Skip to content

[libc][NFC] Use the Sign type for DyadicFloat #78577

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Jan 18, 2024
Merged

[libc][NFC] Use the Sign type for DyadicFloat #78577

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions libc/src/__support/FPUtil/dyadic_float.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -49,8 +49,8 @@ template <size_t Bits> struct DyadicFloat {
normalize();
}

constexpr DyadicFloat(bool s, int e, MantissaType m)
: sign(s ? Sign::NEG : Sign::POS), exponent(e), mantissa(m) {
constexpr DyadicFloat(Sign s, int e, MantissaType m)
: sign(s), exponent(e), mantissa(m) {
normalize();
}

Expand Down
3 changes: 2 additions & 1 deletion libc/src/__support/str_to_float.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -510,6 +510,7 @@ clinger_fast_path(ExpandedFloat<T> init_num,
RoundDirection round = RoundDirection::Nearest) {
using FPBits = typename fputil::FPBits<T>;
using StorageType = typename FPBits::StorageType;
using Sign = fputil::Sign;

StorageType mantissa = init_num.mantissa;
int32_t exp10 = init_num.exponent;
Expand All @@ -522,7 +523,7 @@ clinger_fast_path(ExpandedFloat<T> init_num,
T float_mantissa;
if constexpr (cpp::is_same_v<StorageType, cpp::UInt<128>>) {
float_mantissa = static_cast<T>(fputil::DyadicFloat<128>(
false, 0,
Sign::POS, 0,
fputil::DyadicFloat<128>::MantissaType(
{uint64_t(mantissa), uint64_t(mantissa >> 64)})));
} else {
Expand Down
21 changes: 13 additions & 8 deletions libc/src/math/generic/exp.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,6 +30,7 @@ namespace LIBC_NAMESPACE {
using fputil::DoubleDouble;
using fputil::TripleDouble;
using Float128 = typename fputil::DyadicFloat<128>;
using Sign = fputil::Sign;

// log2(e)
constexpr double LOG2_E = 0x1.71547652b82fep+0;
Expand Down Expand Up @@ -99,14 +100,18 @@ Float128 poly_approx_f128(const Float128 &dx) {
using MType = typename Float128::MantissaType;

constexpr Float128 COEFFS_128[]{
{false, -127, MType({0, 0x8000000000000000})}, // 1.0
{false, -127, MType({0, 0x8000000000000000})}, // 1.0
{false, -128, MType({0, 0x8000000000000000})}, // 0.5
{false, -130, MType({0xaaaaaaaaaaaaaaab, 0xaaaaaaaaaaaaaaaa})}, // 1/6
{false, -132, MType({0xaaaaaaaaaaaaaaab, 0xaaaaaaaaaaaaaaaa})}, // 1/24
{false, -134, MType({0x8888888888888889, 0x8888888888888888})}, // 1/120
{false, -137, MType({0x60b60b60b60b60b6, 0xb60b60b60b60b60b})}, // 1/720
{false, -140, MType({0x00d00d00d00d00d0, 0xd00d00d00d00d00d})}, // 1/5040
{Sign::POS, -127, MType({0, 0x8000000000000000})}, // 1.0
{Sign::POS, -127, MType({0, 0x8000000000000000})}, // 1.0
{Sign::POS, -128, MType({0, 0x8000000000000000})}, // 0.5
{Sign::POS, -130, MType({0xaaaaaaaaaaaaaaab, 0xaaaaaaaaaaaaaaaa})}, // 1/6
{Sign::POS, -132,
MType({0xaaaaaaaaaaaaaaab, 0xaaaaaaaaaaaaaaaa})}, // 1/24
{Sign::POS, -134,
MType({0x8888888888888889, 0x8888888888888888})}, // 1/120
{Sign::POS, -137,
MType({0x60b60b60b60b60b6, 0xb60b60b60b60b60b})}, // 1/720
{Sign::POS, -140,
MType({0x00d00d00d00d00d0, 0xd00d00d00d00d00d})}, // 1/5040
};

