Du bist nicht angemeldet.

Stilllegung des Forums
Das Forum wurde am 05.06.2023 nach über 20 Jahren stillgelegt (weitere Informationen und ein kleiner Rückblick).
Registrierungen, Anmeldungen und Postings sind nicht mehr möglich. Öffentliche Inhalte sind weiterhin zugänglich.
Das Team von spieleprogrammierer.de bedankt sich bei der Community für die vielen schönen Jahre.
Wenn du eine deutschsprachige Spieleentwickler-Community suchst, schau doch mal im Discord und auf ZFX vorbei!

Werbeanzeige

Anonymous

unregistriert

31

21.02.2006, 22:04

FAQ gefährdet

CW_Kovok

Alter Hase

  • »CW_Kovok« ist der Autor dieses Themas

Beiträge: 836

Wohnort: nähe Bonn

Beruf: Schüler

  • Private Nachricht senden

32

25.02.2006, 10:57

Also jetzt ist das so:

C-/C++-Quelltext

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
#include <iostream>
#include <ctime>
#include <cmath>

using namespace std;

const float pi  = atan(1.0);

inline float FastSin0(const float& fAngle)
{
    float fASqr = fAngle*fAngle;
    float fResult = 7.61e-03f;
    
    fResult *= fASqr;
    fResult -= 1.6605e-01f;
    fResult *= fASqr;
    fResult += 1.0f;
    fResult *= fAngle;
    return fResult;
}

inline float FastSin1(const float& fAngle)
{
    float fASqr = fAngle*fAngle;
    float fResult = -2.39e-08;

    fResult *= fASqr;
    fResult += 2.7526e-06f;
    fResult *= fASqr;
    fResult -= 1.98409e-04f;
    fResult *= fASqr;
    fResult += 8.3333315e-03f;
    fResult *= fASqr;
    fResult -= 1.666666664e-01f;
    fResult *= fASqr;
    fResult += 1.0f;
    fResult *= fAngle;

    return fResult;
}

inline float TnSin(float x)
{
    float f2 = x * x;
    float f3 = x * f2;
    float f5 = f3 * f2;
    float f7 = f5 * f2;
    float f9 = f7 * f2;
    x -= 1.0f / 6.0f * f3;
    x += 1.0f / 120.0f * f5;
    x -= 1.0f / 5040.0f * f7;
    x += 1.0f / 362880.0f * f9;
    return x;
}

int main(void)
{
    long start, ende;
    float a1, a2, a3, a4, a5, t = 0.0f;

    start = clock();
    
    for(int j=0; j<1e3; j++)
        for(float i=0.0f; i<pi; i=i+0.0001f)
            t += FastSin0(i);

    ende = clock();

    a1 = float(ende-start);

    cout<<t<<endl;
    cout<<"FastSin0 fertig"<<endl;

    t = 0.0f;
    start = clock();
    
    for(int j=0; j<1e3; j++)
        for(float i=0.0f; i<pi; i=i+0.0001f)
            t += FastSin1(i);

    ende = clock();

    a2 = float(ende-start);

    cout<<t<<endl;
    cout<<"FastSin1 fertig"<<endl;

    t = 0.0f;
    start = clock();
    
    for(int j=0; j<1e3; j++)
        for(float i=0.0f; i<pi; i=i+0.0001f)
            t += sinf(i);

    ende = clock();

    a3 = float(ende-start);

    cout<<t<<endl;
    cout<<"sinf fertig"<<endl;

    t = 0.0f;
    start = clock();
    
    for(int j=0; j<1e3; j++)
        for(float i=0.0f; i<pi; i=i+0.0001f)
            t += sin(i);

    ende = clock();

    a4 = float(ende-start);

    cout<<t<<endl;
    cout<<"sin fertig"<<endl;

    t = 0.0f;
    start = clock();
    
    for(int j=0; j<1e3; j++)
        for(float i=0.0f; i<pi; i=i+0.0001f)
            t += TnSin(i);

    ende = clock();

    a5 = float(ende-start);

    cout<<t<<endl;
    cout<<"TnSin fertig"<<endl;

    cout<<"sinf: "<<a3/CLOCKS_PER_SEC<<" entspricht: "<<a3/a3*100<<"%"<<endl;
    cout<<"sin: "<<a4/CLOCKS_PER_SEC<<" entspricht: "<<a3/a4*100<<"%"<<endl;
    cout<<"FastSin0: "<<a1/CLOCKS_PER_SEC<<" entspricht: "<<a3/a1*100<<"%"<<endl;
    cout<<"FastSin1: "<<a2/CLOCKS_PER_SEC<<" entspricht: "<<a3/a2*100<<"%"<<endl;
    cout<<"TnSin: "<<a5/CLOCKS_PER_SEC<<" entspricht: "<<a3/a5*100<<"%"<<endl;

    float p = 0.0f;

    for(float i=0.0f; i<pi; i+= 0.0001f)
    {
        float j = sin(double(i)) - TnSin(i);
        if(p<j)
            p = j;
    }

    cout<<scientific<<"Genauigkeit TnSin: "<<p<<endl;
   
    p = 0.0f;

    for(float i=0.0f; i<pi; i+= 0.0001f)
    {
        float j = sin(double(i)) - FastSin0(i);
        if(p<j)
            p = j;
    }

    cout<<"Genauigkeit FastSin0: "<<p<<endl;

    p = 0.0f;

    for(float i=0.0f; i<pi; i+= 0.0001f)
    {
        float j = sin(double(i)) - FastSin1(i);
        if(p<j)
            p = j;
    }

    cout<<"Genauigkeit FastSin1: "<<p<<endl;

    p = 0.0f;

    for(float i=0.0f; i<pi; i+= 0.0001f)
    {
        float j = sin(double(i)) - sinf(i);
        if(p<j)
            p = j;
    }

    cout<<"Genauigkeit sinf: "<<p<<endl;

    return 0;
}
Dann erhalte ich als ergebnis:

