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
|
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<title>Smart Pointer Timings</title>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
</head>
<body bgcolor="#FFFFFF">
<h1><img src="../../c++boost.gif" alt="c++boost.gif (8819 bytes)" align="middle" WIDTH="277" HEIGHT="86">Smart Pointer Timings</h1>
<p>In late January 2000, Mark Borgerding put forward a suggestion to boost for
a new design of smart pointer whereby an intrusive doubly linked list is used
to join together all instances of smart pointers sharing a given raw pointer.
This allowed avoidance of the costly heap allocation of a reference count that
occurred in the initial construction of the then current version of boost::shared_ptr.
Of course, nothing is for free and the benefit here was gained at the expense
of increased size and more costly copy operations. A debate ensued on the boost
mailing list and the tests which this page describes were performed to provide
a guide for current and future investigations into smart pointer implementation
strategies.</p>
<p>Thanks are due to <a href="../../people/dave_abrahams.htm">Dave Abrahams</a>,
<a href="../../people/gavin_collings.htm">Gavin Collings</a>,
<a href="../../people/greg_colvin.htm">Greg Colvin</a> and
<a href="../../people/beman_dawes.html">Beman Dawes</a>
for test code and trial implementations, the final version of which can be found
in .zip format <a href="smarttest.zip">here</a>.</p>
<h2>Description</h2>
<p>Two tests were run: the first aimed to obtain timings for two basic individual
operations:</p>
<ol type="i">
<li> Initial construction from raw pointer.</li>
<li> An amortized copy operation consisting of half an assignment and half a
copy construction - designed to reflect average usage.</li>
</ol>
<p>The second attempted to gain more insight into normal usage by timing the fill
and sort algorithms for vectors and lists filled with the various smart pointers.</p>
<p>Five smart pointer implementation strategies were tested:</p>
<ol type="i">
<li>Counted pointer using a heap allocated reference count, this is referred
to as <b>simple counted</b>.</li>
<li>Counted pointer using a special purpose allocator for the reference count
- <b>special counted</b>.</li>
<li>Counted pointer using an intrusive reference count - <b>intrusive</b>.</li>
<li>Linked pointer as described above - <b>linked</b>.</li>
<li>Cyclic pointer, a counted implementation using a std::deque for allocation
with provision for weak pointers and garbage collection of cycles of pointers
- <b>cyclic</b>.</li>
</ol>
<p>on two compilers:</p>
<ol type="i">
<li>MSVC 6.0 service pack 3, using default release optimization mode (/O2 -
optimized for speed, no inlining of functions defined outside a class body
unless specified as inline).</li>
<li>gcc 2.95.2 using full optimization (-O3 -DNDEBUG).</li>
</ol>
<p>Additionally, generated pointer sizes (taking into account struct alignment)
were compared, as were generated code sizes for MSVC mainly by manual inspection
of generated assembly code - a necessity due to function inlining.</p>
<p>All tests were run on a PII-200 running Windows NT version 4.0</p>
<h2> </h2>
<h2>Operation Timing Test Results</h2>
<p>The following graphs show the overall time in nanoseconds to acquire a pointer
(default construction) perform n amortized copy operations on it and finally
release it. The initial allocation time for the contained pointer is not included,
although the time for it's deallocation is. The contained pointer pointed to
a trivial class, but for the inclusion of an intrusive reference count for the
benefit of the intrusive counted shared pointer. A dumb pointer (i.e. a smart
pointer that simply acquires and releases its contained pointer with no extra
overhead) and a raw pointer were also included for comparison.</p>
