1 |
|
---|
2 | /*============================================================================
|
---|
3 |
|
---|
4 | This C source file is part of the SoftFloat IEEE Floating-Point Arithmetic
|
---|
5 | Package, Release 3e, by John R. Hauser.
|
---|
6 |
|
---|
7 | Copyright 2011, 2012, 2013, 2014, 2015, 2016 The Regents of the University of
|
---|
8 | California. All rights reserved.
|
---|
9 |
|
---|
10 | Redistribution and use in source and binary forms, with or without
|
---|
11 | modification, are permitted provided that the following conditions are met:
|
---|
12 |
|
---|
13 | 1. Redistributions of source code must retain the above copyright notice,
|
---|
14 | this list of conditions, and the following disclaimer.
|
---|
15 |
|
---|
16 | 2. Redistributions in binary form must reproduce the above copyright notice,
|
---|
17 | this list of conditions, and the following disclaimer in the documentation
|
---|
18 | and/or other materials provided with the distribution.
|
---|
19 |
|
---|
20 | 3. Neither the name of the University nor the names of its contributors may
|
---|
21 | be used to endorse or promote products derived from this software without
|
---|
22 | specific prior written permission.
|
---|
23 |
|
---|
24 | THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS "AS IS", AND ANY
|
---|
25 | EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
|
---|
26 | WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE, ARE
|
---|
27 | DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE FOR ANY
|
---|
28 | DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
|
---|
29 | (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
|
---|
30 | LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
|
---|
31 | ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
---|
32 | (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
|
---|
33 | SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
---|
34 |
|
---|
35 | =============================================================================*/
|
---|
36 |
|
---|
37 | #include <stdbool.h>
|
---|
38 | #include <stdint.h>
|
---|
39 | #include "platform.h"
|
---|
40 | #include "internals.h"
|
---|
41 | #include "specialize.h"
|
---|
42 | #include "softfloat.h"
|
---|
43 |
|
---|
44 | float16_t f16_rem( float16_t a, float16_t b SOFTFLOAT_STATE_DECL_COMMA )
|
---|
45 | {
|
---|
46 | union ui16_f16 uA;
|
---|
47 | uint_fast16_t uiA;
|
---|
48 | bool signA;
|
---|
49 | int_fast8_t expA;
|
---|
50 | uint_fast16_t sigA;
|
---|
51 | union ui16_f16 uB;
|
---|
52 | uint_fast16_t uiB;
|
---|
53 | int_fast8_t expB;
|
---|
54 | uint_fast16_t sigB;
|
---|
55 | struct exp8_sig16 normExpSig;
|
---|
56 | uint16_t rem;
|
---|
57 | int_fast8_t expDiff;
|
---|
58 | uint_fast16_t q;
|
---|
59 | uint32_t recip32, q32;
|
---|
60 | uint16_t altRem, meanRem;
|
---|
61 | bool signRem;
|
---|
62 | uint_fast16_t uiZ;
|
---|
63 | union ui16_f16 uZ;
|
---|
64 |
|
---|
65 | /*------------------------------------------------------------------------
|
---|
66 | *------------------------------------------------------------------------*/
|
---|
67 | uA.f = a;
|
---|
68 | uiA = uA.ui;
|
---|
69 | signA = signF16UI( uiA );
|
---|
70 | expA = expF16UI( uiA );
|
---|
71 | sigA = fracF16UI( uiA );
|
---|
72 | uB.f = b;
|
---|
73 | uiB = uB.ui;
|
---|
74 | expB = expF16UI( uiB );
|
---|
75 | sigB = fracF16UI( uiB );
|
---|
76 | /*------------------------------------------------------------------------
|
---|
77 | *------------------------------------------------------------------------*/
|
---|
78 | if ( expA == 0x1F ) {
|
---|
79 | if ( sigA || ((expB == 0x1F) && sigB) ) goto propagateNaN;
|
---|
80 | goto invalid;
|
---|
81 | }
|
---|
82 | if ( expB == 0x1F ) {
|
---|
83 | if ( sigB ) goto propagateNaN;
|
---|
84 | return a;
|
---|
85 | }
|
---|