Zitat

2.92468e+006
FastSin0 fertig
2.92422e+006
FastSin1 fertig
2.92422e+006
sinf fertig
2.92422e+006
sin fertig
2.92422e+006
TnSin fertig
sinf: 0.812 entspricht: 100%
sin: 1.063 entspricht: 76.3876%
FastSin0: 0.656 entspricht: 123.78%
FastSin1: 1 entspricht: 81.2%
TnSin: 0.672 entspricht: 120.833%
Genauigkeit TnSin: 1.047701e-007
Genauigkeit FastSin0: 2.530216e-010
Genauigkeit FastSin1: 5.518797e-008
Genauigkeit sinf: 2.979680e-008
Drücken Sie eine beliebige Taste . . .
Immer noch keine 4 bis achtfache geschwindigkeit, wieso?
Was es alles gibt, das ich nich brauche - Aristoteles

David Scherfgen

Administrator

Beiträge: 10 382

Wohnort: Hildesheim

Beruf: Wissenschaftlicher Mitarbeiter

  • Private Nachricht senden

33

25.02.2006, 13:24

Und das hier erhalte ich:

Quellcode

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
2.92893e+06
FastSin0 fertig
2.92859e+06
FastSin1 fertig
2.92859e+06
sinf fertig
2.92859e+06
sin fertig
2.92859e+06
TnSin fertig
sinf: 0.468 entspricht: 100%
sin: 0.469 entspricht: 99.7868%
FastSin0: 0.046 entspricht: 1017.39%
FastSin1: 0.079 entspricht: 592.405%
TnSin: 0.109 entspricht: 429.358%
Genauigkeit TnSin: 1.142471e-09
Genauigkeit FastSin0: 0.000000e+00
Genauigkeit FastSin1: 2.433674e-09
Genauigkeit sinf: 0.000000e+00


Seltsam ist die Genauigkeit von FastSin0!
Ich habe übrigens einen Athlon64 3000+.

CW_Kovok

Alter Hase

  • »CW_Kovok« ist der Autor dieses Themas

Beiträge: 836

Wohnort: nähe Bonn

Beruf: Schüler

  • Private Nachricht senden

34

25.02.2006, 13:32

habe ich auch, welchen compiler benutzt du? Hier das ganze mal inm Releasemode

Zitat

2.92468e+006
FastSin0 fertig
2.92422e+006
FastSin1 fertig
2.92422e+006
sinf fertig
2.92422e+006
sin fertig
2.92422e+006
TnSin fertig
sinf: 0.438 entspricht: 100%
sin: 0.453 entspricht: 96.6887%
FastSin0: 0.219 entspricht: 200%
FastSin1: 0.516 entspricht: 84.8837%
TnSin: 0.328 entspricht: 133.537%
Genauigkeit TnSin: 1.047701e-007
Genauigkeit FastSin0: 2.530216e-010
Genauigkeit FastSin1: 5.518797e-008
Genauigkeit sinf: 2.979680e-008
Drücken Sie eine beliebige Taste . . .
Was es alles gibt, das ich nich brauche - Aristoteles

David Scherfgen

Administrator

Beiträge: 10 382

Wohnort: Hildesheim

Beruf: Wissenschaftlicher Mitarbeiter

  • Private Nachricht senden

35

25.02.2006, 13:45

Visual C++.NET 2003, Release-Mode, nur Standard-Optimierungen

CW_Kovok

Alter Hase

  • »CW_Kovok« ist der Autor dieses Themas

Beiträge: 836

Wohnort: nähe Bonn

Beruf: Schüler

  • Private Nachricht senden

36

25.02.2006, 13:48

komisch, ich benutze visual express 2005, der müsste doch ähnlich gute ergebnisse erzielen oder?

//edit:
fmodf ist ein speedkiller oder?
Was es alles gibt, das ich nich brauche - Aristoteles

CW_Kovok

Alter Hase

  • »CW_Kovok« ist der Autor dieses Themas

Beiträge: 836

Wohnort: nähe Bonn

Beruf: Schüler

  • Private Nachricht senden

37

25.02.2006, 20:44

habe in den optionen unter project->propertys->c++->codegeneration die option floating point model auf fast gestellt mit dem ergebnis:

Zitat

2.92893e+006
FastSin0 fertig
2.92859e+006
FastSin1 fertig
2.92859e+006
sinf fertig
2.92859e+006
sin fertig
2.92859e+006
TnSin fertig
sinf: 0.468 entspricht: 100%
sin: 0.453 entspricht: 103.311%
FastSin0: 0.046 entspricht: 1017.39%
FastSin1: 0.079 entspricht: 592.405%
TnSin: 0.094 entspricht: 497.872%
Genauigkeit TnSin: 1.142471e-009
Genauigkeit FastSin0: 0.000000e+000
Genauigkeit FastSin1: 2.433674e-009
Genauigkeit sinf: 0.000000e+000
Drücken Sie eine beliebige Taste . . .

also deinen sehr ähnliche ergebnisse, jetzt also die frage, welche einstellung entscheiden über die geschwindigkeit, aus der einen gibbet da noch viele andere, die alle irgendwie die geschwindigkeit beeinflussen. Wie kriegt man es denn am schnellsten?
Was es alles gibt, das ich nich brauche - Aristoteles

CW_Kovok

Alter Hase

  • »CW_Kovok« ist der Autor dieses Themas

Beiträge: 836

Wohnort: nähe Bonn

Beruf: Schüler

  • Private Nachricht senden

38

26.02.2006, 12:22

So jetzt mal die volle Version, mit allen optimierungen, auch der oben schon genannten fast einstellung unter code generation(nur ohne assembler):

trigfunc.h:

C-/C++-Quelltext

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
#include <cmath>

const float CE_PI              = (float)(4.0*atan(1.0));
const float CE_HALF_PI         = (float)(2.0*atan(1.0));
const float CE_QUATER_PI       = (float)(atan(1.0));
const float CE_TWO_PI          = (float)(8.0*atan(1.0));

const double CE_DPI            = 4.0*atan(1.0);
const double CE_DHALF_PI       = 2.0*atan(1.0);
const double CE_DQUATER_PI     = atan(1.0);
const double CE_DTWO_PI        = 8.0*atan(1.0);

