, name, bench_path, benchmarks, python_path=None):
super().__init__(name, benchmarks)
self._python_path = python_path
self._harness_path = HARNESS_PATH
self._harness_path = join(SUITE.dir, self._harness_path)
if not self._harness_path:
mx.abort("python harness path not specified!")
if isinstance(bench_path, str):
self._bench_path = {bench: join(SUITE.dir, bench_path) for bench in benchmarks}
else:
self._bench_path = bench_path
assert isinstance(self._bench_path, dict), "bench_path is not a dict, got {}".format(self._bench_path)
assert self._bench_path.keys() == benchmarks.keys(), "not all benchmarks have a path: {}".format(
benchmarks.keys().difference(self._bench_path.keys()))
@staticmethod
def get_bench_name(benchmarks):
return os.path.basename(os.path.splitext(benchmarks[0])[0])
def get_arg(self, bmSuiteArgs, bench_name):
return " ".join(self._benchmarks[bench_name] + bmSuiteArgs)
def createVmCommandLineArgs(self, benchmarks, bmSuiteArgs):
if not benchmarks or len(benchmarks) != 1:
mx.abort("Please run a specific benchmark (mx benchmark {}:<benchmark-name>) or all the benchmarks "
"(mx benchmark {}:*)".format(self.name(), self.name()))
benchmark = benchmarks[0]
vm_args = self.add_graph_dump_option(bmSuiteArgs, benchmark)
vm_args += self.vmArgs(bmSuiteArgs)
run_args = self.runArgs(bmSuiteArgs)
cmd_args = [self._harness_path]
# resolve the harness python path (for external python modules, that may be required by the benchmark)
if self._python_path:
assert isinstance(self._python_path, list), "python_path must be a list"
python_path = []
for pth in self._python_path:
if hasattr(pth, '__call__'):
pth = pth()
assert isinstance(pth, str)
python_path.append(pth)
cmd_args += ['-p', ",".join(python_path)]
# the benchmark
cmd_args += [join(self._bench_path[benchmark], "{}.py".format(benchmark))]
if "-i" not in run_args:
run_args += self._benchmarks[benchmark]
num_iterations = self.defaultIterations(benchmark) + self.getExtraIterationCount(self.defaultIterations(benchmark))
run_args[run_args.index("-i") + 1] = str(num_iterations)
vm_options, run_args = self.postprocess_run_args(run_args)
cmd_args.extend(run_args)
return vm_options + vm_args + cmd_args
def get_vm_registry(self):
return python_vm_registry
@classmethod
def get_benchmark_suites(cls, benchmarks):
assert isinstance(benchmarks, dict), "benchmarks must be a dict: {suite: [path, {bench: args, ... }], ...}"
return [cls(suite_name, suite_info[0], suite_info[1])
for suite_name, suite_info in benchmarks.items()]
class PythonJavaEmbeddingBenchmarkSuite(PythonBaseBenchmarkSuite):
def __init__(self, name, bench_path, benchmarks):
super().__init__(name, benchmarks)
self._bench_path = bench_path
def get_vm_registry(self):
return python_java_embedding_vm_registry
def get_bench_name(self, benchmarks):
return os.path.basename(os.path.splitext(benchmarks[0])[0])
def get_arg(self, bmSuiteArgs, bench_name):
# returns arguments with which the benchmark was running for the results reporting
return " ".join(self._benchmarks[bench_name][1:] + bmSuiteArgs)
def createCommandLineArgs(self, benchmarks, bmSuiteArgs):
benchmark = benchmarks[0]
vm_args = self.add_graph_dump_option(bmSuiteArgs, benchmark)
vm_args += self.vmArgs(bmSuiteArgs)
run_args = self.runArgs(bmSuiteArgs)
if "-i" not in run_args:
# if not passed explicitly by the user, use the default args for this benchmark
run_args += self._benchmarks[benchmark]
# adds default -i value if missing and splits to VM options and args for the harness
vm_options, run_args = self.postprocess_run_args(run_args)
return vm_options + vm_args + ['-path', self._bench_path] + [benchmark] + run_args
@classmethod
def get_benchmark_suites(cls, benchmarks):
assert isinstance(benchmarks, dict), "benchmarks must be a dict: {suite: [path, {bench: args, ... }], ...}"
return [cls(suite_name, suite_info[0], suite_info[1])
for suite_name, suite_info in benchmarks.items()]
class PythonVmWarmupBenchmarkSuite(PythonBenchmarkSuite):
def rules(self, output, benchmarks, bm_suite_args):
bench_name = self.get_bench_name(benchmarks)
arg = self.get_arg(bm_suite_args, bench_name)
return [
# startup (difference between start of VM to end of first iteration)
StdOutRule(
r"### STARTUP +at iteration: (?P<iteration>[0-9]+), +duration: (?P<time>[0-9]+(\.[0-9]+)?$)",
self.with_branch_and_commit_dict({
"benchmark": '{}.{}'.format(self._name, bench_name),
"metric.name": "startup",
"metric.iteration": ("<iteration>", int),
"metric.type": "numeric",
"metric.value": ("<time>", float),
"metric.unit": "s",
"metric.score-function": "id",
"metric.better": "lower",
"config.run-flags": "".join(arg),
})
),
StdOutRule(
r"### EARLY WARMUP +at iteration: (?P<iteration>[0-9]+), +duration: (?P<time>[0-9]+(\.[0-9]+)?$)",
