1MCE::Map(3) User Contributed Perl Documentation MCE::Map(3)
2
6 MCE::Map - Parallel map model similar to the native map function
7
9 This document describes MCE::Map version 1.874
10
12 ## Exports mce_map, mce_map_f, and mce_map_s
13 use MCE::Map;
14 ## Array or array_ref
16 my @a = mce_map { $_ * $_ } 1..10000;
17 my @b = mce_map { $_ * $_ } \@list;
18 ## Important; pass an array_ref for deeply input data
20 my @c = mce_map { $_->[1] *= 2; $_ } [ [ 0, 1 ], [ 0, 2 ], ... ];
21 my @d = mce_map { $_->[1] *= 2; $_ } \@deeply_list;
22 ## File path, glob ref, IO::All::{ File, Pipe, STDIO } obj, or scalar ref
24 ## Workers read directly and not involve the manager process
25 my @e = mce_map_f { chomp; $_ } "/path/to/file"; # efficient
26 ## Involves the manager process, therefore slower
28 my @f = mce_map_f { chomp; $_ } $file_handle;
29 my @g = mce_map_f { chomp; $_ } $io;
30 my @h = mce_map_f { chomp; $_ } \$scalar;
31 ## Sequence of numbers (begin, end [, step, format])
33 my @i = mce_map_s { $_ * $_ } 1, 10000, 5;
34 my @j = mce_map_s { $_ * $_ } [ 1, 10000, 5 ];
35 my @k = mce_map_s { $_ * $_ } {
37 begin => 1, end => 10000, step => 5, format => undef
38 };
39
41 This module provides a parallel map implementation via Many-Core
42 Engine. MCE incurs a small overhead due to passing of data. A fast
43 code block will run faster natively. However, the overhead will likely
44 diminish as the complexity increases for the code.
45 my @m1 = map { $_ * $_ } 1..1000000; ## 0.127 secs
47 my @m2 = mce_map { $_ * $_ } 1..1000000; ## 0.304 secs
48 Chunking, enabled by default, greatly reduces the overhead behind the
50 scene. The time for mce_map below also includes the time for data
51 exchanges between the manager and worker processes. More
52 parallelization will be seen when the code incurs additional CPU time.
53 sub calc {
55 sqrt $_ * sqrt $_ / 1.3 * 1.5 / 3.2 * 1.07
56 }
57 my @m1 = map { calc } 1..1000000; ## 0.367 secs
59 my @m2 = mce_map { calc } 1..1000000; ## 0.365 secs
60 Even faster is mce_map_s; useful when input data is a range of numbers.
62 Workers generate sequences mathematically among themselves without any
63 interaction from the manager process. Two arguments are required for
64 mce_map_s (begin, end). Step defaults to 1 if begin is smaller than
65 end, otherwise -1.
66 my @m3 = mce_map_s { calc } 1, 1000000; ## 0.270 secs
68 Although this document is about MCE::Map, the MCE::Stream module can
70 write results immediately without waiting for all chunks to complete.
71 This is made possible by passing the reference to an array (in this
72 case @m4 and @m5).
73 use MCE::Stream;
75 sub calc {
77 sqrt $_ * sqrt $_ / 1.3 * 1.5 / 3.2 * 1.07
78 }
79 my @m4; mce_stream \@m4, sub { calc }, 1..1000000;
81 ## Completes in 0.272 secs. This is amazing considering the
83 ## overhead for passing data between the manager and workers.
84 my @m5; mce_stream_s \@m5, sub { calc }, 1, 1000000;
86 ## Completed in 0.176 secs. Like with mce_map_s, specifying a
88 ## sequence specification turns out to be faster due to lesser
89 ## overhead for the manager process.
90
92 The following list options which may be overridden when loading the
93 module.
94 use Sereal qw( encode_sereal decode_sereal );
96 use CBOR::XS qw( encode_cbor decode_cbor );
97 use JSON::XS qw( encode_json decode_json );
98 use MCE::Map
100 max_workers => 4, # Default 'auto'
101 chunk_size => 100, # Default 'auto'
102 tmp_dir => "/path/to/app/tmp", # $MCE::Signal::tmp_dir
103 freeze => \&encode_sereal, # \&Storable::freeze
104 thaw => \&decode_sereal # \&Storable::thaw
105 ;
106 From MCE 1.8 onwards, Sereal 3.015+ is loaded automatically if
108 available. Specify "Sereal => 0" to use Storable instead.