//some very fast sin and asin functions

template<typename T> inline T FastSin0(const T& tAngle)
{
    T tASqr = tAngle*tAngle;
    T tResult = (T)7.61e-03;
    
    tResult *= tASqr;
    tResult -= (T)1.6605e-01;
    tResult *= tASqr;
    tResult += (T)1.0;
    tResult *= tAngle;

    return tResult;
}

template<typename T> inline T FastSin1(const T& tAngle)
{
    T tASqr = tAngle*tAngle;
    T tResult = (T)-2.39e-08;

    tResult *= tASqr;
    tResult += (T)2.7526e-06f;
    tResult *= tASqr;
    tResult -= (T)1.98409e-04f;
    tResult *= tASqr;
    tResult += (T)8.3333315e-03f;
    tResult *= tASqr;
    tResult -= (T)1.666666664e-01f;
    tResult *= tASqr;
    tResult += (T)1.0f;
    tResult *= tAngle;

    return tResult;
}

template<typename T> inline T FastSinTn(const T& tAngle)
{
    T tResult = tAngle;
    
    T t2 = tAngle * tAngle;
    T t3 = tAngle * t2;
    T t5 = t3 * t2;
    T t7 = t5 * t2;
    T t9 = t7 * t2;
    
    tResult -= 1.0f / 6.0f * t3;
    tResult += 1.0f / 120.0f * t5;
    tResult -= 1.0f / 5040.0f * t7;
    tResult += 1.0f / 362880.0f * t9;
    
    return tResult; 
}

template<typename T> inline T FastInvSin(const T& tValue)
{
    T tRoot = sqrt(((double)1.0)-tValue);
    T tResult = -(T)0.0187293;

    tResult *= tValue;
    tResult += (T)0.0742610;
    tResult *= tValue;
    tResult -= (T)0.2121144;
    tResult *= tValue;
    tResult += (T)1.5707288;
    tResult = CE_HALF_PI - tRoot*tResult;

    return tResult;
}

//some very fast cos and acos functions

template<typename T> inline T FastCos0(const T& tAngle)
{
    if(tAngle == CE_HALF_PI)
        return 0;

    T tASqr = tAngle*tAngle;
    T tResult = (T)3.705e-02;
    
    tResult *= tASqr;
    tResult -= (T)4.967e-01;
    tResult *= tASqr;
    tResult += (T)1.0;
    
    return tResult;
}

template<typename T> inline T FastCos1(const T& tAngle)
{
    if(tAngle == CE_HALF_PI)
        return 0;

    T tASqr = tAngle*tAngle;
    T tResult = -(T)2.605e-07;

    tResult *= tASqr;
    tResult += (T)2.47609e-05;
    tResult *= tASqr;
    tResult -= (T)1.3888397e-03;
    tResult *= tASqr;
    tResult += (T)4.16666418e-02;
    tResult *= tASqr;
    tResult -= (T)4.999999963e-01;
    tResult *= tASqr;
    tResult += (T)1.0;

    return tResult;
}

template<typename T> inline T FastInvCos(const T& tValue)
{
    T tRoot = sqrt(((double)1.0)-tValue);
    T tResult = -(T)0.0187293;

    tResult *= tValue;
    tResult += (T)0.0742610;
    tResult *= tValue;
    tResult -= (T)0.2121144;
    tResult *= tValue;
    tResult += (T)1.5707288;
    tResult *= tRoot;

    return tResult;
}

und die Datei main.cpp:

C-/C++-Quelltext

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
#include <iostream>
#include <ctime>

#include "trigfunc.h"

using namespace std;

void TestFastSinFloat(long lTestCount, long lBase)
{
    float fSin0, fSin1, fSinTn, fSinf, fSin, fSin0Accuracy, fSin1Accuracy, fSinTnAccuracy, fTemp=0.0f;
    long lStart, lEnd;

    cout<<"fast Sin<float> (between 0 and pi/2):"<<endl;
    cout<<"-------------------------------------"<<endl;
    cout<<endl;
    
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(float i=0; i<CE_HALF_PI; i=i+1.0f/lBase)
        {
            fTemp += FastSin0<float>(i);
        }

    lEnd   = clock();
    fSin0  = float(lEnd-lStart);

    cout<<scientific<<"Sin with FastSin0(sum): "<<fTemp<<" (against unwanted optimizations)"<<endl;

    fTemp  = 0.0f; 
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(float i=0; i<CE_HALF_PI; i=i+1.0f/lBase)
        {
            fTemp += FastSin1<float>(i);
        }

    lEnd   = clock();
    fSin1  = float(lEnd-lStart);

    cout<<"Sin with FastSin1: "<<fTemp<<" (against unwanted optimizations)"<<endl;

    fTemp  = 0.0f; 
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(float i=0; i<CE_HALF_PI; i=i+1.0f/lBase)
        {
            fTemp += FastSinTn<float>(i);
        }

    lEnd   = clock();
    fSinTn = float(lEnd-lStart);

    cout<<"Sin with FastSinTn: "<<fTemp<<" (against unwanted optimizations)"<<endl;

    fTemp  = 0.0f; 
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(float i=0; i<CE_HALF_PI; i=i+1.0f/lBase)
        {
            fTemp += sinf(i);
        }

    lEnd   = clock();
    fSinf  = float(lEnd-lStart);

    cout<<"Sin with sinf: "<<fTemp<<" (against unwanted optimizations)"<<endl;

    fTemp  = 0.0f; 
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(float i=0; i<CE_HALF_PI; i=i+1.0f/lBase)
        {
            fTemp += sin(i);
        }

    lEnd   = clock();
    fSin  = float(lEnd-lStart);

    cout<<"Sin with sin: "<<fTemp<<" (against unwanted optimizations)"<<endl;
 
