1 /* $NetBSD: n_cbrt.S,v 1.7 2013/11/24 15:12:10 martin Exp $ */
3 * Copyright (c) 1985, 1993
4 * The Regents of the University of California. All rights reserved.
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
9 * 1. Redistributions of source code must retain the above copyright
10 * notice, this list of conditions and the following disclaimer.
11 * 2. Redistributions in binary form must reproduce the above copyright
12 * notice, this list of conditions and the following disclaimer in the
13 * documentation and/or other materials provided with the distribution.
14 * 3. Neither the name of the University nor the names of its contributors
15 * may be used to endorse or promote products derived from this software
16 * without specific prior written permission.
18 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
19 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
20 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
22 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
23 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
24 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
25 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
26 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
27 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 * @(#)cbrt.s 8.1 (Berkeley) 6/4/93
33 #include <machine/asm.h>
36 * double cbrt(double arg)
37 * W. Kahan, 10/13/80. revised 1/13/84 for keeping sign symmetry
38 * error check by E LeBlanc, 8/18/82
39 * Revised and tested by K.C. Ng, 5/2/85
40 * Max error less than 0.667 ulps (unit in the last places)
45 calls $2,_C_LABEL(cbrt)
50 WEAK_ALIAS(cbrtl, cbrt)
54 ENTRY(d_cbrt, 0x00c0) # save %r6 & %r7
55 movq 4(%ap),%r0 # %r0 = argument x
58 ENTRY(dcbrt_, 0x00c0) # save %r6 & %r7
59 movq *4(%ap),%r0 # %r0 = argument x
61 dcbrt2: bicw3 $0x807f,%r0,%r2 # biased exponent of x
62 jeql return # dcbrt(0)=0 dcbrt(res)=res. operand
63 bicw3 $0x7fff,%r0,%ap # ap has sign(x)
64 xorw2 %ap,%r0 # %r0 is abs(x)
65 movl %r0,%r2 # %r2 has abs(x)
66 rotl $16,%r2,%r2 # %r2 = |x| with bits unscrambled
67 divl2 $3,%r2 # rough dcbrt with bias/3
68 addl2 B,%r2 # restore bias, diminish fraction
69 rotl $16,%r2,%r2 # %r2=|q|=|dcbrt| to 5 bits
70 mulf3 %r2,%r2,%r3 # %r3 =qq
71 divf2 %r0,%r3 # %r3 = qq/x
73 addf2 C,%r3 # %r3 = s = C + qqq/x
74 divf3 %r3,D,%r4 # %r4 = D/s
76 addf2 %r4,%r3 # %r3 = s + E + D/s
77 divf3 %r3,F,%r3 # %r3 = F / (s + E + D/s)
78 addf2 G,%r3 # %r3 = G + F / (s + E + D/s)
79 mulf2 %r3,%r2 # %r2 = q%r3 = new q to 23 bits
80 clrl %r3 # %r2:%r3 = q as double float
81 muld3 %r2,%r2,%r4 # %r4:%r5 = qq exactly
82 divd2 %r4,%r0 # %r0:%r1 = x/(q*q) rounded
83 subd3 %r2,%r0,%r6 # %r6:%r7 = x/(q*q) - q exactly
84 movq %r2,%r4 # %r4:%r5 = q
85 addw2 $0x80,%r4 # %r4:%r5 = 2 * q
86 addd2 %r0,%r4 # %r4:%r5 = 2*q + x/(q*q)
87 divd2 %r4,%r6 # %r6:%r7 = (x/(q*q)-q)/(2*q+x/(q*q))
88 muld2 %r2,%r6 # %r6:%r7 = q*(x/(q*q)-q)/(2*q+x/(q*q))
89 addd3 %r6,%r2,%r0 # %r0:%r1 = q + %r6:%r7
90 bisw2 %ap,%r0 # restore the sign bit
92 ret # error less than 0.667 ulps
95 B : .long 721142941 # (86-0.03306235651)*(2^23)
96 C : .float 0f0.5428571429 # 19/35
97 D : .float 0f-0.7053061224 # -864/1225
98 E : .float 0f1.414285714 # 99/70
99 F : .float 0f1.607142857 # 45/28
100 G : .float 0f0.3571428571 # 5/14