1vpow_(3MVEC) Vector Math Library Functions vpow_(3MVEC)
2
6 vpow_, vpowf_ - vector power functions
7
9 cc [ flag... ] file... -lmvec [ library... ]
10 void vpow_(int *n, double * restrict x, int *stridex,
12 double * restrict y, int *stridey, double * restrict z,
13 int *stridez);
14 void vpowf_(int *n, float * restrict x, int *stridex,
17 float * restrict y, int *stridey, float * restrict z,
18 int *stridez);
19
22 These functions evaluate the function pow(x, y) for an entire vector of
23 values at once. The first parameter specifies the number of values to
24 compute. Subsequent parameters specify the argument and result vectors.
25 Each vector is described by a pointer to the first element and a
26 stride, which is the increment between successive elements.
27 Specifically, vpow_(n, x, sx, y, sy, z, sz) computes z[i * *sz] =
30 pow(x[i * *sx], y[i * *sy]) for each i = 0, 1, ..., *n - 1. The
31 vpowf_() function performs the same computation for single precision
32 data.
33 These functions are not guaranteed to deliver results that are identi‐
36 cal to the results of the pow(3M) functions given the same arguments.
37 Non-exceptional results, however, are accurate to within a unit in the
38 last place.
39
41 The element count *n must be greater than zero. The strides for the
42 argument and result arrays can be arbitrary integers, but the arrays
43 themselves must not be the same or overlap. A zero stride effectively
44 collapses an entire vector into a single element. A negative stride
45 causes a vector to be accessed in descending memory order, but note
46 that the corresponding pointer must still point to the first element of
47 the vector to be used; if the stride is negative, this will be the
48 highest-addressed element in memory. This convention differs from the
49 Level 1 BLAS, in which array parameters always refer to the lowest-
50 addressed element in memory even when negative increments are used.
51 These functions assume that the default round-to-nearest rounding
54 direction mode is in effect. On x86, these functions also assume that
55 the default round-to-64-bit rounding precision mode is in effect. The
56 result of calling a vector function with a non-default rounding mode in
57 effect is undefined.
58 The results of these functions for special cases and exceptions match
61 that of the pow() functions when the latter are used in a program com‐
62 piled with the cc compiler driver (that is, not SUSv3-conforming) and
63 the expression (math_errhandling & MATH_ERREXCEPT) is non-zero. These
64 functions do not set errno. See pow(3M) for the results for special
65 cases.
66 An application wanting to check for exceptions should call feclearex‐
69 cept(FE_ALL_EXCEPT) before calling these functions. On return, if
70 fetestexcept(FE_INVALID | FE_DIVBYZERO | FE_OVERFLOW | FE_UNDERFLOW) is
71 non-zero, an exception has been raised. The application can then exam‐
72 ine the result or argument vectors for exceptional values. Some vector
73 functions can raise the inexact exception even if all elements of the
74 argument array are such that the numerical results are exact.
75
77 See attributes(5) for descriptions of the following attributes:
78 ┌─────────────────────────────┬─────────────────────────────┐
83 │ ATTRIBUTE TYPE │ ATTRIBUTE VALUE │
84 ├─────────────────────────────┼─────────────────────────────┤
85 │Interface Stability │Committed │
86 ├─────────────────────────────┼─────────────────────────────┤
87 │MT-Level │MT-Safe │
88 └─────────────────────────────┴─────────────────────────────┘
89
91 pow(3M), feclearexcept(3M), fetestexcept(3M), attributes(5)
92SunOS 5.11 16 Jan 2009 vpow_(3MVEC)