Float128 p = fputil::polyeval(dx, COEFFS_128[0], COEFFS_128[1], COEFFS_128[2],
Expand Down
17 changes: 9 additions & 8 deletions libc/src/math/generic/exp10.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,6 +30,7 @@ namespace LIBC_NAMESPACE {
using fputil::DoubleDouble;
using fputil::TripleDouble;
using Float128 = typename fputil::DyadicFloat<128>;
using Sign = fputil::Sign;

// log2(10)
constexpr double LOG2_10 = 0x1.a934f0979a371p+1;
Expand Down Expand Up @@ -101,14 +102,14 @@ Float128 poly_approx_f128(const Float128 &dx) {
using MType = typename Float128::MantissaType;

constexpr Float128 COEFFS_128[]{
{false, -127, MType({0, 0x8000000000000000})}, // 1.0
{false, -126, MType({0xea56d62b82d30a2d, 0x935d8dddaaa8ac16})},
{false, -126, MType({0x80a99ce75f4d5bdb, 0xa9a92639e753443a})},
{false, -126, MType({0x6a4f9d7dbf6c9635, 0x82382c8ef1652304})},
{false, -124, MType({0x345787019216c7af, 0x12bd7609fd98c44c})},
{false, -127, MType({0xcc41ed7e0d27aee5, 0x450a7ff47535d889})},
{false, -130, MType({0x8326bb91a6e7601d, 0xd3f6b844702d636b})},
{false, -130, MType({0xfa7b46df314112a9, 0x45b937f0d05bb1cd})},
{Sign::POS, -127, MType({0, 0x8000000000000000})}, // 1.0
{Sign::POS, -126, MType({0xea56d62b82d30a2d, 0x935d8dddaaa8ac16})},
{Sign::POS, -126, MType({0x80a99ce75f4d5bdb, 0xa9a92639e753443a})},
{Sign::POS, -126, MType({0x6a4f9d7dbf6c9635, 0x82382c8ef1652304})},
{Sign::POS, -124, MType({0x345787019216c7af, 0x12bd7609fd98c44c})},
{Sign::POS, -127, MType({0xcc41ed7e0d27aee5, 0x450a7ff47535d889})},
{Sign::POS, -130, MType({0x8326bb91a6e7601d, 0xd3f6b844702d636b})},
{Sign::POS, -130, MType({0xfa7b46df314112a9, 0x45b937f0d05bb1cd})},
};

Float128 p = fputil::polyeval(dx, COEFFS_128[0], COEFFS_128[1], COEFFS_128[2],
Expand Down
17 changes: 9 additions & 8 deletions libc/src/math/generic/exp2.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,6 +30,7 @@ namespace LIBC_NAMESPACE {
using fputil::DoubleDouble;
using fputil::TripleDouble;
using Float128 = typename fputil::DyadicFloat<128>;
using Sign = fputil::Sign;

// Error bounds:
// Errors when using double precision.
Expand Down Expand Up @@ -90,14 +91,14 @@ Float128 poly_approx_f128(const Float128 &dx) {
using MType = typename Float128::MantissaType;

