Debian Codegraph @

mp/readme

raw
1 README for multiprecision arithmetic directory.
2 README by W. Schelter
3
4 The files in this directory
5
6 gencom.h mp.s sparc.s
7 erreurs.h genport.h
8
9 are from the PARI distribution version 1.34 written by C. Batut, D.
10 Bernardi, H. Cohen and M. Olivier.
11
12 The file mp.c from the 1.34 distribution has been divided into
13 mpi.c: The functions benefiting from assembler or assembler macros.
14 mp2.c: Additional integer arithmetic routines.
15
16 In addition the functions in mpi.c have been somewhat rewritten by
17 Schelter, to improve the efficiency on machines unable to use the 68k
18 assembler in the pari file mp.s. By using gcc and assembler macros,
19 we were able to equal the speed of the excellent pari assembler, on
20 multiply and addition of 100 word bignums. This distribution
21 contains .s files produced by gcc for machines where this has
22 been available. For other machines the functions divul3
23 and mulul3 may be provided as assembler functions, or they may
24 use the code in the libmport.a which is plain C.
25
26 mp_mulul3.c: 64 bit multiply in C
27 mp_divul3.c: 64 bit multiply in C
28 mp_bfffo.c: position of first non zero bit in C
29
30 mpi-*.s: are assembler produced by gcc for several machines.
31
32 Full sources for PARI/GP are available by anonymous ftp from (Internet
33 number 192.33.148.32).
34
35 Authors address:
36
37 Prof. Henri COHEN (re: PARI)
38 UFR de Mathematiques et Informatique
39 Universite Bordeaux I
40 351 Cours de la Liberation
41 33405 TALENCE CEDEX
42 FRANCE
43
44 e-mail: [email protected] (Internet number 192.33.148.32)
45
46 The integration of the pari code into AKCL was done by W. Schelter.
47
48 ---------------------------------------------------------------------------
49 Excerpt from letter from Henri Cohen giving us permission to
50 redistribute this code with AKCL.
51
52 Received-Date: Fri, 25 Jan 91 12:38:02 -0600
53 Date: Fri, 25 Jan 91 19:17:34 +0100
54 From: [email protected] (Systeme PARI)
55 To: [email protected]
56 Subject: Re: AKCL
57
58 Thank you for the clarifications.
59
60 1) -2^31: it was mainly a matter of convenience to exclude that value, so that
61 in effect I don't know which programs should be affected if one allows it. If
62 necessary, I can look into it.
63
64 2) Yes, go ahead means that you may modify and rearrange things in your own
65 way,
66 and redistribute it, of course keeping us informed. On the other hand, please
67 tell us as soon as possible of any bug that you discover that we should
68 eliminate from the main system itself.
69
70 ..
71
72 Sincerely,
73
74 Henri Cohen
75
76
77
78
79
80
81
82
83
84
85