about summary refs log tree commit diff
path: root/REORG.TODO/sysdeps/ia64/fpu/libm_frexpl.S
blob: 3c3bba08225222c089230dd89c7645034e4c3ffe (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
.file "libm_frexpl.s"


// Copyright (c) 2000 - 2003, Intel Corporation
// All rights reserved.
//
// Contributed 2000 by the Intel Numerics Group, Intel Corporation
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
// * Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
//
// * Redistributions in binary form must reproduce the above copyright
// notice, this list of conditions and the following disclaimer in the
// documentation and/or other materials provided with the distribution.
//
// * The name of Intel Corporation may not be used to endorse or promote
// products derived from this software without specific prior written
// permission.

// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL INTEL OR ITS
// CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
// EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
// PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
// OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY OR TORT (INCLUDING
// NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
// SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
//
// Intel Corporation is the author of this code, and requests that all
// problem reports or change requests be submitted to it directly at
// http://www.intel.com/software/products/opensource/libraries/num.htm.
//
// History
//==============================================================
// 02/02/00 Initial version
// 03/20/00 Improved speed
// 06/01/00 Fixed bug when x a double-extended denormal
// 12/08/00 Corrected label on .endp
// 01/23/02 Added handling for int 32 or 64 bits
// 05/20/02 Cleaned up namespace and sf0 syntax
// 02/10/03 Reordered header: .section, .global, .proc, .align
//
// API
//==============================================================
// long double __libm_frexpl(long double x, int* y, int int_type)
// input  floating point f8, pointer to y (r34), int int_type (r35)
// output floating point f8, returns the fraction of x, 0.5 <= fraction < 1.0
// output int* y, returns the true exponent of x
//
// int_type = 0 if int is 32 bits
// int_type = 1 if int is 64 bits
//
// int* y is returned as a 32 bit integer if int_type = 0
// int* y is returned as a 64 bit integer if int_type = 1
//
// Overview of operation
//==============================================================
// break a floating point x number into fraction and an exponent
// The fraction is returned as a long double
// The exponent is returned as an integer pointed to by y
//    This is a true (not a biased exponent) but 0fffe is subtracted
//    as a bias instead of 0xffff. This is because the fraction returned
//    is between 0.5 and 1.0, not the expected IEEE range.
//
// The fraction is 0.5 <= fraction < 1.0
//
// Registers used
//==============================================================
//
// general registers:
// r14  exponent bias for x negative
// r15  exponent bias for x positive
// r16  signexp of x
// r17  exponent mask
// r18  exponent of x
// r19  exponent result
// r20  signexp of 2^64
// r32-33  on input contains the 80-bit IEEE long double that is in f8
// r34  on input pointer to 32-bit or 64-bit integer for exponent
// r35  on input contains 0 if output int is 32 bits, else output int is 64 bits
//
// predicate registers:
// p6   set if x is Nan, zero, or infinity
// p7   set if x negative
// p8   set if x positive
// p9   set if x double-extended denormal
// p10  set if int_type = 0, 32-bit integer
// p11  set if int_type = 1, 64-bit integer
//
// floating-point registers:
// f8  input, output
// f9  normalized x
// f10 signexp for significand result for x positive
// f11 signexp for significand result for x negative
// f12 2^64

.section .text
GLOBAL_LIBM_ENTRY(__libm_frexpl)

// Set signexp for significand result for x>0
// If x is a NaN, zero, or infinity, return it.
// Put 0 in the int pointer.
// x NAN, ZERO, INFINITY?
// Set signexp for significand result for x<0
{ .mfi
        mov         r15 = 0x0fffe
        fclass.m    p6,p7 = f8, 0xe7
        mov         r14 = 0x2fffe
}
// Form signexp of 2^64 in case x double-extended denormal
// Save the normalized value of input in f9
// The normalization also sets fault flags and takes faults if necessary
{ .mfi
        mov         r20 = 0x1003f
        fnorm.s0    f9 = f8
        nop.i 999 ;;
}

// Move signexp for significand result for x>0 to FP reg
// Form 2^64 in case x double-extended denormal
{ .mmi
        setf.exp    f10 = r15
        setf.exp    f12 = r20
        nop.i 999 ;;
}

// Move signexp for significand result for x<0 to FP reg
// p7 if x<0, else p8
// If x=0,nan,inf, set p10 if output int to be 32 bits, or set p11 if 64 bits
{ .mfi
        setf.exp    f11 = r14
(p7)    fcmp.lt.s0  p7,p8 = f8,f0
(p6)    cmp.eq.unc  p10,p11 = r35, r0 ;;
}

// If x NAN, ZERO, INFINITY, set *y=0 and exit
{ .mmb
(p10)   st4         [r34] = r0      // Store *y=0 as 32-bit integer
(p11)   st8         [r34] = r0      // Store *y=0 as 64-bit integer
(p6)    br.ret.spnt b0 ;;
}

// Form exponent mask
// Test for fnorm(x) denormal, means x double-extended denormal
{ .mfi
        mov         r17 = 0x1ffff
        fclass.m    p9,p0 = f9, 0x0b
        nop.i 999 ;;
}

// If x double-extended denormal add 64 to exponent bias for scaling
// If x double-extended denormal multiply x * 2^64 which is normal
// Set p10 if output int to be 32 bits, or set p11 if 64 bits
{ .mfi
(p9)    add         r15 = 64, r15
(p9)    fmpy.s0     f9 = f9, f12
        cmp.eq      p10,p11 = r35, r0 ;;
}

// true exponent stored to int pointer
// the bias is treated as 0xfffe instead of
// normal 0xffff because we want the significand
// to be in the range <=0.5 sig < 1.0
// Store the value of the exponent at the pointer in r34

// If x>0 form significand result
{ .mfi
        nop.m 999
(p8)    fmerge.se   f8 = f10,f9
        nop.i 999  ;;
}

// Get signexp of normalized x
// If x<0 form significand result
{ .mfi
        getf.exp    r16 = f9
(p7)    fmerge.se   f8 = f11,f9
        nop.i 999  ;;
}

// Get exp of normalized x
// Subtract off bias to get true exponent of x
{ .mmi
        and         r18 = r17,r16 ;;
        sub         r19 = r18,r15
        nop.i 999  ;;
}

// Store int *y as a 32-bit integer
// Make the value a long double
{ .mfi
(p10)   st4         [r34] = r19        // Store *y as 32-bit integer
        fnorm.s0    f8 = f8
        nop.i 999
}
{ .mfb
(p11)   st8         [r34] = r19        // Store *y as 64-bit integer
        nop.f 999
        br.ret.sptk b0 ;;
}

GLOBAL_LIBM_END(__libm_frexpl)