    cout<<fixed<<endl;
    cout<<"FastSin0: "<<fSin0/CLOCKS_PER_SEC<<" sec or "<<(1/fSin0)*lBase<<" per second and Achievement: "<<(fSinf/fSin0)*100-100<<"%"<<endl;
    cout<<"FastSin1: "<<fSin1/CLOCKS_PER_SEC<<" sec or "<<(1/fSin1)*lBase<<" per second and Achievement: "<<(fSinf/fSin1)*100-100<<"%"<<endl;
    cout<<"FastSinTn: "<<fSinTn/CLOCKS_PER_SEC<<" sec or "<<(1/fSinTn)*lBase<<" per second and Achievement: "<<(fSinf/fSinTn)*100-100<<"%"<<endl;
    cout<<"sinf: "<<fSinf/CLOCKS_PER_SEC<<" sec or "<<(1/fSinf)*lBase<<" per second and Achievement: "<<(fSinf/fSinf)*100-100<<"%"<<endl;
    cout<<"sin: "<<fSin/CLOCKS_PER_SEC<<" sec or "<<(1/fSin)*lBase<<" per second and Achievement: "<<(fSinf/fSin)*100-100<<"%"<<endl;
    cout<<endl;

    fSin0Accuracy = 0.0f;

    for(float i=1; i<CE_HALF_PI; i=i+1.0f/lBase)
    {
        fTemp = sin(i)-FastSin0<float>(i);

        if(fTemp<0) fTemp = -fTemp;

        if(fSin0Accuracy<fTemp)
            fSin0Accuracy = fTemp;
    }

    fSin1Accuracy = 0.0f;

    for(float i=1; i<CE_HALF_PI; i=i+1.0f/lBase)
    {
        fTemp = sin(i)-FastSin1<float>(i);

        if(fTemp<0) fTemp = -fTemp;

        if(fSin1Accuracy<fTemp)
            fSin1Accuracy = fTemp;
    }

    fSinTnAccuracy = 0.0f;

    for(float i=1; i<CE_HALF_PI; i=i+1.0f/lBase)
    {
        fTemp = sin(i)-FastSinTn<float>(i);

        if(fTemp<0) fTemp = -fTemp;

        if(fSinTnAccuracy<fTemp)
            fSinTnAccuracy = fTemp;
    }

    cout<<scientific<<"Accuracy FastSin0: "<<fSin0Accuracy<<endl;
    cout<<scientific<<"Accuracy FastSin1: "<<fSin1Accuracy<<endl;
    cout<<scientific<<"Accuracy FastSinTn: "<<fSinTnAccuracy<<endl;
    cout<<fixed<<endl;
}

void TestFastSinDouble(long lTestCount, long lBase)
{
    double dSin0, dSin1, dSinTn, dSin, dSin0Accuracy, dSin1Accuracy, dSinTnAccuracy, dTemp=0.0;
    long lStart, lEnd;

    cout<<"fast Sin<double> (between 0 and pi/2): "<<endl;
    cout<<"--------------------------------------"<<endl;
    cout<<endl;
    
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(double i=0; i<CE_DHALF_PI; i=i+1.0/lBase)
        {
            dTemp += FastSin0<double>(i);
        }

    lEnd   = clock();
    dSin0  = double(lEnd-lStart);

    cout<<scientific<<"Sin with FastSin0(sum): "<<dTemp<<" (against unwanted optimizations)"<<endl;

    dTemp  = 0.0; 
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(double i=0; i<CE_DHALF_PI; i=i+1.0/lBase)
        {
            dTemp += FastSin1<double>(i);
        }

    lEnd   = clock();
    dSin1  = double(lEnd-lStart);

    cout<<"Sin with FastSin1: "<<dTemp<<" (against unwanted optimizations)"<<endl;

    dTemp  = 0.0; 
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(double i=0; i<CE_DHALF_PI; i=i+1.0/lBase)
        {
            dTemp += FastSinTn<double>(i);
        }

    lEnd   = clock();
    dSinTn = double(lEnd-lStart);

    cout<<"Sin with FastSinTn: "<<dTemp<<" (against unwanted optimizations)"<<endl;

    dTemp  = 0.0; 
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(double i=0; i<CE_DHALF_PI; i=i+1.0f/lBase)
        {
            dTemp += sin(i);
        }

    lEnd   = clock();
    dSin  = double(lEnd-lStart);

    cout<<"Sin with sin: "<<dTemp<<" (against unwanted optimizations)"<<endl;
  
    cout<<fixed<<endl;
    cout<<"FastSin0: "<<dSin0/CLOCKS_PER_SEC<<" sec or "<<(1/dSin0)*lBase<<" per sec and achieves: "<<(dSin/dSin0)*100-100<<"%"<<endl;
    cout<<"FastSin1: "<<dSin1/CLOCKS_PER_SEC<<" sec or "<<(1/dSin1)*lBase<<" per sec and achieves: "<<(dSin/dSin1)*100-100<<"%"<<endl;
    cout<<"FastSinTn: "<<dSinTn/CLOCKS_PER_SEC<<" sec or "<<(1/dSinTn)*lBase<<" per sec and achieves: "<<(dSin/dSinTn)*100-100<<"%"<<endl;
    cout<<"sin: "<<dSin/CLOCKS_PER_SEC<<" sec or "<<(1/dSin)*lBase<<" per sec and achieves: "<<(dSin/dSin)*100-100<<"%"<<endl;
    cout<<endl;

    dSin0Accuracy = 0.0f;

    for(double i=1; i<CE_DHALF_PI; i=i+1.0f/lBase)
    {
        dTemp = sin(i)-FastSin0<double>(i);

        if(dTemp<0) dTemp = -dTemp;

        if(dSin0Accuracy<dTemp)
            dSin0Accuracy = dTemp;
    }

    dSin1Accuracy = 0.0f;

    for(double i=1; i<CE_DHALF_PI; i=i+1.0f/lBase)
    {
        dTemp = sin(i)-FastSin1<double>(i);

        if(dTemp<0) dTemp = -dTemp;

        if(dSin1Accuracy<dTemp)
            dSin1Accuracy = dTemp;
    }

    dSinTnAccuracy = 0.0f;

    for(double i=1; i<CE_DHALF_PI; i=i+1.0f/lBase)
    {
        dTemp = sin(i)-FastSinTn<double>(i);

