summaryrefslogtreecommitdiffstatshomepage
path: root/src/lib/softfloat/fsincos.c
blob: 889bb56679362732d45272c55847e403afdd5473 (plain) (blame)
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
/*============================================================================
This source file is an extension to the SoftFloat IEC/IEEE Floating-point
Arithmetic Package, Release 2b, written for Bochs (x86 achitecture simulator)
floating point emulation.

THIS SOFTWARE IS DISTRIBUTED AS IS, FOR FREE.  Although reasonable effort has
been made to avoid it, THIS SOFTWARE MAY CONTAIN FAULTS THAT WILL AT TIMES
RESULT IN INCORRECT BEHAVIOR.  USE OF THIS SOFTWARE IS RESTRICTED TO PERSONS
AND ORGANIZATIONS WHO CAN AND WILL TAKE FULL RESPONSIBILITY FOR ALL LOSSES,
COSTS, OR OTHER PROBLEMS THEY INCUR DUE TO THE SOFTWARE, AND WHO FURTHERMORE
EFFECTIVELY INDEMNIFY JOHN HAUSER AND THE INTERNATIONAL COMPUTER SCIENCE
INSTITUTE (possibly via similar legal warning) AGAINST ALL LOSSES, COSTS, OR
OTHER PROBLEMS INCURRED BY THEIR CUSTOMERS AND CLIENTS DUE TO THE SOFTWARE.

Derivative works are acceptable, even for commercial purposes, so long as
(1) the source code for the derivative work includes prominent notice that
the work is derivative, and (2) the source code includes prominent notice with
these four paragraphs for those parts of this code that are retained.
=============================================================================*/

/*============================================================================
 * Written for Bochs (x86 achitecture simulator) by
 *            Stanislav Shwartsman [sshwarts at sourceforge net]
 * ==========================================================================*/

#define FLOAT128

#define USE_estimateDiv128To64
#include "mamesf.h"
#include "softfloat.h"
//#include "softfloat-specialize"
#include "fpu_constant.h"

static const floatx80 floatx80_one = packFloatx80(0, 0x3fff, U64(0x8000000000000000));
static const floatx80 floatx80_default_nan = packFloatx80(0, 0xffff, U64(0xffffffffffffffff));

#define packFloat2x128m(zHi, zLo) {(zHi), (zLo)}
#define PACK_FLOAT_128(hi,lo) packFloat2x128m(LIT64(hi),LIT64(lo))

#define EXP_BIAS 0x3FFF

/*----------------------------------------------------------------------------
| Returns the fraction bits of the extended double-precision floating-point
| value `a'.
*----------------------------------------------------------------------------*/

INLINE bits64 extractFloatx80Frac( floatx80 a )
{
	return a.low;

}

/*----------------------------------------------------------------------------
| Returns the exponent bits of the extended double-precision floating-point
| value `a'.
*----------------------------------------------------------------------------*/

INLINE int32 extractFloatx80Exp( floatx80 a )
{
	return a.high & 0x7FFF;

}

/*----------------------------------------------------------------------------
| Returns the sign bit of the extended double-precision floating-point value
| `a'.
*----------------------------------------------------------------------------*/

INLINE flag extractFloatx80Sign( floatx80 a )
{
	return a.high>>15;

}

/*----------------------------------------------------------------------------
| Takes extended double-precision floating-point  NaN  `a' and returns the
| appropriate NaN result. If `a' is a signaling NaN, the invalid exception
| is raised.
*----------------------------------------------------------------------------*/

INLINE floatx80 propagateFloatx80NaNOneArg(floatx80 a)
{
	if (floatx80_is_signaling_nan(a))
		float_raise(float_flag_invalid);

	a.low |= U64(0xC000000000000000);

	return a;
}

/*----------------------------------------------------------------------------
| Normalizes the subnormal extended double-precision floating-point value
| represented by the denormalized significand `aSig'.  The normalized exponent
| and significand are stored at the locations pointed to by `zExpPtr' and
| `zSigPtr', respectively.
*----------------------------------------------------------------------------*/

void normalizeFloatx80Subnormal(UINT64 aSig, INT32 *zExpPtr, UINT64 *zSigPtr)
{
	int shiftCount = countLeadingZeros64(aSig);
	*zSigPtr = aSig<<shiftCount;
	*zExpPtr = 1 - shiftCount;
}

