-
Notifications
You must be signed in to change notification settings - Fork 10.8k
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.
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
[reland][libc][NFC] Use user defined literals to build 128 and 256 bit constants #81998
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Contributor
gchatelet
commented
Feb 16, 2024
•
edited
edited
- [reland][libc][NFC] Use user defined literals to build 128 and 256 bit constants #81835
- Fix for platforms where uint64_t is not available
…t constants (llvm#81835) This is a reland of llvm#81746
@llvm/pr-subscribers-libc Author: Guillaume Chatelet (gchatelet) Changes
Patch is 386.24 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/81998.diff 17 Files Affected:
diff --git a/libc/src/__support/UInt.h b/libc/src/__support/UInt.h
index f2084300d14208..ae1fe7aaa18287 100644
--- a/libc/src/__support/UInt.h
+++ b/libc/src/__support/UInt.h
@@ -867,14 +867,29 @@ struct BigInt {
LIBC_INLINE const WordType *data() const { return val; }
};
+namespace internal {
+// We default BigInt's WordType to 'uint64_t' or 'uint32_t' depending on type
+// availability.
+template <size_t Bits>
+struct WordTypeSelector : cpp::type_identity<
+#if defined(UINT64_MAX)
+ uint64_t
+#else
+ uint32_t
+#endif
+ > {
+};
+// Except if we request 32 bits explicitly.
+template <> struct WordTypeSelector<32> : cpp::type_identity<uint32_t> {};
+template <size_t Bits>
+using WordTypeSelectorT = typename WordTypeSelector<Bits>::type;
+} // namespace internal
+
template <size_t Bits>
-using UInt =
- typename cpp::conditional_t<Bits == 32, BigInt<32, false, uint32_t>,
- BigInt<Bits, false, uint64_t>>;
+using UInt = BigInt<Bits, false, internal::WordTypeSelectorT<Bits>>;
template <size_t Bits>
-using Int = typename cpp::conditional_t<Bits == 32, BigInt<32, true, uint32_t>,
- BigInt<Bits, true, uint64_t>>;
+using Int = BigInt<Bits, true, internal::WordTypeSelectorT<Bits>>;
// Provides limits of U/Int<128>.
template <> class numeric_limits<UInt<128>> {
diff --git a/libc/src/__support/integer_literals.h b/libc/src/__support/integer_literals.h
index c8e965c1a03a05..08bc6973b1582b 100644
--- a/libc/src/__support/integer_literals.h
+++ b/libc/src/__support/integer_literals.h
@@ -22,19 +22,19 @@
namespace LIBC_NAMESPACE {
LIBC_INLINE constexpr uint8_t operator""_u8(unsigned long long value) {
- return value;
+ return static_cast<uint8_t>(value);
}
LIBC_INLINE constexpr uint16_t operator""_u16(unsigned long long value) {
- return value;
+ return static_cast<uint16_t>(value);
}
LIBC_INLINE constexpr uint32_t operator""_u32(unsigned long long value) {
- return value;
+ return static_cast<uint32_t>(value);
}
LIBC_INLINE constexpr uint64_t operator""_u64(unsigned long long value) {
- return value;
+ return static_cast<uint64_t>(value);
}
namespace internal {
@@ -64,6 +64,7 @@ template <typename T, int base> struct DigitBuffer {
: 0;
LIBC_INLINE_VAR static constexpr size_t MAX_DIGITS =
sizeof(T) * CHAR_BIT / BITS_PER_DIGIT;
+ LIBC_INLINE_VAR static constexpr uint8_t INVALID_DIGIT = 255;
uint8_t digits[MAX_DIGITS] = {};
size_t size = 0;
@@ -74,7 +75,7 @@ template <typename T, int base> struct DigitBuffer {
}
// Returns the digit for a particular character.
- // Returns 255 if the character is invalid.
+ // Returns INVALID_DIGIT if the character is invalid.
LIBC_INLINE static constexpr uint8_t get_digit_value(const char c) {
const auto to_lower = [](char c) { return c | 32; };
const auto is_digit = [](char c) { return c >= '0' && c <= '9'; };
@@ -82,10 +83,10 @@ template <typename T, int base> struct DigitBuffer {
return ('a' <= c && c <= 'z') || ('A' <= c && c <= 'Z');
};
if (is_digit(c))
- return c - '0';
+ return static_cast<uint8_t>(c - '0');
if (base > 10 && is_alpha(c))
- return to_lower(c) - 'a' + 10;
- return 255;
+ return static_cast<uint8_t>(to_lower(c) - 'a' + 10);
+ return INVALID_DIGIT;
}
// Adds a single character to this buffer.