<table border="0" align="center">
<tr>
<td width="20" height="20"> </td>
<td> </td>
<td width="20"> </td>
</tr>
<tr>
<td width="20"> </td>
<td><img src="msvcspeed.gif" width="560" height="355" alt="MSVC speed graph"></td>
<td width="20"> </td>
</tr>
<tr>
<td height="20"> </td>
<td> </td>
<td> </td>
</tr>
<tr>
<td> </td>
<td><img src="gccspeed.gif" width="560" height="355" alt="GCC speed graph"></td>
<td> </td>
</tr>
<tr>
<td height="20"> </td>
<td> </td>
<td> </td>
</tr>
</table>
<p> </p>
<p>Fitting straight lines to the above plots gives the following figures for initialization
and amortized copy operation for the two compilers (times in nanoseconds, errors
at two standard deviations) : -</p>
<p> </p>
<h4 align="center">MSVC</h4>
<table align="center" cellpadding="5" cellspacing="0" class="codetable" width="400">
<tr>
<th width="120">
<div align="right"></div>
</th>
<th class="codetabletop" width="120">
<div align="center">initialization</div>
</th>
<th class="codetabletop" width="120">copy operation</th>
</tr>
<tr>
<th class="codetableleft">
<div align="right">simple counted</div>
</th>
<td class="codetablecell" align="center">3000 +/- 170</td>
<td class="codetablecell" align="center">104 +/- 31</td>
</tr>
<tr>
<th class="codetableleft">
<div align="right">special counted</div>
</th>
<td class="codetablecell" align="center">1330 +/- 50</td>
<td class="codetablecell" align="center">85 +/- 9</td>
</tr>
<tr>
<th class="codetableleft">
<div align="right">intrusive</div>
</th>
<td class="codetablecell" align="center">1000 +/- 20</td>
<td class="codetablecell" align="center">71 +/- 3</td>
</tr>
<tr>
<th class="codetableleft" align="right">linked</th>
<td class="codetablecell" align="center">970 +/- 60</td>
<td class="codetablecell" align="center">136 +/- 10</td>
</tr>
<tr>
<th class="codetableleft" align="right">cyclic</th>
<td class="codetablecell" align="center">1290 +/- 70</td>
<td class="codetablecell" align="center">112 +/- 12</td>
</tr>
<tr>
<th class="codetableleft" align="right">dumb</th>
<td class="codetablecell" align="center">1020 +/- 20</td>
<td class="codetablecell" align="center">10 +/- 4</td>
</tr>
<tr>
<th class="codetableleft">
<div align="right">raw</div>
</th>
<td class="codetablecell" align="center">1038 +/- 30</td>
<td class="codetablecell" align="center">10 +/- 5</td>
</tr>
</table>
<h4 align="center"> </h4>
<h4 align="center">GCC</h4>
<table align="center" cellpadding="5" cellspacing="0" class="codetable" width="400">
<tr>
<th width="120">
<div align="right"></div>
</th>
<th class="codetabletop" width="120">
<div align="center">initialization</div>
</th>
<th class="codetabletop" width="120">copy operation</th>
</tr>
<tr>
<th class="codetableleft">
<div align="right">simple counted</div>
</th>
<td class="codetablecell" align="center">4620 +/- 150</td>
<td class="codetablecell" align="center">301 +/- 28</td>
</tr>
<tr>
<th class="codetableleft">
<div align="right">special counted</div>
</th>
<td class="codetablecell" align="center">1990 +/- 40</td>
<td class="codetablecell" align="center">264 +/- 7</td>
</tr>
<tr>
<th class="codetableleft">
<div align="right">intrusive</div>
</th>
<td class="codetablecell" align="center">1590 +/- 70</td>
<td class="codetablecell" align="center">181 +/- 12</td>
</tr>
<tr>
<th class="codetableleft" align="right">linked</th>
<td class="codetablecell" align="center">1470 +/- 140</td>
<td class="codetablecell" align="center">345 +/- 26</td>
</tr>
<tr>
<th class="codetableleft" align="right">cyclic</th>
<td class="codetablecell" align="center">2180 +/- 100</td>
<td class="codetablecell" align="center">330 +/- 18</td>
</tr>
<tr>
<th class="codetableleft" align="right">dumb</th>
<td class="codetablecell" align="center">1590 +/- 70</td>
<td class="codetablecell" align="center">74 +/- 12</td>
</tr>
<tr>
<th class="codetableleft" align="right">
<div align="right">raw</div>