constexpr Float128 COEFFS_128[]{
{false, -127, MType({0, 0x8000000000000000})}, // 1.0
{false, -128, MType({0xc9e3b39803f2f6af, 0xb17217f7d1cf79ab})},
{false, -128, MType({0xde2d60dd9c9a1d9f, 0x3d7f7bff058b1d50})},
{false, -132, MType({0x9d3b15d9e7fb6897, 0xe35846b82505fc59})},
{false, -134, MType({0x184462f6bcd2b9e7, 0x9d955b7dd273b94e})},
{false, -137, MType({0x39ea1bb964c51a89, 0xaec3ff3c53398883})},
{false, -138, MType({0x842c53418fa8ae61, 0x2861225f345c396a})},
{false, -144, MType({0x7abeb5abd5ad2079, 0xffe5fe2d109a319d})},
{Sign::POS, -127, MType({0, 0x8000000000000000})}, // 1.0
{Sign::POS, -128, MType({0xc9e3b39803f2f6af, 0xb17217f7d1cf79ab})},
{Sign::POS, -128, MType({0xde2d60dd9c9a1d9f, 0x3d7f7bff058b1d50})},
{Sign::POS, -132, MType({0x9d3b15d9e7fb6897, 0xe35846b82505fc59})},
{Sign::POS, -134, MType({0x184462f6bcd2b9e7, 0x9d955b7dd273b94e})},
{Sign::POS, -137, MType({0x39ea1bb964c51a89, 0xaec3ff3c53398883})},
{Sign::POS, -138, MType({0x842c53418fa8ae61, 0x2861225f345c396a})},
{Sign::POS, -144, MType({0x7abeb5abd5ad2079, 0xffe5fe2d109a319d})},
};

Float128 p = fputil::polyeval(dx, COEFFS_128[0], COEFFS_128[1], COEFFS_128[2],
Expand Down
21 changes: 13 additions & 8 deletions libc/src/math/generic/expm1.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,6 +38,7 @@ namespace LIBC_NAMESPACE {
using fputil::DoubleDouble;
using fputil::TripleDouble;
using Float128 = typename fputil::DyadicFloat<128>;
using Sign = fputil::Sign;

// log2(e)
constexpr double LOG2_E = 0x1.71547652b82fep+0;
Expand Down Expand Up @@ -109,13 +110,17 @@ Float128 poly_approx_f128(const Float128 &dx) {
using MType = typename Float128::MantissaType;

constexpr Float128 COEFFS_128[]{
{false, -127, MType({0, 0x8000000000000000})}, // 1.0
{false, -128, MType({0, 0x8000000000000000})}, // 0.5
{false, -130, MType({0xaaaaaaaaaaaaaaab, 0xaaaaaaaaaaaaaaaa})}, // 1/6
{false, -132, MType({0xaaaaaaaaaaaaaaab, 0xaaaaaaaaaaaaaaaa})}, // 1/24
{false, -134, MType({0x8888888888888889, 0x8888888888888888})}, // 1/120
{false, -137, MType({0x60b60b60b60b60b6, 0xb60b60b60b60b60b})}, // 1/720
{false, -140, MType({0x00d00d00d00d00d0, 0xd00d00d00d00d00d})}, // 1/5040
{Sign::POS, -127, MType({0, 0x8000000000000000})}, // 1.0
{Sign::POS, -128, MType({0, 0x8000000000000000})}, // 0.5
{Sign::POS, -130, MType({0xaaaaaaaaaaaaaaab, 0xaaaaaaaaaaaaaaaa})}, // 1/6
{Sign::POS, -132,
MType({0xaaaaaaaaaaaaaaab, 0xaaaaaaaaaaaaaaaa})}, // 1/24
{Sign::POS, -134,
MType({0x8888888888888889, 0x8888888888888888})}, // 1/120
{Sign::POS, -137,
MType({0x60b60b60b60b60b6, 0xb60b60b60b60b60b})}, // 1/720
{Sign::POS, -140,
MType({0x00d00d00d00d00d0, 0xd00d00d00d00d00d})}, // 1/5040
};

Float128 p = fputil::polyeval(dx, COEFFS_128[0], COEFFS_128[1], COEFFS_128[2],
Expand Down Expand Up @@ -165,7 +170,7 @@ Float128 expm1_f128(double x, double kd, int idx1, int idx2) {
Float128 exp_mid = fputil::quick_mul(exp_mid1, exp_mid2);

int hi = static_cast<int>(kd) >> 12;
Float128 minus_one{true, -127 - hi, MType({0, 0x8000000000000000})};
Float128 minus_one{Sign::NEG, -127 - hi, MType({0, 0x8000000000000000})};

Float128 exp_mid_m1 = fputil::quick_add(exp_mid, minus_one);

Expand Down
Loading