1MCE::Stream(3) User Contributed Perl Documentation MCE::Stream(3)
2
3
4
6 MCE::Stream - Parallel stream model for chaining multiple maps and
7 greps
8
10 This document describes MCE::Stream version 1.862
11
13 ## Exports mce_stream, mce_stream_f, mce_stream_s
14 use MCE::Stream;
15
16 my (@m1, @m2, @m3);
17
18 ## Default mode is map and processed from right-to-left
19 @m1 = mce_stream sub { $_ * 3 }, sub { $_ * 2 }, 1..10000;
20 mce_stream \@m2, sub { $_ * 3 }, sub { $_ * 2 }, 1..10000;
21
22 ## Native Perl
23 @m3 = map { $_ * $_ } grep { $_ % 5 == 0 } 1..10000;
24
25 ## Streaming grep and map in parallel
26 mce_stream \@m3,
27 { mode => 'map', code => sub { $_ * $_ } },
28 { mode => 'grep', code => sub { $_ % 5 == 0 } }, 1..10000;
29
30 ## Array or array_ref
31 my @a = mce_stream sub { $_ * $_ }, 1..10000;
32 my @b = mce_stream sub { $_ * $_ }, \@list;
33
34 ## Important; pass an array_ref for deeply input data
35 my @c = mce_stream sub { $_->[1] *= 2; $_ }, [ [ 0, 1 ], [ 0, 2 ], ... ];
36 my @d = mce_stream sub { $_->[1] *= 2; $_ }, \@deeply_list;
37
38 ## File path, glob ref, IO::All::{ File, Pipe, STDIO } obj, or scalar ref
39 ## Workers read directly and not involve the manager process
40 my @e = mce_stream_f sub { chomp; $_ }, "/path/to/file"; # efficient
41
42 ## Involves the manager process, therefore slower
43 my @f = mce_stream_f sub { chomp; $_ }, $file_handle;
44 my @g = mce_stream_f sub { chomp; $_ }, $io;
45 my @h = mce_stream_f sub { chomp; $_ }, \$scalar;
46
47 ## Sequence of numbers (begin, end [, step, format])
48 my @i = mce_stream_s sub { $_ * $_ }, 1, 10000, 5;
49 my @j = mce_stream_s sub { $_ * $_ }, [ 1, 10000, 5 ];
50
51 my @k = mce_stream_s sub { $_ * $_ }, {
52 begin => 1, end => 10000, step => 5, format => undef
53 };
54
56 This module allows one to stream multiple map and/or grep operations in
57 parallel. Code blocks run simultaneously from right-to-left. The
58 results are appended immediately when providing a reference to an
59 array.
60
61 ## Appends are serialized, even out-of-order ok, but immediately.
62 ## Out-of-order chunks are held temporarily until ordered chunks
63 ## arrive.
64
65 mce_stream \@a, sub { $_ }, sub { $_ }, sub { $_ }, 1..10000;
66
67 ## input
68 ## chunk1 input
69 ## chunk3 chunk2 input
70 ## chunk2 chunk2 chunk3 input
71 ## append1 chunk3 chunk1 chunk4 input
72 ## append2 chunk1 chunk5 chunk5 input
73 ## append3 chunk5 chunk4 chunk6 ...
74 ## append4 chunk4 chunk6 ...
75 ## append5 chunk6 ...
76 ## append6 ...
77 ## ...
78 ##
79
80 MCE incurs a small overhead due to passing of data. A fast code block
81 will run faster natively when chaining multiple map functions. However,
82 the overhead will likely diminish as the complexity increases for the
83 code.
84
85 ## 0.334 secs -- baseline using the native map function
86 my @m1 = map { $_ * 4 } map { $_ * 3 } map { $_ * 2 } 1..1000000;
87
88 ## 0.427 secs -- this is quite amazing considering data passing
89 my @m2 = mce_stream
90 sub { $_ * 4 }, sub { $_ * 3 }, sub { $_ * 2 }, 1..1000000;
91
92 ## 0.355 secs -- appends to @m3 immediately, not after running
93 my @m3; mce_stream \@m3,
94 sub { $_ * 4 }, sub { $_ * 3 }, sub { $_ * 2 }, 1..1000000;
95
96 Even faster is mce_stream_s; useful when input data is a range of
97 numbers. Workers generate sequences mathematically among themselves
98 without any interaction from the manager process. Two arguments are
99 required for mce_stream_s (begin, end). Step defaults to 1 if begin is
100 smaller than end, otherwise -1.