</th>
<td class="codetablecell" align="center">1430 +/- 60</td>
<td class="codetablecell" align="center">27 +/- 11</td>
</tr>
</table>
<p>Note that the above times include a certain amount of loop overhead etc. for
each operation. An estimate of the pure smart pointer operation time 'overhead'
can be obtained by subtracting the dumb or raw figure from the smart pointer
time of interest.</p>
<h3>Detail</h3>
<p>The test involved iterating a loop which creates raw pointers. These were then
shared among a varying number (set size) of smart pointers. A range of set sizes
was used and then a line fitted to get a linear relation with number of initializations
and copy-operations. A spreadsheet was used for the line fit, and to produce
the performance graphs above.</p>
<h2> </h2>
<h2>Container Test Results</h2>
<p>To gain some insight in to operation within real life programs, this test was
devised. Smart pointers were used to fill standard containers which were then
sorted.</p>
<p>In this case, the contained pointer pointed to a class which initializes a
private data member to a random value in its default constructor. This value
is used subsequently for the sort comparison test. The class also contains an
intrusive reference count for the benefit of the intrusive counted pointer.</p>
<p> All times are in seconds for 300,000 contained pointers.</p>
<h4 align="center">GCC</h4>
<table align="center" cellpadding="5" cellspacing="0" class="codetable" width="500">
<tr>
<th> </th>
<th class="codetabletop" colspan="2">vector</th>
<th class="codetabletop" colspan="2">list</th>
</tr>
<tr>
<th width="120">
<div align="right"></div>
</th>
<th class="codetabletop2" width="80">
<div align="center">fill</div>
</th>
<th class="codetabletop2" width="80">sort</th>
<th class="codetabletop2" width="80">fill</th>
<th class="codetabletop2" width="80">sort</th>
</tr>
<tr>
<th class="codetableleft">
<div align="right">simple counted</div>
</th>
<td class="codetablecell" align="center">46.54</td>
<td class="codetablecell" align="center">2.44</td>
<td class="codetablecell" align="center">47.09</td>
<td class="codetablecell" align="center">3.22</td>
</tr>
<tr>
<th class="codetableleft">
<div align="right">special counted</div>
</th>
<td class="codetablecell" align="center">14.02</td>
<td class="codetablecell" align="center">2.83</td>
<td class="codetablecell" align="center">7.28</td>
<td class="codetablecell" align="center">3.21</td>
</tr>
<tr>
<th class="codetableleft">
<div align="right">intrusive</div>
</th>
<td class="codetablecell" align="center">12.15</td>
<td class="codetablecell" align="center">1.91</td>
<td class="codetablecell" align="center">7.99</td>
<td class="codetablecell" align="center">3.08</td>
</tr>
<tr>
<th class="codetableleft" align="right">linked</th>
<td class="codetablecell" align="center">12.46</td>
<td class="codetablecell" align="center">2.32</td>
<td class="codetablecell" align="center">8.14</td>
<td class="codetablecell" align="center">3.27</td>
</tr>
<tr>
<th class="codetableleft" align="right">cyclic</th>
<td class="codetablecell" align="center">22.60</td>
<td class="codetablecell" align="center">3.19</td>
<td class="codetablecell" align="center">1.63</td>
<td class="codetablecell" align="center">3.18</td>
</tr>
<tr>
<th class="codetableleft" align="right">
<div align="right">raw</div>
</th>
<td class="codetablecell" align="center">11.81</td>
<td class="codetablecell" align="center">0.24</td>
<td class="codetablecell" align="center">27.51</td>
<td class="codetablecell" align="center">0.77</td>
</tr>
</table>
<p> </p>
<h4 align="center">MSVC</h4>
<table align="center" cellpadding="5" cellspacing="0" class="codetable" width="500">
<tr>
<th> </th>
<th class="codetabletop" colspan="2">vector</th>
<th class="codetabletop" colspan="2">list</th>
</tr>
<tr>
<th width="120">
<div align="right"></div>
</th>
