Case vs. hashset benchmark

benchmark.lpr

program benchmark;

{$mode objfpc}{$H+}

uses
  {$IFDEF UNIX}{$IFDEF UseCThreads}
  cthreads,
  {$ENDIF}{$ENDIF}
  Classes, xquery.internals.common, bbutils, bbdebugtools,htmlInformation, SysUtils
  { you can add units after this };

//{$define K3}
//{$define K6}
//{$define K9}
{$define K15}

{$ifdef K3}{$define K3P}{$endif}
{$ifdef K6}{$define K3P}{$define K6P}{$endif}
{$ifdef K9}{$define K3P}{$define K6P}{$define K9P}{$endif}
{$ifdef K12}{$define K3P}{$define K6P}{$define K9P}{$define K12P}{$endif}
{$ifdef K15}{$define K3P}{$define K6P}{$define K9P}{$define K12P}{$define K15P}{$endif}

const n = 10000000;
      k = {$ifdef K3}3{$endif}{$ifdef K6}6{$endif}{$ifdef K9}9{$endif}{$ifdef K12}12{$endif}{$ifdef K15}15{$endif};
      nomatchfiller = 5;
type HTMLNodeNameHashs = object
  const img = $4FACAFC2;
  const table = $57CFB523;
  const a = $820103F0;
  const thead = $2D298F5C;
  const tbody = $BB316BB0;
  const br = $2CF50F7A;
  const td = $A664EDFC;
  const tr = $B93B93AD;
  const th = $1483CA3C;
  const basefont = $C997A27A;
  const p = $B7656EB4;

  //invisible
  const area = $B61A9737;
  const base = $36BAA821;
  const datalist = $41BB801A;
  const head = $FB1A74A6;
  const link = $21E329D3;
  const meta = $53F6A414;
  const noembed = $35DC71D8;
  const noframes = $8EF9275D;
  const param = $EA036F5E;
  const rp = $065D2F8B;
  const script = $75469FD3;
  const source = $B04BAA1E;
  const style = $244E4D3D;
  const template = $08F14C20;
  const track = $AB8D6A26;
  const title = $FE8D4719;
end;

function lowerContains(const s: string): boolean;
begin
  case lowercase(s) of
  {$ifdef K3P}
  'img': result := true;
  'table': result := true;
  'a': result := true;
  {$endif}
  {$ifdef K6P}
  'thead': result := true;
  'tbody': result := true;
  'br': result := true;
  {$endif}
  {$ifdef K9P}
  'script': result := true;
  'source': result := true;
  'meta': result := true;
  {$endif}
  {$ifdef K12P}
  'basefont': result := true;
  'p': result := true;
  'area': result := true;
  {$endif}
  {$ifdef K15P}
  'tr': result := true;
  'td': result := true;
  'th': result := true;
  {$endif}
  else result := false;
  end;
end;

function caseHashContains(const hash: longword; const s: string): boolean;
begin
  case hash of
  {$ifdef K3P}
  HTMLNodeNameHashs.img: result := striEqual(s, 'img');
  HTMLNodeNameHashs.table: result := striEqual(s, 'table');
  HTMLNodeNameHashs.a: result := striEqual(s, 'a');
  {$endif}
  {$ifdef K6P}
  HTMLNodeNameHashs.thead: result := striEqual(s, 'thead');
  HTMLNodeNameHashs.tbody: result := striEqual(s, 'tbody');
  HTMLNodeNameHashs.br: result := striEqual(s, 'br');
  {$endif}
  {$ifdef K9P}
  HTMLNodeNameHashs.script: result := striEqual(s, 'script');
  HTMLNodeNameHashs.source: result := striEqual(s, 'source');
  HTMLNodeNameHashs.meta: result := striEqual(s, 'meta');
  {$endif}
  {$ifdef K12P}
  HTMLNodeNameHashs.basefont: result := striEqual(s, 'basefont');
  HTMLNodeNameHashs.p: result := striEqual(s, 'p');
  HTMLNodeNameHashs.area: result := striEqual(s, 'area');
  {$endif}
  {$ifdef K15P}
  HTMLNodeNameHashs.tr: result := striEqual(s, 'tr');
  HTMLNodeNameHashs.td: result := striEqual(s, 'td');
  HTMLNodeNameHashs.th: result := striEqual(s, 'th');
  {$endif}
  else result := false;
  end;
end;