        if(dTemp<0) dTemp = -dTemp;

        if(dSinTnAccuracy<dTemp)
            dSinTnAccuracy = dTemp;
    }

    cout<<scientific<<"Accuracy FastSin0: "<<dSin0Accuracy<<endl;
    cout<<scientific<<"Accuracy FastSin1: "<<dSin1Accuracy<<endl;
    cout<<scientific<<"Accuracy FastSinTn: "<<dSinTnAccuracy<<endl;
    cout<<fixed<<endl;
}

void TestFastCosFloat(long lTestCount, long lBase)
{
    float fCos0, fCos1, fCosf, fCos, fCos0Accuracy, fCos1Accuracy, fTemp=0.0f;
    long lStart, lEnd;

    cout<<"fast Cos<float> (between 0 and pi/2): "<<endl;
    cout<<"-------------------------------------"<<endl;
    cout<<endl;
    
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(float i=0; i<CE_HALF_PI; i=i+1.0f/lBase)
        {
            fTemp += FastCos0<float>(i);
        }

    lEnd   = clock();
    fCos0  = float(lEnd-lStart);

    cout<<scientific<<"Cos with FastCos0(sum): "<<fTemp<<" (against unwanted optimizations)"<<endl;

    fTemp  = 0.0f; 
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(float i=0; i<CE_HALF_PI; i=i+1.0f/lBase)
        {
            fTemp += FastCos1<float>(i);
        }

    lEnd   = clock();
    fCos1  = float(lEnd-lStart);

    cout<<"Cos with FastCos1: "<<fTemp<<" (against unwanted optimizations)"<<endl;

    fTemp  = 0.0f; 
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(float i=0; i<CE_HALF_PI; i=i+1.0f/lBase)
        {
            fTemp += cosf(i);
        }

    lEnd   = clock();
    fCosf  = float(lEnd-lStart);

    cout<<"Cos with cosf: "<<fTemp<<" (against unwanted optimizations)"<<endl;

    fTemp  = 0.0f; 
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(float i=0; i<CE_HALF_PI; i=i+1.0f/lBase)
        {
            fTemp += cos(i);
        }

    lEnd   = clock();
    fCos  = float(lEnd-lStart);

    cout<<"Cos with cos: "<<fTemp<<" (against unwanted optimizations)"<<endl;
 
    cout<<fixed<<endl;
    cout<<"FastCos0: "<<fCos0/CLOCKS_PER_SEC<<" sec or "<<(1/fCos0)*lBase<<" per second and Achievement: "<<(fCosf/fCos0)*100-100<<"%"<<endl;
    cout<<"FastCos1: "<<fCos1/CLOCKS_PER_SEC<<" sec or "<<(1/fCos1)*lBase<<" per second and Achievement: "<<(fCosf/fCos1)*100-100<<"%"<<endl;
    cout<<"cosf: "<<fCosf/CLOCKS_PER_SEC<<" sec or "<<(1/fCosf)*lBase<<" per second and Achievement: "<<(fCosf/fCosf)*100-100<<"%"<<endl;
    cout<<"cos: "<<fCos/CLOCKS_PER_SEC<<" sec or "<<(1/fCos)*lBase<<" per second and Achievement: "<<(fCosf/fCos)*100-100<<"%"<<endl;
    cout<<endl;

    fCos0Accuracy = 0.0f;

    for(float i=1; i<CE_HALF_PI; i=i+1.0f/lBase)
    {
        fTemp = cos(i)-FastCos0<float>(i);

        if(fTemp<0) fTemp = -fTemp;

        if(fCos0Accuracy<fTemp)
            fCos0Accuracy = fTemp;
    }

    fCos1Accuracy = 0.0f;

    for(float i=1; i<CE_HALF_PI; i=i+1.0f/lBase)
    {
        fTemp = cos(i)-FastCos1<float>(i);

        if(fTemp<0) fTemp = -fTemp;

        if(fCos1Accuracy<fTemp)
            fCos1Accuracy = fTemp;
    }

    cout<<scientific<<"Accuracy FastCos0: "<<fCos0Accuracy<<endl;
    cout<<scientific<<"Accuracy FastCos1: "<<fCos1Accuracy<<endl;
    cout<<fixed<<endl;
}

void TestFastCosDouble(long lTestCount, long lBase)
{
    double dCos0, dCos1, dCos, dCos0Accuracy, dCos1Accuracy, dTemp=0.0;
    long lStart, lEnd;

    cout<<"fast Cos<double> (between 0 and pi/2):"<<endl;
    cout<<"--------------------------------------"<<endl;
    cout<<endl;
    
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(double i=0; i<CE_DHALF_PI; i=i+1.0/lBase)
        {
            dTemp += FastCos0<double>(i);
        }

    lEnd   = clock();
    dCos0  = double(lEnd-lStart);

    cout<<scientific<<"Cos with FastCos0(sum): "<<dTemp<<" (against unwanted optimizations)"<<endl;

    dTemp  = 0.0; 
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(double i=0; i<CE_DHALF_PI; i=i+1.0/lBase)
        {
            dTemp += FastCos1<double>(i);
        }

    lEnd   = clock();
    dCos1  = double(lEnd-lStart);

    cout<<"Cos with FastCos1: "<<dTemp<<" (against unwanted optimizations)"<<endl;

    dTemp  = 0.0; 
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(double i=0; i<CE_DHALF_PI; i=i+1.0f/lBase)
        {
            dTemp += cos(i);
        }

    lEnd   = clock();
    dCos  = double(lEnd-lStart);

    cout<<"Cos with cos: "<<dTemp<<" (against unwanted optimizations)"<<endl;
  