101
102 ## 0.278 secs -- numbers are generated mathematically via sequence
103 my @m4; mce_stream_s \@m4,
104 sub { $_ * 4 }, sub { $_ * 3 }, sub { $_ * 2 }, 1, 1000000;
105
107 The following list options which may be overridden when loading the
108 module.
109
110 use Sereal qw( encode_sereal decode_sereal );
111 use CBOR::XS qw( encode_cbor decode_cbor );
112 use JSON::XS qw( encode_json decode_json );
113
114 use MCE::Stream
115 max_workers => 8, # Default 'auto'
116 chunk_size => 500, # Default 'auto'
117 tmp_dir => "/path/to/app/tmp", # $MCE::Signal::tmp_dir
118 freeze => \&encode_sereal, # \&Storable::freeze
119 thaw => \&decode_sereal, # \&Storable::thaw
120 default_mode => 'grep', # Default 'map'
121 fast => 1 # Default 0 (fast dequeue)
122 ;
123
124 From MCE 1.8 onwards, Sereal 3.015+ is loaded automatically if
125 available. Specify "Sereal => 0" to use Storable instead.
126
127 use MCE::Stream Sereal => 0;
128
130 MCE::Stream->init ( options )
131 MCE::Stream::init { options }
132
133 The init function accepts a hash of MCE options. The gather and
134 bounds_only options, if specified, are ignored due to being used
135 internally by the module (not shown below).
136
137 use MCE::Stream;
138
139 MCE::Stream::init {
140 chunk_size => 1, max_workers => 4,
141
142 user_begin => sub {
143 print "## ", MCE->wid, " started\n";
144 },
145
146 user_end => sub {
147 print "## ", MCE->wid, " completed\n";
148 }
149 };
150
151 my @a = mce_stream sub { $_ * $_ }, 1..100;
152
153 print "\n", "@a", "\n";
154
155 -- Output
156
157 ## 1 started
158 ## 2 started
159 ## 3 started
160 ## 4 started
161 ## 3 completed
162 ## 1 completed
163 ## 2 completed
164 ## 4 completed
165
166 1 4 9 16 25 36 49 64 81 100 121 144 169 196 225 256 289 324 361
167 400 441 484 529 576 625 676 729 784 841 900 961 1024 1089 1156
168 1225 1296 1369 1444 1521 1600 1681 1764 1849 1936 2025 2116 2209
169 2304 2401 2500 2601 2704 2809 2916 3025 3136 3249 3364 3481 3600
170 3721 3844 3969 4096 4225 4356 4489 4624 4761 4900 5041 5184 5329
171 5476 5625 5776 5929 6084 6241 6400 6561 6724 6889 7056 7225 7396
172 7569 7744 7921 8100 8281 8464 8649 8836 9025 9216 9409 9604 9801
173 10000
174
175 Like with MCE::Stream::init above, MCE options may be specified using
176 an anonymous hash for the first argument. Notice how both max_workers
177 and task_name can take an anonymous array for setting values uniquely
178 per each code block.
179
180 Remember that MCE::Stream processes from right-to-left when setting the
181 individual values.
182
183 use MCE::Stream;
184
185 my @a = mce_stream {
186 task_name => [ 'c', 'b', 'a' ],
187 max_workers => [ 2, 4, 3, ],
188
189 user_end => sub {
190 my ($mce, $task_id, $task_name) = @_;
191 print "$task_id - $task_name completed\n";
192 },
193
194 task_end => sub {
195 my ($mce, $task_id, $task_name) = @_;
196 MCE->print("$task_id - $task_name ended\n");
197 }
198 },
199 sub { $_ * 4 }, ## 2 workers, named c
200 sub { $_ * 3 }, ## 4 workers, named b
201 sub { $_ * 2 }, 1..10000; ## 3 workers, named a
202
203 -- Output
204
205 0 - a completed
206 0 - a completed
207 0 - a completed
208 0 - a ended
209 1 - b completed
210 1 - b completed
211 1 - b completed
212 1 - b completed
213 1 - b ended
214 2 - c completed
215 2 - c completed
216 2 - c ended
217
218 Note that the anonymous hash, for specifying options, also comes first
219 when passing an array reference.
220
221 my @a; mce_stream {
222 ...
223 }, \@a, sub { ... }, sub { ... }, 1..10000;
224
226 Scripts using MCE::Stream can be written using the long or short form.
227 The long form becomes relevant when mixing modes. Again, processing
228 occurs from right-to-left.