function caseFirstLetter(const s: string): boolean;
begin
  if length(s) = 0 then exit(false);
  case s[1] of
  {$if defined(K3P)}
  'i', 'I': result := striEqual(s, 'img');
  't', 'T': result := striEqual(s, 'table')
                       {$ifdef K6P}or striEqual(s, 'thead') or striEqual(s, 'tbody') {$endif}
                       {$ifdef K15P}or striEqual(s, 'tr') or striEqual(s, 'td') or striEqual(s, 'th')  {$endif}
                      ;
  'a', 'A': result := striEqual(s, 'a') {$ifdef K12P}or striEqual(s, 'area'){$endif};
  {$endif}
  {$if defined(K6P)}
  'b', 'B': result := striEqual(s, 'br') {$ifdef K12P}or striEqual(s, 'basefont'){$endif};
  {$endif}
  {$if defined(K9P)}
  's', 'S': result := striEqual(s, 'script') or striEqual(s, 'source');
  'm', 'M': result := striEqual(s, 'meta');
  {$endif}
  {$if defined(K12P)}
  'p', 'P': result := striEqual(s, 'p');
  {$endif}
  else result := false;
  end;
end;

var kinds: array[0..20] of string = ('img', 'table', 'a',
                                     'thead', 'tbody', 'br',
                                     'script', 'source', 'meta',
                                     'basefont', 'p', 'area',
                                     'tr', 'td', 'th',
                                     'base', 'datalist', 'head', 'link', 'noembed', 'noframes');

var test: array of record
hash: longword;
s: string;
end;
  hashset: TXQHashsetStrCaseInsensitiveASCII;
  i, count: Integer;
begin
  writeln(n);
  writeln(k);
  logging := true;
  setlength(test, n);
  for i := 0 to high(test) do begin
    test[i].s := kinds[Random(k + nomatchfiller)];
    if random(10) < 5 then test[i].s := uppercase(test[i].s);
    test[i].hash := nodeNameHash(test[i].s);
  end;

  hashset.init;
  for i := 0 to k - 1 do begin
    hashset.include(kinds[i]);
  end;


  writeln;writeln('case hash');
  startTiming();
  count := 0;
  for i := 0 to high(test) do
    if caseHashContains(nodeNameHash(test[i].s), test[i].s) then inc(count);
  stopTiming();
  writeln(count);

  writeln;writeln('case hash (cached)');
  startTiming();
  count := 0;
  for i := 0 to high(test) do
    if caseHashContains(test[i].hash, test[i].s) then inc(count);
  stopTiming();
  writeln(count);

  writeln;writeln('hashset');
  startTiming();
  count := 0;
  for i := 0 to high(test) do
    if hashset.contains(test[i].s) then inc(count);
  stopTiming();
  writeln(count);

  writeln;writeln('hashset (cached)');
  startTiming();
  count := 0;
  for i := 0 to high(test) do
    if hashset.findEntityWithHash(pointer(test[i].s), length(test[i].s), test[i].hash) <> nil then inc(count);
  stopTiming();
  writeln(count);

  writeln;writeln('case letter');
  startTiming();
  count := 0;
  for i := 0 to high(test) do
    if caseFirstLetter(test[i].s) then inc(count);
  stopTiming();
  writeln(count);

  writeln;writeln('case lowercase');
  startTiming();
  count := 0;
  for i := 0 to high(test) do
    if lowerContains(test[i].s) then inc(count);
  stopTiming();
  writeln(count);

end.

results.txt

=========================================O1=================================


10000000
3

case hash
2020-03-28:22:05:55:266 (1):  started timing of global
2020-03-28:22:05:55:851 (1):  stopped timing of global run-time: 585 ms
3749857

case hash (cached)
2020-03-28:22:05:55:851 (1):  started timing of global
2020-03-28:22:05:56:044 (1):  stopped timing of global run-time: 193 ms
3749857

hashset
2020-03-28:22:05:56:044 (1):  started timing of global
2020-03-28:22:05:56:972 (1):  stopped timing of global run-time: 928 ms
3749857

hashset (cached)
2020-03-28:22:05:56:972 (1):  started timing of global
2020-03-28:22:05:57:385 (1):  stopped timing of global run-time: 413 ms
3749857

case letter
2020-03-28:22:05:57:385 (1):  started timing of global
2020-03-28:22:05:57:601 (1):  stopped timing of global run-time: 216 ms
3749857

case lowercase
2020-03-28:22:05:57:601 (1):  started timing of global
2020-03-28:22:05:58:475 (1):  stopped timing of global run-time: 874 ms
3749857