<th class="codetabletop2" width="80">
<div align="center">fill</div>
</th>
<th class="codetabletop2" width="80">sort</th>
<th class="codetabletop2" width="80">fill</th>
<th class="codetabletop2" width="80">sort</th>
</tr>
<tr>
<th class="codetableleft">
<div align="right">simple counted</div>
</th>
<td class="codetablecell" align="center">1.83</td>
<td class="codetablecell" align="center">2.37</td>
<td class="codetablecell" align="center">1.86</td>
<td class="codetablecell" align="center">4.85</td>
</tr>
<tr>
<th class="codetableleft">
<div align="right">special counted</div>
</th>
<td class="codetablecell" align="center">1.04</td>
<td class="codetablecell" align="center">2.35</td>
<td class="codetablecell" align="center">1.38</td>
<td class="codetablecell" align="center">4.58</td>
</tr>
<tr>
<th class="codetableleft">
<div align="right">intrusive</div>
</th>
<td class="codetablecell" align="center">1.04</td>
<td class="codetablecell" align="center">1.84</td>
<td class="codetablecell" align="center">1.16</td>
<td class="codetablecell" align="center">4.29</td>
</tr>
<tr>
<th class="codetableleft" align="right">linked</th>
<td class="codetablecell" align="center">1.08</td>
<td class="codetablecell" align="center">2.00</td>
<td class="codetablecell" align="center">1.21</td>
<td class="codetablecell" align="center">4.33</td>
</tr>
<tr>
<th class="codetableleft" align="right">cyclic</th>
<td class="codetablecell" align="center">1.38</td>
<td class="codetablecell" align="center">2.84</td>
<td class="codetablecell" align="center">1.47</td>
<td class="codetablecell" align="center">4.73</td>
</tr>
<tr>
<th class="codetableleft" align="right">
<div align="right">raw</div>
</th>
<td class="codetablecell" align="center">0.67</td>
<td class="codetablecell" align="center">0.28</td>
<td class="codetablecell" align="center">1.24</td>
<td class="codetablecell" align="center">1.81</td>
</tr>
</table>
<p> </p>
<h2>Code Size</h2>
<p>The following code sizes were determined by inspection of generated code for
MSVC only. Sizes are given in the form N / M / I where:</p>
<ul type="circle">
<li> N is the instruction count of the operation</li>
<li>M is the size of the code in bytes</li>
<li>I determines whether generated code was inlined or not I = inline, O = "outline"</li>
</ul>
<p> </p>
<table align="center" cellpadding="5" cellspacing="0" class="codetable" width="570">
<tr>
<th height="28" width="140">
<div align="right"></div>
</th>
<th height="28" class="codetabletop" width="80">
<div align="center">ptr()</div>
</th>
<th height="28" class="codetabletop" width="80">ptr(p)</th>
<th height="28" class="codetabletop" width="80">ptr(ptr)</th>
<th height="28" class="codetabletop" width="80">op=()</th>
<th height="28" class="codetabletop" width="80">
<div align="center">~ptr()</div>
</th>
</tr>
<tr>
<th class="codetableleft">
<div align="right">simple counted</div>
</th>
<td class="codetablecell" align="center">38/110/O</td>
<td class="codetablecell" align="center">38/110/O</td>
<td class="codetablecell" align="center">9/23/I</td>
<td class="codetablecell" align="center">22/57/I</td>
<td class="codetablecell" align="center">17/40/I</td>
</tr>
<tr>
<th class="codetableleft">
<div align="right">special counted</div>
</th>
<td class="codetablecell" align="center"><font size="-1">50/141/O</font></td>
<td class="codetablecell" align="center"><font size="-1">50/141/O</font></td>
<td class="codetablecell" align="center"><font size="-1">9/23/I</font></td>
<td class="codetablecell" align="center"><font size="-1">23/64/I</font></td>
<td class="codetablecell" align="center"><font size="-1">13/38/I</font></td>
</tr>
<tr>
<th class="codetableleft">
<div align="right">intrusive</div>
</th>
<td class="codetablecell" align="center"><font size="-1">1/2/I</font></td>
<td class="codetablecell" align="center"><font size="-1">3/6/I</font></td>