    cout<<fixed<<endl;
    cout<<"FastCos0: "<<dCos0/CLOCKS_PER_SEC<<" sec or "<<(1/dCos0)*lBase<<" per sec and achieves: "<<(dCos/dCos0)*100-100<<"%"<<endl;
    cout<<"FastCos1: "<<dCos1/CLOCKS_PER_SEC<<" sec or "<<(1/dCos1)*lBase<<" per sec and achieves: "<<(dCos/dCos1)*100-100<<"%"<<endl;
    cout<<"cos: "<<dCos/CLOCKS_PER_SEC<<" sec or "<<(1/dCos)*lBase<<" per sec and achieves: "<<(dCos/dCos)*100-100<<"%"<<endl;
    cout<<endl;

    dCos0Accuracy = 0.0f;

    for(double i=1; i<CE_DHALF_PI; i=i+1.0f/lBase)
    {
        dTemp = cos(i)-FastCos0<double>(i);

        if(dTemp<0) dTemp = -dTemp;

        if(dCos0Accuracy<dTemp)
            dCos0Accuracy = dTemp;
    }

    dCos1Accuracy = 0.0f;

    for(double i=1; i<CE_DHALF_PI; i=i+1.0f/lBase)
    {
        dTemp = cos(i)-FastCos1<double>(i);

        if(dTemp<0) dTemp = -dTemp;

        if(dCos1Accuracy<dTemp)
            dCos1Accuracy = dTemp;
    }

    cout<<scientific<<"Accuracy FastCos0: "<<dCos0Accuracy<<endl;
    cout<<scientific<<"Accuracy FastCos1: "<<dCos1Accuracy<<endl;
    cout<<fixed<<endl;
}

void TestFastInvSinFloat(long lTestCount, long lBase)
{
    float fInvSin,  fAsinf, fAsin, fInvSinAccuracy, fTemp = 0.0f;
    long lStart, lEnd;

    cout<<"fast Asin<float> (between 0 and 1):"<<endl;
    cout<<"-----------------------------------"<<endl;
    cout<<endl;
    
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(float i=0.0f; i<=1.0f; i=i+1.0f/lBase)
        {
            fTemp += FastInvSin<float>(i);
        }

    lEnd    = clock();
    fInvSin = float(lEnd-lStart);

    cout<<scientific<<"asin with FastInvSin(sum): "<<fTemp<<" (against unwanted optimizations)"<<endl;

    fTemp  = 0.0f; 
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(float i=0.0f; i<=1.0f; i=i+1.0f/lBase)
        {
            fTemp += asinf(i);
        }

    lEnd   = clock();
    fAsinf = float(lEnd-lStart);

    cout<<"asin with asinf: "<<fTemp<<" (against unwanted optimizations)"<<endl;

    fTemp  = 0.0f; 
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(float i=0.0f; i<=1.0f; i=i+1.0f/lBase)
        {
            fTemp += asin(i);
        }

    lEnd   = clock();
    fAsin  = float(lEnd-lStart);

    cout<<"asin with asin: "<<fTemp<<" (against unwanted optimizations)"<<endl;

    cout<<fixed<<endl;
    cout<<"FastInvSin: "<<fInvSin/CLOCKS_PER_SEC<<" sec or "<<(1/fInvSin)*lBase<<" per second and Achievement: "<<(fAsinf/fInvSin)*100-100<<"%"<<endl;
    cout<<"asinf: "<<fAsinf/CLOCKS_PER_SEC<<" sec or "<<(1/fAsinf)*lBase<<" per second and Achievement: "<<(fAsinf/fAsinf)*100-100<<"%"<<endl;
    cout<<"asin: "<<fAsin/CLOCKS_PER_SEC<<" sec or "<<(1/fAsin)*lBase<<" per second and Achievement: "<<(fAsinf/fAsin)*100-100<<"%"<<endl;
    cout<<endl;

    fInvSinAccuracy = 0.0f;

    for(float i=0.0f; i<=1.0f; i=i+1.0f/lBase)
    {
        fTemp = asin(i)-FastInvSin<float>(i);

        if(fTemp<0) fTemp = -fTemp;

        if(fInvSinAccuracy<fTemp)
            fInvSinAccuracy = fTemp;
    }

    cout<<scientific<<"Accuracy FastInvSin: "<<fInvSinAccuracy<<endl;
    cout<<fixed<<endl;
}

void TestFastInvSinDouble(long lTestCount, long lBase)
{
    double fInvSin, fAsin, fInvSinAccuracy, fTemp = 0.0f;
    long lStart, lEnd;

    cout<<"fast Asin<double> (between 0 and 1):"<<endl;
    cout<<"------------------------------------"<<endl;
    cout<<endl;
    
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(double i=0.0f; i<=1.0f; i=i+1.0f/lBase)
        {
            fTemp += FastInvSin<double>(i);
        }

    lEnd    = clock();
    fInvSin = double(lEnd-lStart);

    cout<<scientific<<"asin with FastInvSin(sum): "<<fTemp<<" (against unwanted optimizations)"<<endl;

    fTemp  = 0.0f; 
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(double i=0.0f; i<=1.0f; i=i+1.0f/lBase)
        {
            fTemp += asin(i);
        }

    lEnd   = clock();
    fAsin  = double(lEnd-lStart);

    cout<<"asin with asin: "<<fTemp<<" (against unwanted optimizations)"<<endl;

    cout<<fixed<<endl;
    cout<<"FastInvSin: "<<fInvSin/CLOCKS_PER_SEC<<" sec or "<<(1/fInvSin)*lBase<<" per second and Achievement: "<<(fAsin/fInvSin)*100-100<<"%"<<endl;
    cout<<"asin: "<<fAsin/CLOCKS_PER_SEC<<" sec or "<<(1/fAsin)*lBase<<" per second and Achievement: "<<(fAsin/fAsin)*100-100<<"%"<<endl;
    cout<<endl;

    fInvSinAccuracy = 0.0f;

    for(double i=0.0f; i<=1.0f; i=i+1.0f/lBase)
    {
        fTemp = asin(i)-FastInvSin<double>(i);

        if(fTemp<0) fTemp = -fTemp;

        if(fInvSinAccuracy<fTemp)
            fInvSinAccuracy = fTemp;
    }

    cout<<scientific<<"Accuracy FastInvSin: "<<fInvSinAccuracy<<endl;
    cout<<fixed<<endl;
}

void TestFastInvCosFloat(long lTestCount, long lBase)
{
    float fInvCos,  fAcosf, fAcos, fInvCosAccuracy, fTemp = 0.0f;
    long lStart, lEnd;

    cout<<"fast acos<float> (between 0 and 1):"<<endl;
    cout<<"-----------------------------------"<<endl;
    cout<<endl;
    