--------------------------------------------------------------------


10000000
9

case hash
2020-03-28:22:06:21:523 (1):  started timing of global
2020-03-28:22:06:22:131 (1):  stopped timing of global run-time: 608 ms
6426313

case hash (cached)
2020-03-28:22:06:22:131 (1):  started timing of global
2020-03-28:22:06:22:367 (1):  stopped timing of global run-time: 236 ms
6426313

hashset
2020-03-28:22:06:22:367 (1):  started timing of global
2020-03-28:22:06:23:230 (1):  stopped timing of global run-time: 862 ms
6426313

hashset (cached)
2020-03-28:22:06:23:230 (1):  started timing of global
2020-03-28:22:06:23:686 (1):  stopped timing of global run-time: 457 ms
6426313

case letter
2020-03-28:22:06:23:686 (1):  started timing of global
2020-03-28:22:06:24:022 (1):  stopped timing of global run-time: 336 ms
6426313

case lowercase
2020-03-28:22:06:24:022 (1):  started timing of global
2020-03-28:22:06:25:299 (1):  stopped timing of global run-time: 1276 ms
6426313


--------------------------------------------------------------------------


10000000
15

case hash
2020-03-28:22:06:55:562 (1):  started timing of global
2020-03-28:22:06:56:207 (1):  stopped timing of global run-time: 644 ms
7499955

case hash (cached)
2020-03-28:22:06:56:207 (1):  started timing of global
2020-03-28:22:06:56:488 (1):  stopped timing of global run-time: 281 ms
7499955

hashset
2020-03-28:22:06:56:488 (1):  started timing of global
2020-03-28:22:06:57:431 (1):  stopped timing of global run-time: 943 ms
7499955

hashset (cached)
2020-03-28:22:06:57:431 (1):  started timing of global
2020-03-28:22:06:57:997 (1):  stopped timing of global run-time: 567 ms
7499955

case letter
2020-03-28:22:06:57:997 (1):  started timing of global
2020-03-28:22:06:58:424 (1):  stopped timing of global run-time: 427 ms
7499955

case lowercase
2020-03-28:22:06:58:424 (1):  started timing of global
2020-03-28:22:06:59:978 (1):  stopped timing of global run-time: 1553 ms
7499955


=========================================O2=================================


10000000
3

case hash
2020-03-28:22:08:10:222 (1):  started timing of global
2020-03-28:22:08:10:519 (1):  stopped timing of global run-time: 296 ms
3749857

case hash (cached)
2020-03-28:22:08:10:519 (1):  started timing of global
2020-03-28:22:08:10:685 (1):  stopped timing of global run-time: 166 ms
3749857

hashset
2020-03-28:22:08:10:685 (1):  started timing of global
2020-03-28:22:08:11:120 (1):  stopped timing of global run-time: 435 ms
3749857

hashset (cached)
2020-03-28:22:08:11:120 (1):  started timing of global
2020-03-28:22:08:11:433 (1):  stopped timing of global run-time: 313 ms
3749857

case letter
2020-03-28:22:08:11:433 (1):  started timing of global
2020-03-28:22:08:11:640 (1):  stopped timing of global run-time: 207 ms
3749857

case lowercase
2020-03-28:22:08:11:640 (1):  started timing of global
2020-03-28:22:08:12:539 (1):  stopped timing of global run-time: 899 ms
3749857

