llvm-project/libc/utils/FPUtil/NearestIntegerOperations.h
Siva Chandra Reddy 139018265b [libc] Add implementations long double fabsl and truncl functions.
Current implementations of single precision and double precision
floating point operations operate on bits of the integer type of
same size. The code made use of magic masks which were listed as
literal integer values. This is not possible in the case of long
double type as the mantissa of quad-precision long double type used
on non-x86 architectures is wider that the widest integer type for
which we can list literal values. So, in this patch, to avoid
using magic masks specified with literal values, we use packed
bit-field struct types and let the compiler generate the masks.
This new scheme allows us to implement long double flavors of the
various floating point operations. To keep the size of the patch
small, only the implementations of fabs and trunc have been
switched to the new scheme. In following patches, all exisiting
implementations will be switched to the new scheme.

Reviewers: asteinhauser

Differential Revision: https://reviews.llvm.org/D82036
2020-06-18 11:08:26 -07:00

167 lines
4.6 KiB
C++

//===-- Nearest integer floating-point operations ---------------*- C++ -*-===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
#ifndef LLVM_LIBC_UTILS_FPUTIL_NEAREST_INTEGER_OPERATIONS_H
#define LLVM_LIBC_UTILS_FPUTIL_NEAREST_INTEGER_OPERATIONS_H
#include "ClassificationFunctions.h"
#include "FPBits.h"
#include "FloatOperations.h"
#include "FloatProperties.h"
#include "utils/CPP/TypeTraits.h"
namespace __llvm_libc {
namespace fputil {
template <typename T,
cpp::EnableIfType<cpp::IsFloatingPointType<T>::Value, int> = 0>
static inline T trunc(T x) {
FPBits<T> bits(x);
// If x is infinity or NaN, return it.
// If it is zero also we should return it as is, but the logic
// later in this function takes care of it. But not doing a zero
// check, we improve the run time of non-zero values.
if (bits.isInfOrNaN())
return x;
int exponent = bits.getExponent();
// If the exponent is greater than the most negative mantissa
// exponent, then x is already an integer.
if (exponent >= static_cast<int>(MantissaWidth<T>::value))
return x;
// If the exponent is such that abs(x) is less than 1, then return 0.
if (exponent <= -1) {
if (bits.sign)
return T(-0.0);
else
return T(0.0);
}
int trimSize = MantissaWidth<T>::value - exponent;
bits.mantissa = (bits.mantissa >> trimSize) << trimSize;
return bits;
}
template <typename T,
cpp::EnableIfType<cpp::IsFloatingPointType<T>::Value, int> = 0>
static inline T ceil(T x) {
using Properties = FloatProperties<T>;
using BitsType = typename FloatProperties<T>::BitsType;
BitsType bits = valueAsBits(x);
// If x is infinity NaN or zero, return it.
if (bitsAreInfOrNaN(bits) || bitsAreZero(bits))
return x;
bool isNeg = bits & Properties::signMask;
int exponent = getExponentFromBits(bits);
// If the exponent is greater than the most negative mantissa
// exponent, then x is already an integer.
if (exponent >= static_cast<int>(Properties::mantissaWidth))
return x;
if (exponent <= -1) {
if (isNeg)
return T(-0.0);
else
return T(1.0);
}
uint32_t trimSize = Properties::mantissaWidth - exponent;
// If x is already an integer, return it.
if ((bits << (Properties::bitWidth - trimSize)) == 0)
return x;
BitsType truncBits = (bits >> trimSize) << trimSize;
T truncValue = valueFromBits(truncBits);
// If x is negative, the ceil operation is equivalent to the trunc operation.
if (isNeg)
return truncValue;
return truncValue + T(1.0);
}
template <typename T,
cpp::EnableIfType<cpp::IsFloatingPointType<T>::Value, int> = 0>
static inline T floor(T x) {
auto bits = valueAsBits(x);
if (FloatProperties<T>::signMask & bits) {
return -ceil(-x);
} else {
return trunc(x);
}
}
template <typename T,
cpp::EnableIfType<cpp::IsFloatingPointType<T>::Value, int> = 0>
static inline T round(T x) {
using Properties = FloatProperties<T>;
using BitsType = typename FloatProperties<T>::BitsType;
BitsType bits = valueAsBits(x);
// If x is infinity, NaN or zero, return it.
if (bitsAreInfOrNaN(bits) || bitsAreZero(bits))
return x;
bool isNeg = bits & Properties::signMask;
int exponent = getExponentFromBits(bits);
// If the exponent is greater than the most negative mantissa
// exponent, then x is already an integer.
if (exponent >= static_cast<int>(Properties::mantissaWidth))
return x;
if (exponent == -1) {
// Absolute value of x is greater than equal to 0.5 but less than 1.
if (isNeg)
return T(-1.0);
else
return T(1.0);
}
if (exponent <= -2) {
// Absolute value of x is less than 0.5.
if (isNeg)
return T(-0.0);
else
return T(0.0);
}
uint32_t trimSize = Properties::mantissaWidth - exponent;
// If x is already an integer, return it.
if ((bits << (Properties::bitWidth - trimSize)) == 0)
return x;
BitsType truncBits = (bits >> trimSize) << trimSize;
T truncValue = valueFromBits(truncBits);
if ((bits & (BitsType(1) << (trimSize - 1))) == 0) {
// Franctional part is less than 0.5 so round value is the
// same as the trunc value.
return truncValue;
}
if (isNeg)
return truncValue - T(1.0);
else
return truncValue + T(1.0);
}
} // namespace fputil
} // namespace __llvm_libc
#endif // LLVM_LIBC_UTILS_FPUTIL_NEAREST_INTEGER_OPERATIONS_H