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(float i=0.0f; i<=1.0f; i=i+1.0f/lBase)
        {
            fTemp += FastInvCos<float>(i);
        }

    lEnd    = clock();
    fInvCos = float(lEnd-lStart);

    cout<<scientific<<"acos with FastInvCos(sum): "<<fTemp<<" (against unwanted optimizations)"<<endl;

    fTemp  = 0.0f; 
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(float i=0.0f; i<=1.0f; i=i+1.0f/lBase)
        {
            fTemp += acosf(i);
        }

    lEnd   = clock();
    fAcosf = float(lEnd-lStart);

    cout<<"acos with acosf: "<<fTemp<<" (against unwanted optimizations)"<<endl;

    fTemp  = 0.0f; 
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(float i=0.0f; i<=1.0f; i=i+1.0f/lBase)
        {
            fTemp += acos(i);
        }

    lEnd   = clock();
    fAcos  = float(lEnd-lStart);

    cout<<"acos with acos: "<<fTemp<<" (against unwanted optimizations)"<<endl;

    cout<<fixed<<endl;
    cout<<"FastInvCos: "<<fInvCos/CLOCKS_PER_SEC<<" sec or "<<(1/fInvCos)*lBase<<" per second and Achievement: "<<(fAcosf/fInvCos)*100-100<<"%"<<endl;
    cout<<"acosf: "<<fAcosf/CLOCKS_PER_SEC<<" sec or "<<(1/fAcosf)*lBase<<" per second and Achievement: "<<(fAcosf/fAcosf)*100-100<<"%"<<endl;
    cout<<"acos: "<<fAcos/CLOCKS_PER_SEC<<" sec or "<<(1/fAcos)*lBase<<" per second and Achievement: "<<(fAcosf/fAcos)*100-100<<"%"<<endl;
    cout<<endl;

    fInvCosAccuracy = 0.0f;

    for(float i=0.0f; i<=1.0f; i=i+1.0f/lBase)
    {
        fTemp = acos(i)-FastInvCos<float>(i);

        if(fTemp<0) fTemp = -fTemp;

        if(fInvCosAccuracy<fTemp)
            fInvCosAccuracy = fTemp;
    }

    cout<<scientific<<"Accuracy FastInvCos: "<<fInvCosAccuracy<<endl;
    cout<<fixed<<endl;
}

void TestFastInvCosDouble(long lTestCount, long lBase)
{
    double fInvCos, fAcos, fInvCosAccuracy, fTemp = 0.0f;
    long lStart, lEnd;

    cout<<"fast acos<double> (between 0 and 1):"<<endl;
    cout<<"------------------------------------"<<endl;
    cout<<endl;
    
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(double i=0.0f; i<=1.0f; i=i+1.0f/lBase)
        {
            fTemp += FastInvCos<double>(i);
        }

    lEnd    = clock();
    fInvCos = double(lEnd-lStart);

    cout<<scientific<<"acos with FastInvCos(sum): "<<fTemp<<" (against unwanted optimizations)"<<endl;

    fTemp  = 0.0f; 
    lStart = clock();

    for(long j=0; j<lTestCount; j++)
        for(double i=0.0f; i<=1.0f; i=i+1.0f/lBase)
        {
            fTemp += acos(i);
        }

    lEnd   = clock();
    fAcos  = double(lEnd-lStart);

    cout<<"acos with acos: "<<fTemp<<" (against unwanted optimizations)"<<endl;

    cout<<fixed<<endl;
    cout<<"FastInvCos: "<<fInvCos/CLOCKS_PER_SEC<<" sec or "<<(1/fInvCos)*lBase<<" per second and Achievement: "<<(fAcos/fInvCos)*100-100<<"%"<<endl;
    cout<<"acos: "<<fAcos/CLOCKS_PER_SEC<<" sec or "<<(1/fAcos)*lBase<<" per second and Achievement: "<<(fAcos/fAcos)*100-100<<"%"<<endl;
    cout<<endl;

    fInvCosAccuracy = 0.0f;

    for(double i=0.0f; i<=1.0f; i=i+1.0f/lBase)
    {
        fTemp = acos(i)-FastInvCos<double>(i);

        if(fTemp<0) fTemp = -fTemp;

        if(fInvCosAccuracy<fTemp)
            fInvCosAccuracy = fTemp;
    }

    cout<<scientific<<"Accuracy FastInvCos: "<<fInvCosAccuracy<<endl;
    cout<<fixed<<endl;
}

int main(void)
{
    TestFastSinFloat(1e3, 1e4);
    TestFastSinDouble(1e3, 1e4);
    TestFastCosFloat(1e3, 1e4);
    TestFastCosDouble(1e3, 1e4);
    TestFastInvSinFloat(1e3, 1e4);
    TestFastInvSinDouble(1e3, 1e4);
    TestFastInvCosFloat(1e3, 1e4);
    TestFastInvCosDouble(1e3, 1e4);

    return 0;
}


mit folgendem Ergebnis:

Zitat


fast Sin<float> (between 0 and pi/2):
-------------------------------------

Sin with FastSin0(sum): 9.999758e+006 (against unwanted optimizations)
Sin with FastSin1: 9.999537e+006 (against unwanted optimizations)
Sin with FastSinTn: 9.999541e+006 (against unwanted optimizations)
Sin with sinf: 9.999537e+006 (against unwanted optimizations)
Sin with sin: 9.999537e+006 (against unwanted optimizations)

FastSin0: 0.093000 sec or 107.526878 per second and Achievement: 673.118286%
FastSin1: 0.172000 sec or 58.139534 per second and Achievement: 318.023254%
FastSinTn: 0.235000 sec or 42.553192 per second and Achievement: 205.957443%
sinf: 0.719000 sec or 13.908206 per second and Achievement: 0.000000%
sin: 0.734000 sec or 13.623979 per second and Achievement: -2.043597%