-------------------------------------------------------------------------

10000000
9

case hash
2020-03-28:22:08:40:747 (1):  started timing of global
2020-03-28:22:08:41:200 (1):  stopped timing of global run-time: 453 ms
6426313

case hash (cached)
2020-03-28:22:08:41:200 (1):  started timing of global
2020-03-28:22:08:41:454 (1):  stopped timing of global run-time: 254 ms
6426313

hashset
2020-03-28:22:08:41:454 (1):  started timing of global
2020-03-28:22:08:41:964 (1):  stopped timing of global run-time: 511 ms
6426313

hashset (cached)
2020-03-28:22:08:41:964 (1):  started timing of global
2020-03-28:22:08:42:347 (1):  stopped timing of global run-time: 382 ms
6426313

case letter
2020-03-28:22:08:42:347 (1):  started timing of global
2020-03-28:22:08:42:680 (1):  stopped timing of global run-time: 334 ms
6426313

case lowercase
2020-03-28:22:08:42:680 (1):  started timing of global
2020-03-28:22:08:44:027 (1):  stopped timing of global run-time: 1346 ms
6426313

-----------------------------------------------------------------------

10000000
15

case hash
2020-03-28:22:09:06:896 (1):  started timing of global
2020-03-28:22:09:07:311 (1):  stopped timing of global run-time: 416 ms
7499955

case hash (cached)
2020-03-28:22:09:07:312 (1):  started timing of global
2020-03-28:22:09:07:581 (1):  stopped timing of global run-time: 269 ms
7499955

hashset
2020-03-28:22:09:07:581 (1):  started timing of global
2020-03-28:22:09:08:113 (1):  stopped timing of global run-time: 533 ms
7499955

hashset (cached)
2020-03-28:22:09:08:113 (1):  started timing of global
2020-03-28:22:09:08:509 (1):  stopped timing of global run-time: 395 ms
7499955

case letter
2020-03-28:22:09:08:509 (1):  started timing of global
2020-03-28:22:09:08:889 (1):  stopped timing of global run-time: 380 ms
7499955

case lowercase
2020-03-28:22:09:08:889 (1):  started timing of global
2020-03-28:22:09:10:530 (1):  stopped timing of global run-time: 1642 ms
7499955



=========================================O3=================================

10000000
3

case hash
2020-03-28:22:10:24:670 (1):  started timing of global
2020-03-28:22:10:24:972 (1):  stopped timing of global run-time: 303 ms
3749857

case hash (cached)
2020-03-28:22:10:24:972 (1):  started timing of global
2020-03-28:22:10:25:197 (1):  stopped timing of global run-time: 224 ms
3749857

hashset
2020-03-28:22:10:25:197 (1):  started timing of global
2020-03-28:22:10:25:644 (1):  stopped timing of global run-time: 447 ms
3749857

hashset (cached)
2020-03-28:22:10:25:644 (1):  started timing of global
2020-03-28:22:10:25:952 (1):  stopped timing of global run-time: 309 ms
3749857

case letter
2020-03-28:22:10:25:952 (1):  started timing of global
2020-03-28:22:10:26:162 (1):  stopped timing of global run-time: 210 ms
3749857

case lowercase
2020-03-28:22:10:26:162 (1):  started timing of global
2020-03-28:22:10:27:080 (1):  stopped timing of global run-time: 917 ms
3749857


---------------------------------------------------------------

10000000
9

case hash
2020-03-28:22:10:04:294 (1):  started timing of global
2020-03-28:22:10:04:646 (1):  stopped timing of global run-time: 351 ms
6426313

case hash (cached)
2020-03-28:22:10:04:646 (1):  started timing of global
2020-03-28:22:10:04:859 (1):  stopped timing of global run-time: 214 ms
6426313

hashset
2020-03-28:22:10:04:859 (1):  started timing of global
2020-03-28:22:10:05:367 (1):  stopped timing of global run-time: 508 ms
6426313

hashset (cached)
2020-03-28:22:10:05:367 (1):  started timing of global
2020-03-28:22:10:05:745 (1):  stopped timing of global run-time: 377 ms
6426313

case letter
2020-03-28:22:10:05:745 (1):  started timing of global
2020-03-28:22:10:06:075 (1):  stopped timing of global run-time: 331 ms
6426313

case lowercase
2020-03-28:22:10:06:075 (1):  started timing of global
2020-03-28:22:10:07:355 (1):  stopped timing of global run-time: 1279 ms
6426313