<td class="codetablecell" align="center"><font size="-1">3/6/I</font></td>
<td class="codetablecell" align="center"><font size="-1">6/11/I</font></td>
<td class="codetablecell" align="center"><font size="-1">6/11/I</font></td>
</tr>
<tr>
<th class="codetableleft">
<div align="right">linked</div>
</th>
<td class="codetablecell" align="center"><font size="-1">5/19/I</font></td>
<td class="codetablecell" align="center"><font size="-1">5/15/I</font></td>
<td class="codetablecell" align="center"><font size="-1">10/30/I</font></td>
<td class="codetablecell" align="center"><font size="-1">27/59/I</font></td>
<td class="codetablecell" align="center"><font size="-1">14/38/I</font></td>
</tr>
</table>
<p>During the code inspection, a couple of minor points were noticed: -</p>
<ul>
<li>Function inlining was critical to performance.</li>
<li>For MSVC, at least, a "delete 0" caused execution of 11 assembly
instructions, including a function call. So in cases where performance is
at an absolute premium it can be worth inserting the extra manual test.</li>
</ul>
<h2> </h2>
<h2>Data Size</h2>
<p>The following smart pointer sizes were obtained in bytes</p>
<table align="center" cellpadding="5" cellspacing="0" class="codetable" width="270">
<tr>
<th height="28" width="150">
<div align="right"></div>
</th>
<th height="28" class="codetabletop" width="60">
<div align="center">MSVC</div>
</th>
<th height="28" class="codetabletop" width="60">
<div align="center">GCC</div>
</th>
</tr>
<tr>
<th class="codetableleft">
<div align="right">simple counted</div>
</th>
<td class="codetablecell">
<div align="center">8</div>
</td>
<td class="codetablecell">
<div align="center">8</div>
</td>
</tr>
<tr>
<th class="codetableleft">
<div align="right">special counted</div>
</th>
<td class="codetablecell">
<div align="center">8</div>
</td>
<td class="codetablecell">
<div align="center">12</div>
</td>
</tr>
<tr>
<th class="codetableleft">
<div align="right">intrusive</div>
</th>
<td class="codetablecell">
<div align="center">4</div>
</td>
<td class="codetablecell">
<div align="center">4</div>
</td>
</tr>
<tr>
<th class="codetableleft">
<div align="right">linked</div>
</th>
<td class="codetablecell">
<div align="center">12</div>
</td>
<td class="codetablecell">
<div align="center">12</div>
</td>
</tr>
<tr>
<th class="codetableleft">
<div align="right">cyclic</div>
</th>
<td class="codetablecell">
<div align="center">8</div>
</td>
<td class="codetablecell">
<div align="center">8</div>
</td>
</tr>
</table>
<h2> </h2>
<h2>Summary</h2>
<p>The timing results mainly speak for themselves: clearly an intrusive pointer
outperforms all others and a simple heap based counted pointer has poor performance
relative to other implementations. The selection of an optimal non-intrusive
smart pointer implementation is more application dependent, however. Where small
numbers of copies are expected, it is likely that the linked implementation
will be favoured. Conversely, for larger numbers of copies a counted pointer
with some type of special purpose allocator looks like a win. Other factors
to bear in mind are: -</p>
<ul>
<li>Deterministic individual, as opposed to amortized, operation time. This
weighs against any implementation depending on an allocator.</li>
<li>Multithreaded synchronization. This weighs against an implementation which
spreads its information as in the case of linked pointer.</li>
</ul>
<hr>
<p>Revised <!--webbot bot="Timestamp" S-Type="EDITED" S-Format="%d %B %Y" startspan -->19 August 2001<!--webbot bot="Timestamp" endspan i-checksum="14767" -->
</p>
<p> Copyright Gavin Collings 2000. Permission to copy, use, modify, sell
and distribute this document is granted provided this copyright notice appears in all
copies. This document is provided "as is" without express or implied warranty,
and with no claim as to its suitability for any purpose.</p>
</body>
</html>
|