/* reduce trigonometric function argument using 128-bit precision
   M_PI approximation */
static UINT64 argument_reduction_kernel(UINT64 aSig0, int Exp, UINT64 *zSig0, UINT64 *zSig1)
{
	UINT64 term0, term1, term2;
	UINT64 aSig1 = 0;

	shortShift128Left(aSig1, aSig0, Exp, &aSig1, &aSig0);
	UINT64 q = estimateDiv128To64(aSig1, aSig0, FLOAT_PI_HI);
	mul128By64To192(FLOAT_PI_HI, FLOAT_PI_LO, q, &term0, &term1, &term2);
	sub128(aSig1, aSig0, term0, term1, zSig1, zSig0);
	while ((INT64)(*zSig1) < 0) {
		--q;
		add192(*zSig1, *zSig0, term2, 0, FLOAT_PI_HI, FLOAT_PI_LO, zSig1, zSig0, &term2);
	}
	*zSig1 = term2;
	return q;
}

static int reduce_trig_arg(int expDiff, int &zSign, UINT64 &aSig0, UINT64 &aSig1)
{
	UINT64 term0, term1, q = 0;

	if (expDiff < 0) {
		shift128Right(aSig0, 0, 1, &aSig0, &aSig1);
		expDiff = 0;
	}
	if (expDiff > 0) {
		q = argument_reduction_kernel(aSig0, expDiff, &aSig0, &aSig1);
	}
	else {
		if (FLOAT_PI_HI <= aSig0) {
			aSig0 -= FLOAT_PI_HI;
			q = 1;
		}
	}

	shift128Right(FLOAT_PI_HI, FLOAT_PI_LO, 1, &term0, &term1);
	if (! lt128(aSig0, aSig1, term0, term1))
	{
		int lt = lt128(term0, term1, aSig0, aSig1);
		int eq = eq128(aSig0, aSig1, term0, term1);

		if ((eq && (q & 1)) || lt) {
			zSign = !zSign;
			++q;
		}
		if (lt) sub128(FLOAT_PI_HI, FLOAT_PI_LO, aSig0, aSig1, &aSig0, &aSig1);
	}

	return (int)(q & 3);
}

#define SIN_ARR_SIZE 11
#define COS_ARR_SIZE 11

static float128 sin_arr[SIN_ARR_SIZE] =
{
	PACK_FLOAT_128(0x3fff000000000000, 0x0000000000000000), /*  1 */
	PACK_FLOAT_128(0xbffc555555555555, 0x5555555555555555), /*  3 */
	PACK_FLOAT_128(0x3ff8111111111111, 0x1111111111111111), /*  5 */
	PACK_FLOAT_128(0xbff2a01a01a01a01, 0xa01a01a01a01a01a), /*  7 */
	PACK_FLOAT_128(0x3fec71de3a556c73, 0x38faac1c88e50017), /*  9 */
	PACK_FLOAT_128(0xbfe5ae64567f544e, 0x38fe747e4b837dc7), /* 11 */
	PACK_FLOAT_128(0x3fde6124613a86d0, 0x97ca38331d23af68), /* 13 */
	PACK_FLOAT_128(0xbfd6ae7f3e733b81, 0xf11d8656b0ee8cb0), /* 15 */
	PACK_FLOAT_128(0x3fce952c77030ad4, 0xa6b2605197771b00), /* 17 */
	PACK_FLOAT_128(0xbfc62f49b4681415, 0x724ca1ec3b7b9675), /* 19 */
	PACK_FLOAT_128(0x3fbd71b8ef6dcf57, 0x18bef146fcee6e45)  /* 21 */
};

