v/vlib/math/bits.v

64 lines
2.2 KiB
V
Raw Permalink Normal View History

2022-01-04 10:21:08 +01:00
// Copyright (c) 2019-2022 Alexander Medvednikov. All rights reserved.
// Use of this source code is governed by an MIT license
// that can be found in the LICENSE file.
module math
const (
uvnan = u64(0x7FF8000000000001)
uvinf = u64(0x7FF0000000000000)
uvneginf = u64(0xFFF0000000000000)
uvone = u64(0x3FF0000000000000)
mask = 0x7FF
shift = 64 - 11 - 1
bias = 1023
normalize_smallest_mask = u64(u64(1) << 52)
sign_mask = u64(0x8000000000000000) // (u64(1) << 63)
frac_mask = u64((u64(1) << u64(shift)) - u64(1))
)
// inf returns positive infinity if sign >= 0, negative infinity if sign < 0.
pub fn inf(sign int) f64 {
v := if sign >= 0 { math.uvinf } else { math.uvneginf }
return f64_from_bits(v)
}
// nan returns an IEEE 754 ``not-a-number'' value.
pub fn nan() f64 {
return f64_from_bits(math.uvnan)
}
// is_nan reports whether f is an IEEE 754 ``not-a-number'' value.
pub fn is_nan(f f64) bool {
// IEEE 754 says that only NaNs satisfy f != f.
// To avoid the floating-point hardware, could use:
// x := f64_bits(f);
// return u32(x>>shift)&mask == mask && x != uvinf && x != uvneginf
return f != f
}
// is_inf reports whether f is an infinity, according to sign.
// If sign > 0, is_inf reports whether f is positive infinity.
// If sign < 0, is_inf reports whether f is negative infinity.
// If sign == 0, is_inf reports whether f is either infinity.
pub fn is_inf(f f64, sign int) bool {
// Test for infinity by comparing against maximum float.
// To avoid the floating-point hardware, could use:
// x := f64_bits(f);
// return sign >= 0 && x == uvinf || sign <= 0 && x == uvneginf;
return (sign >= 0 && f > max_f64) || (sign <= 0 && f < -max_f64)
}
pub fn is_finite(f f64) bool {
return !is_nan(f) && !is_inf(f, 0)
}
// normalize returns a normal number y and exponent exp
// satisfying x == y × 2**exp. It assumes x is finite and non-zero.
pub fn normalize(x f64) (f64, int) {
smallest_normal := 2.2250738585072014e-308 // 2**-1022
if abs(x) < smallest_normal {
return x * math.normalize_smallest_mask, -52
}
return x, 0
}