229
230 my @m3 = mce_stream
231 { mode => 'map', code => sub { $_ * $_ } },
232 { mode => 'grep', code => sub { $_ % 5 == 0 } }, 1..10000;
233
234 my @m4; mce_stream \@m4,
235 { mode => 'map', code => sub { $_ * $_ } },
236 { mode => 'grep', code => sub { $_ % 5 == 0 } }, 1..10000;
237
238 For multiple grep blocks, the short form can be used. Simply specify
239 the default mode for the module. The two valid values for default_mode
240 is 'grep' and 'map'.
241
242 use MCE::Stream default_mode => 'grep';
243
244 my @f = mce_stream_f sub { /ending$/ }, sub { /^starting/ }, $file;
245
246 The following assumes 'map' for default_mode in order to demonstrate
247 all the possibilities for providing input data.
248
249 MCE::Stream->run ( sub { code }, list )
250 mce_stream sub { code }, list
251
252 Input data may be defined using a list or an array reference. Unlike
253 MCE::Loop, Flow, and Step, specifying a hash reference as input data
254 isn't allowed.
255
256 ## Array or array_ref
257 my @a = mce_stream sub { $_ * 2 }, 1..1000;
258 my @b = mce_stream sub { $_ * 2 }, \@list;
259
260 ## Important; pass an array_ref for deeply input data
261 my @c = mce_stream sub { $_->[1] *= 2; $_ }, [ [ 0, 1 ], [ 0, 2 ], ... ];
262 my @d = mce_stream sub { $_->[1] *= 2; $_ }, \@deeply_list;
263
264 ## Not supported
265 my @z = mce_stream sub { ... }, \%hash;
266
267 MCE::Stream->run_file ( sub { code }, file )
268 mce_stream_f sub { code }, file
269
270 The fastest of these is the /path/to/file. Workers communicate the next
271 offset position among themselves with zero interaction by the manager
272 process.
273
274 "IO::All" { File, Pipe, STDIO } is supported since MCE 1.845.
275
276 my @c = mce_stream_f sub { chomp; $_ . "\r\n" }, "/path/to/file"; # faster
277 my @d = mce_stream_f sub { chomp; $_ . "\r\n" }, $file_handle;
278 my @e = mce_stream_f sub { chomp; $_ . "\r\n" }, $io; # IO::All
279 my @f = mce_stream_f sub { chomp; $_ . "\r\n" }, \$scalar;
280
281 MCE::Stream->run_seq ( sub { code }, $beg, $end [, $step, $fmt ] )
282 mce_stream_s sub { code }, $beg, $end [, $step, $fmt ]
283
284 Sequence may be defined as a list, an array reference, or a hash
285 reference. The functions require both begin and end values to run.
286 Step and format are optional. The format is passed to sprintf (% may be
287 omitted below).
288
289 my ($beg, $end, $step, $fmt) = (10, 20, 0.1, "%4.1f");
290
291 my @f = mce_stream_s sub { $_ }, $beg, $end, $step, $fmt;
292 my @g = mce_stream_s sub { $_ }, [ $beg, $end, $step, $fmt ];
293
294 my @h = mce_stream_s sub { $_ }, {
295 begin => $beg, end => $end, step => $step, format => $fmt
296 };
297
298 MCE::Stream->run ( { input_data => iterator }, sub { code } )
299 mce_stream { input_data => iterator }, sub { code }
300
301 An iterator reference may be specified for input_data. The only other
302 way is to specify input_data via MCE::Stream::init. This prevents
303 MCE::Stream from configuring the iterator reference as another user
304 task which will not work.
305
306 Iterators are described under section "SYNTAX for INPUT_DATA" at
307 MCE::Core.
308
309 MCE::Stream::init {
310 input_data => iterator
311 };
312
313 my @a = mce_stream sub { $_ * 3 }, sub { $_ * 2 };
314
316 MCE::Stream->finish
317 MCE::Stream::finish
318
319 Workers remain persistent as much as possible after running. Shutdown
320 occurs automatically when the script terminates. Call finish when
321 workers are no longer needed.
322
323 use MCE::Stream;
324
325 MCE::Stream::init {
326 chunk_size => 20, max_workers => 'auto'
327 };
328
329 my @a = mce_stream { ... } 1..100;
330
331 MCE::Stream::finish;
332
334 MCE, MCE::Core
335
337 Mario E. Roy, <marioeroy AT gmail DOT com>
338
339
340
341perl v5.30.0 2019-09-19 MCE::Stream(3)