static float128 cos_arr[COS_ARR_SIZE] =
{
	PACK_FLOAT_128(0x3fff000000000000, 0x0000000000000000), /*  0 */
	PACK_FLOAT_128(0xbffe000000000000, 0x0000000000000000), /*  2 */
	PACK_FLOAT_128(0x3ffa555555555555, 0x5555555555555555), /*  4 */
	PACK_FLOAT_128(0xbff56c16c16c16c1, 0x6c16c16c16c16c17), /*  6 */
	PACK_FLOAT_128(0x3fefa01a01a01a01, 0xa01a01a01a01a01a), /*  8 */
	PACK_FLOAT_128(0xbfe927e4fb7789f5, 0xc72ef016d3ea6679), /* 10 */
	PACK_FLOAT_128(0x3fe21eed8eff8d89, 0x7b544da987acfe85), /* 12 */
	PACK_FLOAT_128(0xbfda93974a8c07c9, 0xd20badf145dfa3e5), /* 14 */
	PACK_FLOAT_128(0x3fd2ae7f3e733b81, 0xf11d8656b0ee8cb0), /* 16 */
	PACK_FLOAT_128(0xbfca6827863b97d9, 0x77bb004886a2c2ab), /* 18 */
	PACK_FLOAT_128(0x3fc1e542ba402022, 0x507a9cad2bf8f0bb)  /* 20 */
};

extern float128 OddPoly (float128 x, float128 *arr, unsigned n);

/* 0 <= x <= pi/4 */
INLINE float128 poly_sin(float128 x)
{
	//                 3     5     7     9     11     13     15
	//                x     x     x     x     x      x      x
	// sin (x) ~ x - --- + --- - --- + --- - ---- + ---- - ---- =
	//                3!    5!    7!    9!    11!    13!    15!
	//
	//                 2     4     6     8     10     12     14
	//                x     x     x     x     x      x      x
	//   = x * [ 1 - --- + --- - --- + --- - ---- + ---- - ---- ] =
	//                3!    5!    7!    9!    11!    13!    15!
	//
	//           3                          3
	//          --       4k                --        4k+2
	//   p(x) = >  C  * x   > 0     q(x) = >  C   * x     < 0
	//          --  2k                     --  2k+1
	//          k=0                        k=0
	//
	//                          2
	//   sin(x) ~ x * [ p(x) + x * q(x) ]
	//

	return OddPoly(x, sin_arr, SIN_ARR_SIZE);
}

extern float128 EvenPoly(float128 x, float128 *arr, unsigned n);

/* 0 <= x <= pi/4 */
INLINE float128 poly_cos(float128 x)
{
	//                 2     4     6     8     10     12     14
	//                x     x     x     x     x      x      x
	// cos (x) ~ 1 - --- + --- - --- + --- - ---- + ---- - ----
	//                2!    4!    6!    8!    10!    12!    14!
	//
	//           3                          3
	//          --       4k                --        4k+2
	//   p(x) = >  C  * x   > 0     q(x) = >  C   * x     < 0
	//          --  2k                     --  2k+1
	//          k=0                        k=0
	//
	//                      2
	//   cos(x) ~ [ p(x) + x * q(x) ]
	//

	return EvenPoly(x, cos_arr, COS_ARR_SIZE);
}

INLINE void sincos_invalid(floatx80 *sin_a, floatx80 *cos_a, floatx80 a)
{
	if (sin_a) *sin_a = a;
	if (cos_a) *cos_a = a;
}

INLINE void sincos_tiny_argument(floatx80 *sin_a, floatx80 *cos_a, floatx80 a)
{
	if (sin_a) *sin_a = a;
	if (cos_a) *cos_a = floatx80_one;
}

static floatx80 sincos_approximation(int neg, float128 r, UINT64 quotient)
{
	if (quotient & 0x1) {
		r = poly_cos(r);
		neg = 0;
	} else  {
		r = poly_sin(r);
	}

	floatx80 result = float128_to_floatx80(r);
	if (quotient & 0x2)
		neg = ! neg;

	if (neg)
		result = floatx80_chs(result);

	return result;
}

// =================================================
// SFFSINCOS               Compute sin(x) and cos(x)
// =================================================