Accuracy FastSin0: 1.641530e-004
Accuracy FastSin1: 1.884717e-008
Accuracy FastSinTn: 3.525043e-006

fast Sin<double> (between 0 and pi/2):
--------------------------------------

Sin with FastSin0(sum): 9.999758e+006 (against unwanted optimizations)
Sin with FastSin1: 9.999537e+006 (against unwanted optimizations)
Sin with FastSinTn: 9.999541e+006 (against unwanted optimizations)
Sin with sin: 9.999537e+006 (against unwanted optimizations)

FastSin0: 0.094000 sec or 106.382979 per sec and achieves: 663.829787%
FastSin1: 0.172000 sec or 58.139535 per sec and achieves: 317.441860%
FastSinTn: 0.187000 sec or 53.475936 per sec and achieves: 283.957219%
sin: 0.718000 sec or 13.927577 per sec and achieves: 0.000000%

Accuracy FastSin0: 1.641486e-004
Accuracy FastSin1: 1.884715e-008
Accuracy FastSinTn: 3.540202e-006

fast Cos<float> (between 0 and pi/2):
-------------------------------------

Cos with FastCos0(sum): 1.000008e+007 (against unwanted optimizations)
Cos with FastCos1: 1.018673e+007 (against unwanted optimizations)
Cos with cosf: 1.000050e+007 (against unwanted optimizations)
Cos with cos: 1.000050e+007 (against unwanted optimizations)

FastCos0: 0.109000 sec or 91.743118 per second and Achievement: 588.073364%
FastCos1: 0.234000 sec or 42.735043 per second and Achievement: 220.512817%
cosf: 0.750000 sec or 13.333333 per second and Achievement: 0.000000%
cos: 0.750000 sec or 13.333333 per second and Achievement: 0.000000%

Accuracy FastCos0: 1.187912e-003
Accuracy FastCos1: 6.481527e-009

fast Cos<double> (between 0 and pi/2):
--------------------------------------

Cos with FastCos0(sum): 1.000008e+007 (against unwanted optimizations)
Cos with FastCos1: 1.000050e+007 (against unwanted optimizations)
Cos with cos: 1.000050e+007 (against unwanted optimizations)

FastCos0: 0.125000 sec or 80.000000 per sec and achieves: 500.000000%
FastCos1: 0.218000 sec or 45.871560 per sec and achieves: 244.036697%
cos: 0.750000 sec or 13.333333 per sec and achieves: 0.000000%

Accuracy FastCos0: 1.187937e-003
Accuracy FastCos1: 2.308052e-009

fast Asin<float> (between 0 and 1):
-----------------------------------

asin with FastInvSin(sum): 5.708732e+006 (against unwanted optimizations)
asin with asinf: 5.670888e+006 (against unwanted optimizations)
asin with asin: 5.670888e+006 (against unwanted optimizations)

FastInvSin: 0.156000 sec or 64.102562 per second and Achievement: 340.384613%
asinf: 0.687000 sec or 14.556041 per second and Achievement: 0.000000%
asin: 0.672000 sec or 14.880953 per second and Achievement: 2.232143%

Accuracy FastInvSin: 6.759167e-005

fast Asin<double> (between 0 and 1):
------------------------------------

asin with FastInvSin(sum): 5.708732e+006 (against unwanted optimizations)
asin with asin: 5.708752e+006 (against unwanted optimizations)

FastInvSin: 0.156000 sec or 64.102564 per second and Achievement: 330.769231%
asin: 0.672000 sec or 14.880952 per second and Achievement: 0.000000%

Accuracy FastInvSin: 6.757051e-005

fast acos<float> (between 0 and 1):
-----------------------------------

acos with FastInvCos(sum): 1.000080e+007 (against unwanted optimizations)
acos with acosf: 9.948369e+006 (against unwanted optimizations)
acos with acos: 9.948369e+006 (against unwanted optimizations)

FastInvCos: 0.156000 sec or 64.102562 per second and Achievement: 360.897430%
acosf: 0.719000 sec or 13.908206 per second and Achievement: 0.000000%
acos: 0.688000 sec or 14.534883 per second and Achievement: 4.505814%

Accuracy FastInvCos: 6.754795e-005

fast acos<double> (between 0 and 1):
------------------------------------

acos with FastInvCos(sum): 1.000080e+007 (against unwanted optimizations)
acos with acos: 1.000078e+007 (against unwanted optimizations)

FastInvCos: 0.156000 sec or 64.102564 per second and Achievement: 341.025641%
acos: 0.688000 sec or 14.534884 per second and Achievement: 0.000000%

Accuracy FastInvCos: 6.752679e-005

Drücken Sie eine beliebige Taste . . .


Das lässt sich doch schon sehen oder? Fällt jemandem noch eine optimierung ein?

//edit:

FastSin0: aus D. H. Eberly 3D Game Engine Architecture theor. Fehler: 1,7*10e-4 erw. Gewinn: 4.0
FastSin1: aus D. H. Eberly 3D Game Engine Architecture theor. Fehler: 1,9*10e-8 erw. Gewinn: 2.8
FastSinTn: Taylorentwicklung von D. Scherfgen(s. Thread) theor. Fehler: ? erw. Gewinn: ?
FastCos0: aus D. H. Eberly 3D Game Engine Architecture theor. Fehler: 1,2*10e-3 erw. Gewinn: 4.5
FastCos1: aus D. H. Eberly 3D Game Engine Architecture theor. Fehler: 6,5*10e-9 erw. Gewinn: 2.8
FastInvSin: aus D. H. Eberly 3D Game Engine Architecture theor. Fehler: 6,8*10e-5 erw. Gewinn: 7.5
FastInvCos: aus D. H. Eberly 3D Game Engine Architecture theor. Fehler: 1,3*10e-7 erw. Gewinn: 5.8
Was es alles gibt, das ich nich brauche - Aristoteles

Werbeanzeige