1 IeeeCC754 tests the correct implementation of the ieee floating
2 point constant evaluation.
4 Call make; make test to run the test.
6 Or do the test by hand:
8 How to call the driver program IeeeCC754
9 ========================================
13 IeeeCC754 -c {-s|-d|-l|-q|-m|{-e <int> -t <int> [-h]}}
14 -r {n|p|m|z} -i -n {i|o|x|z|u|tiny|nan|inf|snz}
17 IeeeCC754 -u -r {n|p|m|z} -i -n {i|o|x|z|u|tiny|nan|inf|snz}
20 IeeeCC754 -o <file> {-s|-d|-l|-q|-m|{-e <int> -t <int> [-h]}}
21 -r {n|p|m|z} -i -n {i|o|x|z|u|tiny|nan|inf|snz}
24 The required argument 'testfile' is a file of test vectors in extended
25 Coonen syntax (the default) or SUN-UCB format.
27 The options of the driver program can be subdivided in three categories,
30 (1) Options to specify the purpose of the run. Use precisely one of the
31 following three options:
32 -c : perform testing: file of test vectors is in Coonen syntax
33 -u : perform testing: file of test vectors is in SUN-UCB format
34 -o <file> : do not perform testing but only translate testdata from
35 Coonen syntax to hexadecimal SUN-UCB format and output
38 (2) Options to specify the precision and exponent range of the source (and
39 for some conversions also the destination) floating-point format, when
40 the file of test vectors is in extended Coonen syntax.
41 When the file of test vectors is in SUN-UCB format, the precision and
42 exponent range are specified in the testfile itself, and the options
43 below should be skipped.
44 -s : single precision (same as -e 8 -t 24 -h)
45 -d : double precision (same as -e 11 -t 53 -h)
46 -l : long double precision (same as -e 15 -t 64)
47 -q : quadruple precision (same as -e 15 -t 113 -h)
48 -m : 240 bit multiprecision (same as -e 15 -t 240)
49 -e <int> : provide <int> bits to represent exponent
50 -t <int> : provide <int> bits precision
51 -h : leading bit is hidden
52 -ds : single precision destination
53 -dd : double precision destination
54 -dl : long double precision destination
55 -dq : quadruple precision destination
56 -dm : 240 bit multiprecision destination
57 -de <int> : size of destination exponent
58 -dt <int> : destination precision
59 -dh : leading bit is hidden in destination
61 (3) Options to influence the actual testing phase (optional)
62 -r {n|p|m|z}: test only the specified rounding modes
63 -n {i|o|x|z|u|tiny|nan|inf|snz}: do not test the specified
64 exceptions, denormalized numbers, NaNs, signed
65 infinities or signed zeroes
66 -j {o|u|i|z}: jump/skip test vectors raising the overflow, underflow,
67 invalid or divide by zero exception
68 -ieee: test conversions only within range specified by IEEE
70 -f logfile: output log of testing to 'logfile'; the default value
71 for 'logfile' is ieee.log