//
// Uses the following identities:
// ----------------------------------------------------------
//
//  sin(-x) = -sin(x)
//  cos(-x) =  cos(x)
//
//  sin(x+y) = sin(x)*cos(y)+cos(x)*sin(y)
//  cos(x+y) = sin(x)*sin(y)+cos(x)*cos(y)
//
//  sin(x+ pi/2)  =  cos(x)
//  sin(x+ pi)    = -sin(x)
//  sin(x+3pi/2)  = -cos(x)
//  sin(x+2pi)    =  sin(x)
//

int sf_fsincos(floatx80 a, floatx80 *sin_a, floatx80 *cos_a)
{
	UINT64 aSig0, aSig1 = 0;
	INT32 aExp, zExp, expDiff;
	int aSign, zSign;
	int q = 0;

	aSig0 = extractFloatx80Frac(a);
	aExp = extractFloatx80Exp(a);
	aSign = extractFloatx80Sign(a);

	/* invalid argument */
	if (aExp == 0x7FFF) {
		if ((UINT64) (aSig0<<1)) {
			sincos_invalid(sin_a, cos_a, propagateFloatx80NaNOneArg(a));
			return 0;
		}

		float_raise(float_flag_invalid);
		sincos_invalid(sin_a, cos_a, floatx80_default_nan);
		return 0;
	}

	if (aExp == 0) {
		if (aSig0 == 0) {
			sincos_tiny_argument(sin_a, cos_a, a);
			return 0;
		}

//        float_raise(float_flag_denormal);

		/* handle pseudo denormals */
		if (! (aSig0 & U64(0x8000000000000000)))
		{
			float_raise(float_flag_inexact);
			if (sin_a)
				float_raise(float_flag_underflow);
			sincos_tiny_argument(sin_a, cos_a, a);
			return 0;
		}

		normalizeFloatx80Subnormal(aSig0, &aExp, &aSig0);
	}

	zSign = aSign;
	zExp = EXP_BIAS;
	expDiff = aExp - zExp;

	/* argument is out-of-range */
	if (expDiff >= 63)
		return -1;

	float_raise(float_flag_inexact);

	if (expDiff < -1) {    // doesn't require reduction
		if (expDiff <= -68) {
			a = packFloatx80(aSign, aExp, aSig0);
			sincos_tiny_argument(sin_a, cos_a, a);
			return 0;
		}
		zExp = aExp;
	}
	else {
		q = reduce_trig_arg(expDiff, zSign, aSig0, aSig1);
	}

	/* **************************** */
	/* argument reduction completed */
	/* **************************** */

	/* using float128 for approximation */
	float128 r = normalizeRoundAndPackFloat128(0, zExp-0x10, aSig0, aSig1);

	if (aSign) q = -q;
	if (sin_a) *sin_a = sincos_approximation(zSign, r,   q);
	if (cos_a) *cos_a = sincos_approximation(zSign, r, q+1);

	return 0;
}

int floatx80_fsin(floatx80 &a)
{
	return sf_fsincos(a, &a, 0);
}

int floatx80_fcos(floatx80 &a)
{
	return sf_fsincos(a, 0, &a);
}

// =================================================
// FPTAN                 Compute tan(x)
// =================================================