86 | /*------------------------------------------------------------------------
|
---|
87 | *------------------------------------------------------------------------*/
|
---|
88 | if ( ! expB ) {
|
---|
89 | if ( ! sigB ) goto invalid;
|
---|
90 | normExpSig = softfloat_normSubnormalF16Sig( sigB );
|
---|
91 | expB = normExpSig.exp;
|
---|
92 | sigB = normExpSig.sig;
|
---|
93 | }
|
---|
94 | if ( ! expA ) {
|
---|
95 | if ( ! sigA ) return a;
|
---|
96 | normExpSig = softfloat_normSubnormalF16Sig( sigA );
|
---|
97 | expA = normExpSig.exp;
|
---|
98 | sigA = normExpSig.sig;
|
---|
99 | }
|
---|
100 | /*------------------------------------------------------------------------
|
---|
101 | *------------------------------------------------------------------------*/
|
---|
102 | rem = sigA | 0x0400;
|
---|
103 | sigB |= 0x0400;
|
---|
104 | expDiff = expA - expB;
|
---|
105 | if ( expDiff < 1 ) {
|
---|
106 | if ( expDiff < -1 ) return a;
|
---|
107 | sigB <<= 3;
|
---|
108 | if ( expDiff ) {
|
---|
109 | rem <<= 2;
|
---|
110 | q = 0;
|
---|
111 | } else {
|
---|
112 | rem <<= 3;
|
---|
113 | q = (sigB <= rem);
|
---|
114 | if ( q ) rem -= sigB;
|
---|
115 | }
|
---|
116 | } else {
|
---|
117 | recip32 = softfloat_approxRecip32_1( (uint_fast32_t) sigB<<21 );
|
---|
118 | /*--------------------------------------------------------------------
|
---|
119 | | Changing the shift of `rem' here requires also changing the initial
|
---|
120 | | subtraction from `expDiff'.
|
---|
121 | *--------------------------------------------------------------------*/
|
---|
122 | rem <<= 4;
|
---|
123 | expDiff -= 31;
|
---|
124 | /*--------------------------------------------------------------------
|
---|
125 | | The scale of `sigB' affects how many bits are obtained during each
|
---|
126 | | cycle of the loop. Currently this is 29 bits per loop iteration,
|
---|
127 | | which is believed to be the maximum possible.
|
---|
128 | *--------------------------------------------------------------------*/
|
---|
129 | sigB <<= 3;
|
---|
130 | for (;;) {
|
---|
131 | q32 = (rem * (uint_fast64_t) recip32)>>16;
|
---|
132 | if ( expDiff < 0 ) break;
|
---|
133 | rem = -((uint_fast16_t) q32 * sigB);
|
---|
134 | expDiff -= 29;
|
---|
135 | }
|
---|
136 | /*--------------------------------------------------------------------
|
---|
137 | | (`expDiff' cannot be less than -30 here.)
|
---|
138 | *--------------------------------------------------------------------*/
|
---|
139 | q32 >>= ~expDiff & 31;
|
---|
140 | q = q32;
|
---|
141 | rem = (rem<<(expDiff + 30)) - q * sigB;
|
---|
142 | }
|
---|
143 | /*------------------------------------------------------------------------
|
---|
144 | *------------------------------------------------------------------------*/
|
---|
145 | do {
|
---|
146 | altRem = rem;
|
---|
147 | ++q;
|
---|
148 | rem -= sigB;
|
---|
149 | } while ( ! (rem & 0x8000) );
|
---|
150 | meanRem = rem + altRem;
|
---|
151 | if ( (meanRem & 0x8000) || (! meanRem && (q & 1)) ) rem = altRem;
|
---|
152 | signRem = signA;
|
---|
153 | if ( 0x8000 <= rem ) {
|
---|
154 | signRem = ! signRem;
|
---|
155 | rem = -rem;
|
---|
156 | }
|
---|
157 | return softfloat_normRoundPackToF16( signRem, expB, rem SOFTFLOAT_STATE_ARG_COMMA );
|
---|
158 | /*------------------------------------------------------------------------
|
---|
159 | *------------------------------------------------------------------------*/
|
---|
160 | propagateNaN:
|
---|
161 | uiZ = softfloat_propagateNaNF16UI( uiA, uiB SOFTFLOAT_STATE_ARG_COMMA );
|
---|
162 | goto uiZ;
|
---|
163 | invalid:
|
---|
164 | softfloat_raiseFlags( softfloat_flag_invalid SOFTFLOAT_STATE_ARG_COMMA );
|
---|
165 | uiZ = defaultNaNF16UI;
|
---|
166 | uiZ:
|
---|
167 | uZ.ui = uiZ;
|
---|
168 | return uZ.f;
|
---|
169 |
|
---|
170 | }
|
---|
171 |
|
---|