self.with_branch_and_commit_dict({
"benchmark": '{}.{}'.format(self._name, bench_name),
"metric.name": "early-warmup",
"metric.iteration": ("<iteration>", int),
"metric.type": "numeric",
"metric.value": ("<time>", float),
"metric.unit": "s",
"metric.score-function": "id",
"metric.better": "lower",
"config.run-flags": "".join(arg),
})
),
StdOutRule(
r"### LATE WARMUP +at iteration: (?P<iteration>[0-9]+), +duration: (?P<time>[0-9]+(\.[0-9]+)?$)",
self.with_branch_and_commit_dict({
"benchmark": '{}.{}'.format(self._name, bench_name),
"metric.name": "late-warmup",
"metric.iteration": ("<iteration>", int),
"metric.type": "numeric",
"metric.value": ("<time>", float),
"metric.unit": "s",
"metric.score-function": "id",
"metric.better": "lower",
"config.run-flags": "".join(arg),
})
),
]
class PythonParserBenchmarkSuite(PythonBaseBenchmarkSuite): # pylint: disable=too-many-ancestors
def get_vm_registry(self):
return java_vm_registry
def get_bench_name(self, benchmarks):
return benchmarks[0]
def createCommandLineArgs(self, benchmarks, bmSuiteArgs):
vmArgs = self.vmArgs(bmSuiteArgs)
dists = ["GRAALPYTHON", "GRAALPYTHON-LAUNCHER"]
if mx.suite("tools", fatalIfMissing=False):
dists.extend(('CHROMEINSPECTOR', 'TRUFFLE_PROFILER'))
vmArgs += [
"-Dorg.graalvm.language.python.home=%s" % mx.dependency("GRAALPYTHON_GRAALVM_SUPPORT").get_output(),
]
vmArgs += mx.get_runtime_jvm_args(dists + ['com.oracle.graal.python.benchmarks'], jdk=mx.get_jdk())
jmh_entry = ["com.oracle.graal.python.benchmarks.parser.ParserBenchRunner"]
runArgs = self.runArgs(bmSuiteArgs)
bench_name = benchmarks[0]
bench_args = self._benchmarks[bench_name]
return vmArgs + jmh_entry + runArgs + [bench_name] + bench_args
def get_arg(self, bmSuiteArgs, bench_name):
return " ".join(self._benchmarks[bench_name][1:] + bmSuiteArgs)
@classmethod
def get_benchmark_suites(cls, benchmarks):
assert isinstance(benchmarks, dict), "benchmarks must be a dict: {suite: {bench: args, ... }, ...}"
return [cls(suite_name, suite_info[0]) for suite_name, suite_info in benchmarks.items()]
class PythonJMHDistMxBenchmarkSuite(mx_benchmark.JMHDistBenchmarkSuite):
def name(self):
return "python-jmh"
def group(self):
return "Graal"
def subgroup(self):
return "graalpython"
def filter_distribution(self, dist):
# Note: for some reason the GRAALPYTHON_BENCH is filtered out by the base class,
# but by overriding this method we fix that and also get the nice property
# that one cannot accidentally run some other JMH benchmarks via this class
return dist.name == 'GRAALPYTHON_BENCH'
class LiveHeapTracker(mx_benchmark.Tracker):
def __init__(self, bmSuite):
super().__init__(bmSuite)
self.out_file = None
def map_command(self, cmd):
bench_name = self.bmSuite.currently_running_benchmark() if self.bmSuite else "benchmark"
if self.bmSuite:
bench_name = f"{self.bmSuite.name()}-{bench_name}"
ts = datetime.now().strftime("%Y%m%d-%H%M%S")
jmap_command = mx.get_jdk().exe_path('jmap')
self.out_file = os.path.join(os.getcwd(), f"heap_tracker_{bench_name}_{ts}.txt")
iterations = 3
return [sys.executable, str(DIR / 'live_heap_tracker.py'), self.out_file, str(iterations), jmap_command, *cmd]
def get_rules(self, bmSuiteArgs):
return [self.LiveHeapRule(self, bmSuiteArgs)]
class LiveHeapRule(mx_benchmark.Rule):
def __init__(self, tracker, bmSuiteArgs):
self.tracker = tracker
self.bmSuiteArgs = bmSuiteArgs
def parse(self, text):
with open(self.tracker.out_file) as f:
heap_mb = [int(line.strip()) / (1024 ** 2) for line in f if line]
os.unlink(self.tracker.out_file)
self.tracker.out_file = None
deciles = statistics.quantiles(heap_mb, n=10)
print(f"Heap size deciles (MiB): {deciles}")
return [
PythonBaseBenchmarkSuite.with_branch_and_commit_dict({
"benchmark": self.tracker.bmSuite.currently_running_benchmark(),
"bench-suite": self.tracker.bmSuite.benchSuiteName(self.bmSuiteArgs),
"config.vm-flags": ' '.join(self.tracker.bmSuite.vmArgs(self.bmSuiteArgs)),
"metric.name": "allocated-memory",
"metric.value": deciles[-1],
"metric.unit": "MB",
"metric.type": "numeric",
"metric.score-function": "id",
"metric.better": "lower",
"metric.iteration": 0
})
]
class PythonHeapBenchmarkSuite(PythonBaseBenchmarkSuite):
def __init__(self, name, bench_path, benchmarks):
super().__init__(name, benchmarks)
super().register_tracker('live-heap', LiveHeapTracker)
self._bench_path = bench_path
def get_vm_registry(self):
return python_vm_registry
def rules(self, output, benchmarks, bm_suite_args):
return [] # Tracker will add a rule
def register_tracker(self, name, tracker_type):
# We don't want any other trackers
pass
def createCommandLineArgs(self, benchmarks, bmSuiteArgs):
benchmark = benchmarks[0]
You can’t perform that action at this time.