109 use MCE::Map Sereal => 0;
111
113 MCE::Map->init ( options )
114 MCE::Map::init { options }
115 The init function accepts a hash of MCE options. The gather option, if
117 specified, is ignored due to being used internally by the module.
118 use MCE::Map;
120 MCE::Map->init(
122 chunk_size => 1, max_workers => 4,
123 user_begin => sub {
125 print "## ", MCE->wid, " started\n";
126 },
127 user_end => sub {
129 print "## ", MCE->wid, " completed\n";
130 }
131 );
132 my @a = mce_map { $_ * $_ } 1..100;
134 print "\n", "@a", "\n";
136 -- Output
138 ## 2 started
140 ## 1 started
141 ## 3 started
142 ## 4 started
143 ## 1 completed
144 ## 4 completed
145 ## 2 completed
146 ## 3 completed
147 1 4 9 16 25 36 49 64 81 100 121 144 169 196 225 256 289 324 361
149 400 441 484 529 576 625 676 729 784 841 900 961 1024 1089 1156
150 1225 1296 1369 1444 1521 1600 1681 1764 1849 1936 2025 2116 2209
151 2304 2401 2500 2601 2704 2809 2916 3025 3136 3249 3364 3481 3600
152 3721 3844 3969 4096 4225 4356 4489 4624 4761 4900 5041 5184 5329
153 5476 5625 5776 5929 6084 6241 6400 6561 6724 6889 7056 7225 7396
154 7569 7744 7921 8100 8281 8464 8649 8836 9025 9216 9409 9604 9801
155 10000
156
158 MCE::Map->run ( sub { code }, list )
159 mce_map { code } list
160 Input data may be defined using a list or an array reference. Unlike
162 MCE::Loop, Flow, and Step, specifying a hash reference as input data
163 isn't allowed.
164 ## Array or array_ref
166 my @a = mce_map { $_ * 2 } 1..1000;
167 my @b = mce_map { $_ * 2 } \@list;
168 ## Important; pass an array_ref for deeply input data
170 my @c = mce_map { $_->[1] *= 2; $_ } [ [ 0, 1 ], [ 0, 2 ], ... ];
171 my @d = mce_map { $_->[1] *= 2; $_ } \@deeply_list;
172 ## Not supported
174 my @z = mce_map { ... } \%hash;
175 MCE::Map->run_file ( sub { code }, file )
177 mce_map_f { code } file
178 The fastest of these is the /path/to/file. Workers communicate the next
180 offset position among themselves with zero interaction by the manager
181 process.
182 "IO::All" { File, Pipe, STDIO } is supported since MCE 1.845.
184 my @c = mce_map_f { chomp; $_ . "\r\n" } "/path/to/file"; # faster
186 my @d = mce_map_f { chomp; $_ . "\r\n" } $file_handle;
187 my @e = mce_map_f { chomp; $_ . "\r\n" } $io; # IO::All
188 my @f = mce_map_f { chomp; $_ . "\r\n" } \$scalar;
189 MCE::Map->run_seq ( sub { code }, $beg, $end [, $step, $fmt ] )
191 mce_map_s { code } $beg, $end [, $step, $fmt ]
192 Sequence may be defined as a list, an array reference, or a hash
194 reference. The functions require both begin and end values to run.
195 Step and format are optional. The format is passed to sprintf (% may be
196 omitted below).
197 my ($beg, $end, $step, $fmt) = (10, 20, 0.1, "%4.1f");
199 my @f = mce_map_s { $_ } $beg, $end, $step, $fmt;
201 my @g = mce_map_s { $_ } [ $beg, $end, $step, $fmt ];
202 my @h = mce_map_s { $_ } {
204 begin => $beg, end => $end,
205 step => $step, format => $fmt
206 };
207 MCE::Map->run ( sub { code }, iterator )
209 mce_map { code } iterator
210 An iterator reference may be specified for input_data. Iterators are
212 described under section "SYNTAX for INPUT_DATA" at MCE::Core.
213 my @a = mce_map { $_ * 2 } make_iterator(10, 30, 2);
215
217 MCE::Map->finish
218 MCE::Map::finish
219 Workers remain persistent as much as possible after running. Shutdown
221 occurs automatically when the script terminates. Call finish when
222 workers are no longer needed.
223 use MCE::Map;
225 MCE::Map->init(
227 chunk_size => 20, max_workers => 'auto'
228 );
229 my @a = mce_map { ... } 1..100;
231 MCE::Map->finish;
233
235 MCE, MCE::Core
236
238 Mario E. Roy, <marioeroy AT gmail DOT com>
239perl v5.32.0 2020-08-19 MCE::Map(3)