//
// Uses the following identities:
//
// 1. ----------------------------------------------------------
//
//  sin(-x) = -sin(x)
//  cos(-x) =  cos(x)
//
//  sin(x+y) = sin(x)*cos(y)+cos(x)*sin(y)
//  cos(x+y) = sin(x)*sin(y)+cos(x)*cos(y)
//
//  sin(x+ pi/2)  =  cos(x)
//  sin(x+ pi)    = -sin(x)
//  sin(x+3pi/2)  = -cos(x)
//  sin(x+2pi)    =  sin(x)
//
// 2. ----------------------------------------------------------
//
//           sin(x)
//  tan(x) = ------
//           cos(x)
//

int floatx80_ftan(floatx80 &a)
{
	UINT64 aSig0, aSig1 = 0;
	INT32 aExp, zExp, expDiff;
	int aSign, zSign;
	int q = 0;

	aSig0 = extractFloatx80Frac(a);
	aExp = extractFloatx80Exp(a);
	aSign = extractFloatx80Sign(a);

	/* invalid argument */
	if (aExp == 0x7FFF) {
		if ((UINT64) (aSig0<<1))
		{
			a = propagateFloatx80NaNOneArg(a);
			return 0;
		}

		float_raise(float_flag_invalid);
		a = floatx80_default_nan;
		return 0;
	}

	if (aExp == 0) {
		if (aSig0 == 0) return 0;
//        float_raise(float_flag_denormal);
		/* handle pseudo denormals */
		if (! (aSig0 & U64(0x8000000000000000)))
		{
			float_raise(float_flag_inexact | float_flag_underflow);
			return 0;
		}
		normalizeFloatx80Subnormal(aSig0, &aExp, &aSig0);
	}

	zSign = aSign;
	zExp = EXP_BIAS;
	expDiff = aExp - zExp;

	/* argument is out-of-range */
	if (expDiff >= 63)
		return -1;

	float_raise(float_flag_inexact);

	if (expDiff < -1) {    // doesn't require reduction
		if (expDiff <= -68) {
			a = packFloatx80(aSign, aExp, aSig0);
			return 0;
		}
		zExp = aExp;
	}
	else {
		q = reduce_trig_arg(expDiff, zSign, aSig0, aSig1);
	}

	/* **************************** */
	/* argument reduction completed */
	/* **************************** */

	/* using float128 for approximation */
	float128 r = normalizeRoundAndPackFloat128(0, zExp-0x10, aSig0, aSig1);

	float128 sin_r = poly_sin(r);
	float128 cos_r = poly_cos(r);

	if (q & 0x1) {
		r = float128_div(cos_r, sin_r);
		zSign = ! zSign;
	} else {
		r = float128_div(sin_r, cos_r);
	}

	a = float128_to_floatx80(r);
	if (zSign)
		a = floatx80_chs(a);

	return 0;
}

//                            2         3         4               n
// f(x) ~ C + (C * x) + (C * x) + (C * x) + (C * x) + ... + (C * x)
//         0    1         2         3         4               n
//
//          --       2k                --        2k+1
//   p(x) = >  C  * x           q(x) = >  C   * x
//          --  2k                     --  2k+1
//
//   f(x) ~ [ p(x) + x * q(x) ]
//

float128 EvalPoly(float128 x, float128 *arr, unsigned n)
{
	float128 x2 = float128_mul(x, x);
	unsigned i;

	assert(n > 1);

	float128 r1 = arr[--n];
	i = n;
	while(i >= 2) {
		r1 = float128_mul(r1, x2);
		i -= 2;
		r1 = float128_add(r1, arr[i]);
	}
	if (i) r1 = float128_mul(r1, x);

	float128 r2 = arr[--n];
	i = n;
	while(i >= 2) {
		r2 = float128_mul(r2, x2);
		i -= 2;
		r2 = float128_add(r2, arr[i]);
	}
	if (i) r2 = float128_mul(r2, x);

	return float128_add(r1, r2);
}

//                  2         4         6         8               2n
// f(x) ~ C + (C * x) + (C * x) + (C * x) + (C * x) + ... + (C * x)
//         0    1         2         3         4               n
//
//          --       4k                --        4k+2
//   p(x) = >  C  * x           q(x) = >  C   * x
//          --  2k                     --  2k+1
//
//                    2
//   f(x) ~ [ p(x) + x * q(x) ]
//

float128 EvenPoly(float128 x, float128 *arr, unsigned n)
{
		return EvalPoly(float128_mul(x, x), arr, n);
}