--------------------------------------------------------------------------

10000000
15

case hash
2020-03-28:22:09:29:971 (1):  started timing of global
2020-03-28:22:09:30:347 (1):  stopped timing of global run-time: 376 ms
7499955

case hash (cached)
2020-03-28:22:09:30:347 (1):  started timing of global
2020-03-28:22:09:30:611 (1):  stopped timing of global run-time: 265 ms
7499955

hashset
2020-03-28:22:09:30:612 (1):  started timing of global
2020-03-28:22:09:31:139 (1):  stopped timing of global run-time: 527 ms
7499955

hashset (cached)
2020-03-28:22:09:31:140 (1):  started timing of global
2020-03-28:22:09:31:534 (1):  stopped timing of global run-time: 394 ms
7499955

case letter
2020-03-28:22:09:31:534 (1):  started timing of global
2020-03-28:22:09:31:896 (1):  stopped timing of global run-time: 363 ms
7499955

case lowercase
2020-03-28:22:09:31:896 (1):  started timing of global
2020-03-28:22:09:33:589 (1):  stopped timing of global run-time: 1692 ms
7499955



=========================================O4=================================

10000000
3

case hash
2020-03-28:22:10:47:365 (1):  started timing of global
2020-03-28:22:10:47:661 (1):  stopped timing of global run-time: 295 ms
3749857

case hash (cached)
2020-03-28:22:10:47:661 (1):  started timing of global
2020-03-28:22:10:47:830 (1):  stopped timing of global run-time: 170 ms
3749857

hashset
2020-03-28:22:10:47:830 (1):  started timing of global
2020-03-28:22:10:48:272 (1):  stopped timing of global run-time: 442 ms
3749857

hashset (cached)
2020-03-28:22:10:48:272 (1):  started timing of global
2020-03-28:22:10:48:595 (1):  stopped timing of global run-time: 323 ms
3749857

case letter
2020-03-28:22:10:48:595 (1):  started timing of global
2020-03-28:22:10:48:801 (1):  stopped timing of global run-time: 205 ms
3749857

case lowercase
2020-03-28:22:10:48:801 (1):  started timing of global
2020-03-28:22:10:49:729 (1):  stopped timing of global run-time: 928 ms
3749857


-------------------------------------------------------------------

10000000
9

case hash
2020-03-28:22:11:08:496 (1):  started timing of global
2020-03-28:22:11:08:853 (1):  stopped timing of global run-time: 357 ms
6426313

case hash (cached)
2020-03-28:22:11:08:853 (1):  started timing of global
2020-03-28:22:11:09:066 (1):  stopped timing of global run-time: 214 ms
6426313

hashset
2020-03-28:22:11:09:066 (1):  started timing of global
2020-03-28:22:11:09:564 (1):  stopped timing of global run-time: 498 ms
6426313

hashset (cached)
2020-03-28:22:11:09:564 (1):  started timing of global
2020-03-28:22:11:09:946 (1):  stopped timing of global run-time: 382 ms
6426313

case letter
2020-03-28:22:11:09:946 (1):  started timing of global
2020-03-28:22:11:10:287 (1):  stopped timing of global run-time: 340 ms
6426313

case lowercase
2020-03-28:22:11:10:287 (1):  started timing of global
2020-03-28:22:11:11:572 (1):  stopped timing of global run-time: 1286 ms
6426313


-----------------------------------------------------------------------


10000000
15

case hash
2020-03-28:22:11:28:465 (1):  started timing of global
2020-03-28:22:11:28:843 (1):  stopped timing of global run-time: 379 ms
7499955

case hash (cached)
2020-03-28:22:11:28:843 (1):  started timing of global
2020-03-28:22:11:29:107 (1):  stopped timing of global run-time: 263 ms
7499955

hashset
2020-03-28:22:11:29:107 (1):  started timing of global
2020-03-28:22:11:29:633 (1):  stopped timing of global run-time: 526 ms
7499955

hashset (cached)
2020-03-28:22:11:29:633 (1):  started timing of global
2020-03-28:22:11:30:026 (1):  stopped timing of global run-time: 394 ms
7499955

case letter
2020-03-28:22:11:30:026 (1):  started timing of global
2020-03-28:22:11:30:386 (1):  stopped timing of global run-time: 359 ms
7499955

case lowercase
2020-03-28:22:11:30:386 (1):  started timing of global
2020-03-28:22:11:31:953 (1):  stopped timing of global run-time: 1568 ms
7499955