1 This directory contains source for several test programs:
3 dt is for conversion to/from double; it permits input of pairs of
4 32-bit hex integers as #hhhhhhhh hhhhhhhh (i.e., the initial '#'
5 indicates hex input). No initial # ==> decimal input.
6 After the input number is an optional : mode ndigits
7 (colon, and decimal integers for parameters "mode" and "ndigits"
10 Qtest, ddtest, dtest, ftest, xLtest and xtest are for conversion to/from
12 f IEEE single precision
13 d IEEE double precision
14 xL IEEE extended precision, as on Motorola 680x0 chips
15 x IEEE extended precision, as on Intel 80x87 chips or
16 software emulation of Motorola 680x0 chips
17 Q quad precision, as on Sun Sparc chips
18 dd double double, pairs of IEEE double numbers
19 whose sum is the desired value
21 They're all similar, except for the precision. They test both
22 directed roundings and interval input (the strtoI* routines).
23 Lines that begin with "r" specify or interrogate the desired rounding
31 These are the FPI_Round_* values in gdota.h. The "r" value is sticky:
32 it stays in effect til changed. To change the value, give a line that
33 starts with r followed by 0, 1, 2, or 3. To check the value, give "r"
36 Lines that begin with n followed by a number specify the ndig
37 argument for subsequent calls to the relevant g_*fmt routine.
39 Lines that start with # followed by the appropriate number of
40 hexadecimal strings (see the comments) give the big-endian
41 internal representation of the desired number.
43 When routines Qtest, xLtest, and xtest are used on machines whose
44 long double is of type "quad" (for Qtest) or "extended" (for x*test),
45 they try to print with %Lg as another way to show binary values.
47 Program ddtest also accepts (white-space separated) pairs of decimal
48 input numbers; it converts both with strtod and feeds the result
51 Program dItest exercises strtodI and strtoId.
53 Programs dItestsi and ddtestsi are for testing the sudden-underflow
54 logic (on double and double-double conversions).
56 Program strtodt tests strtod on some hard cases (in file testnos3)
57 posted by Fred Tydeman to comp.arch.arithmetic on 26 Feb. 1996.
58 To get correct results on Intel (x86) systems, the rounding precision
59 must be set to 53 bits. This can be done, e.g., by invoking
60 fpinit_ASL(), whose source appears in
61 http://www.netlib.org/ampl/solvers/fpinit.c .
63 The obad directory shows results expected on (at least some) Intel x86
64 Linux systems and may not be relevant to other systems.
66 You can optionally compile getround.c with -DHonor_FLT_ROUNDS
67 to manually test strtof, strtod, etc., using fegetround().
68 You can also or alternatively compile getround.c with
69 -DUSE_MY_LOCALE (when ../gdtoa.a is compiled with -DUSE_LOCALE)
70 to test multi-byte decimal points.
72 If in the parent directory, you have sucessfully invoked "make Printf"
73 to add a "printf" (called Printf and accessed via ../stdio1.h), then
74 here you can use "make pf_test" and (if you have both a 64-bit long
75 double and a 113-bit "quad" double type) "make pf_testLq" for a brief
76 test of %g and %a variants in Printf.
78 These are simple test programs, not meant for exhaustive testing,
79 but for manually testing "interesting" cases. Paxson's testbase
80 is good for more exhaustive testing, in part with random inputs.
81 See ftp://ftp.ee.lbl.gov/testbase-report.ps.Z .