//                        3         5         7         9               2n+1
// f(x) ~ (C * x) + (C * x) + (C * x) + (C * x) + (C * x) + ... + (C * x)
//          0         1         2         3         4               n
//                        2         4         6         8               2n
//      = x * [ C + (C * x) + (C * x) + (C * x) + (C * x) + ... + (C * x)
//               0    1         2         3         4               n
//
//          --       4k                --        4k+2
//   p(x) = >  C  * x           q(x) = >  C   * x
//          --  2k                     --  2k+1
//
//                        2
//   f(x) ~ x * [ p(x) + x * q(x) ]
//

float128 OddPoly(float128 x, float128 *arr, unsigned n)
{
		return float128_mul(x, EvenPoly(x, arr, n));
}

/*----------------------------------------------------------------------------
| Scales extended double-precision floating-point value in operand `a' by 
| value `b'. The function truncates the value in the second operand 'b' to 
| an integral value and adds that value to the exponent of the operand 'a'.
| The operation performed according to the IEC/IEEE Standard for Binary 
| Floating-Point Arithmetic.
*----------------------------------------------------------------------------*/

extern floatx80 propagateFloatx80NaN( floatx80 a, floatx80 b );

floatx80 floatx80_scale(floatx80 a, floatx80 b)
{
    sbits32 aExp, bExp;
    bits64 aSig, bSig;

    // handle unsupported extended double-precision floating encodings
/*    if (floatx80_is_unsupported(a) || floatx80_is_unsupported(b))
    {
        float_raise(float_flag_invalid);
        return floatx80_default_nan;
    }*/

    aSig = extractFloatx80Frac(a);
    aExp = extractFloatx80Exp(a);
    int aSign = extractFloatx80Sign(a);
    bSig = extractFloatx80Frac(b);
    bExp = extractFloatx80Exp(b);
    int bSign = extractFloatx80Sign(b);

    if (aExp == 0x7FFF) {
        if ((bits64) (aSig<<1) || ((bExp == 0x7FFF) && (bits64) (bSig<<1))) 
        {
            return propagateFloatx80NaN(a, b);
        }
        if ((bExp == 0x7FFF) && bSign) {
            float_raise(float_flag_invalid);
            return floatx80_default_nan;
        }
        if (bSig && (bExp == 0)) float_raise(float_flag_denormal);
        return a;
    }
    if (bExp == 0x7FFF) {
        if ((bits64) (bSig<<1)) return propagateFloatx80NaN(a, b);
        if ((aExp | aSig) == 0) {
            if (! bSign) {
                float_raise(float_flag_invalid);
                return floatx80_default_nan;
            }
            return a;
        }
        if (aSig && (aExp == 0)) float_raise(float_flag_denormal);
        if (bSign) return packFloatx80(aSign, 0, 0);
        return packFloatx80(aSign, 0x7FFF, U64(0x8000000000000000));
    }
    if (aExp == 0) {
        if (aSig == 0) return a;
        float_raise(float_flag_denormal);
        normalizeFloatx80Subnormal(aSig, &aExp, &aSig);
    }
    if (bExp == 0) {
        if (bSig == 0) return a;
        float_raise(float_flag_denormal);
        normalizeFloatx80Subnormal(bSig, &bExp, &bSig);
    }

    if (bExp > 0x400E) {   
        /* generate appropriate overflow/underflow */
        return roundAndPackFloatx80(80, aSign, 
                          bSign ? -0x3FFF : 0x7FFF, aSig, 0);
    }
    if (bExp < 0x3FFF) return a;

    int shiftCount = 0x403E - bExp;
    bSig >>= shiftCount;
    sbits32 scale = bSig;
    if (bSign) scale = -scale; /* -32768..32767 */
    return
        roundAndPackFloatx80(80, aSign, aExp+scale, aSig, 0);
}