1vexp_(3MVEC) Vector Math Library Functions vexp_(3MVEC)
2
6 vexp_, vexpf_ - vector exponential functions
7
9 cc [ flag... ] file... -lmvec [ library... ]
10 void vexp_(int *n, double * restrict x, int *stridex,
12 double * restrict y, int *stridey);
13 void vexpf_(int *n, float * restrict x, int *stridex,
16 float * restrict y, int *stridey);
17
20 These functions evaluate the function exp(x) for an entire vector of
21 values at once. The first parameter specifies the number of values to
22 compute. Subsequent parameters specify the argument and result vectors.
23 Each vector is described by a pointer to the first element and a
24 stride, which is the increment between successive elements.
25 Specifically, vexp_(n, x, sx, y, sy) computes y[i * *sy] = exp(x[i *
28 *sx]) for each i = 0, 1, ..., *n - 1. The vexpf_() function performs
29 the same computation for single precision data.
30 These functions are not guaranteed to deliver results that are identi‐
33 cal to the results of the exp(3M) functions given the same arguments.
34 Non-exceptional results, however, are accurate to within a unit in the
35 last place.
36
38 The element count *n must be greater than zero. The strides for the
39 argument and result arrays can be arbitrary integers, but the arrays
40 themselves must not be the same or overlap. A zero stride effectively
41 collapses an entire vector into a single element. A negative stride
42 causes a vector to be accessed in descending memory order, but note
43 that the corresponding pointer must still point to the first element of
44 the vector to be used; if the stride is negative, this will be the
45 highest-addressed element in memory. This convention differs from the
46 Level 1 BLAS, in which array parameters always refer to the lowest-
47 addressed element in memory even when negative increments are used.
48 These functions assume that the default round-to-nearest rounding
51 direction mode is in effect. On x86, these functions also assume that
52 the default round-to-64-bit rounding precision mode is in effect. The
53 result of calling a vector function with a non-default rounding mode in
54 effect is undefined.
55 On SPARC, the vexpf_() function delivers +0 rather than a subnormal
58 result for arguments in the range -103.2789 <= x <= -87.3365. Other‐
59 wise, these functions handle special cases and exceptions in the same
60 way as the exp() functions when c99 MATHERREXCEPT conventions are in
61 effect. See exp(3M) for the results for special cases.
62 An application wanting to check for exceptions should call feclearex‐
65 cept(FE_ALL_EXCEPT) before calling these functions. On return, if
66 fetestexcept(FE_INVALID | FE_DIVBYZERO | FE_OVERFLOW | FE_UNDERFLOW) is
67 non-zero, an exception has been raised. The application can then exam‐
68 ine the result or argument vectors for exceptional values. Some vector
69 functions can raise the inexact exception even if all elements of the
70 argument array are such that the numerical results are exact.
71
73 See attributes(5) for descriptions of the following attributes:
74 ┌─────────────────────────────┬─────────────────────────────┐
79 │ ATTRIBUTE TYPE │ ATTRIBUTE VALUE │
80 ├─────────────────────────────┼─────────────────────────────┤
81 │Interface Stability │Committed │
82 ├─────────────────────────────┼─────────────────────────────┤
83 │MT-Level │MT-Safe │
84 └─────────────────────────────┴─────────────────────────────┘
85
87 exp(3M), feclearexcept(3M), fetestexcept(3M), attributes(5)
88SunOS 5.11 14 Dec 2007 vexp_(3MVEC)