@@ -93,7 +94,7 @@ template <typename T, int base> struct DigitBuffer {
if (c == '\'')
return; // ' is valid but not taken into account.
const uint8_t value = get_digit_value(c);
- if (value == 255 || size >= MAX_DIGITS) {
+ if (value == INVALID_DIGIT || size >= MAX_DIGITS) {
// During constant evaluation `__builtin_unreachable` will halt the
// compiler as it is not executable. This is preferable over `assert` that
// will only trigger in debug mode. Also we can't use `static_assert`
@@ -114,32 +115,32 @@ template <typename T> struct Parser {
}
};
-// Specialization for cpp::BigInt<N, false, uint64_t>.
+// Specialization for cpp::UInt<N>.
// Because this code runs at compile time we try to make it efficient. For
// binary and hexadecimal formats we read digits by chunks of 64 bits and
// produce the BigInt internal representation direcly. For decimal numbers we
// go the slow path and use slower BigInt arithmetic.
-template <size_t N>
-struct Parser<LIBC_NAMESPACE::cpp::BigInt<N, false, uint64_t>> {
- using UIntT = cpp::BigInt<N, false, uint64_t>;
+template <size_t N> struct Parser<LIBC_NAMESPACE::cpp::UInt<N>> {
+ using UIntT = cpp::UInt<N>;
template <int base> static constexpr UIntT parse(const char *str) {
const DigitBuffer<UIntT, base> buffer(str);
if constexpr (base == 10) {
// Slow path, we sum and multiply BigInt for each digit.
return accumulate<UIntT>(base, buffer.digits, buffer.size);
} else {
- // Fast path, we consume blocks of uint64_t and creates the BigInt's
+ // Fast path, we consume blocks of WordType and creates the BigInt's
// internal representation directly.
- using U64ArrayT = cpp::array<uint64_t, UIntT::WORD_COUNT>;
- U64ArrayT array;
+ using WordArrayT = decltype(UIntT::val);
+ using WordType = typename WordArrayT::value_type;
+ WordArrayT array = {};
size_t size = buffer.size;
const uint8_t *digit_ptr = buffer.digits + size;
for (size_t i = 0; i < array.size(); ++i) {
- constexpr size_t U64_DIGITS = DigitBuffer<uint64_t, base>::MAX_DIGITS;
- const size_t chunk = size > U64_DIGITS ? U64_DIGITS : size;
+ constexpr size_t DIGITS = DigitBuffer<WordType, base>::MAX_DIGITS;
+ const size_t chunk = size > DIGITS ? DIGITS : size;
digit_ptr -= chunk;
size -= chunk;
- array[i] = accumulate<uint64_t>(base, digit_ptr, chunk);
+ array[i] = accumulate<WordType>(base, digit_ptr, chunk);
}
return UIntT(array);
}
diff --git a/libc/src/math/generic/CMakeLists.txt b/libc/src/math/generic/CMakeLists.txt
index 259ae1c2793439..07885ae6a3e242 100644
--- a/libc/src/math/generic/CMakeLists.txt
+++ b/libc/src/math/generic/CMakeLists.txt
@@ -622,6 +622,8 @@ add_entrypoint_object(
DEPENDS
.common_constants
.explogxf
+ libc.include.errno
+ libc.include.math
libc.src.__support.CPP.bit
libc.src.__support.CPP.optional
libc.src.__support.FPUtil.dyadic_float
@@ -632,10 +634,9 @@ add_entrypoint_object(
libc.src.__support.FPUtil.polyeval
libc.src.__support.FPUtil.rounding_mode
libc.src.__support.FPUtil.triple_double
+ libc.src.__support.integer_literals
libc.src.__support.macros.optimization
- libc.include.errno
libc.src.errno.errno
- libc.include.math
COMPILE_OPTIONS
-O3
)
@@ -672,6 +673,8 @@ add_entrypoint_object(
DEPENDS
.common_constants
.explogxf
+ libc.include.errno
+ libc.include.math
libc.src.__support.CPP.bit
libc.src.__support.CPP.optional
libc.src.__support.FPUtil.dyadic_float
@@ -682,10 +685,9 @@ add_entrypoint_object(
libc.src.__support.FPUtil.polyeval
libc.src.__support.FPUtil.rounding_mode
libc.src.__support.FPUtil.triple_double
+ libc.src.__support.integer_literals
libc.src.__support.macros.optimization
- libc.include.errno
libc.src.errno.errno
- libc.include.math
COMPILE_OPTIONS
-O3
)
@@ -731,6 +733,8 @@ add_entrypoint_object(
DEPENDS
.common_constants
.explogxf
+ libc.include.errno
+ libc.include.math
libc.src.__support.CPP.bit
libc.src.__support.CPP.optional
libc.src.__support.FPUtil.dyadic_float
@@ -741,10 +745,9 @@ add_entrypoint_object(
libc.src.__support.FPUtil.polyeval
libc.src.__support.FPUtil.rounding_mode
libc.src.__support.FPUtil.triple_double
+ libc.src.__support.integer_literals
libc.src.__support.macros.optimization
- libc.include.errno
libc.src.errno.errno
- libc.include.math
COMPILE_OPTIONS
-O3
)
@@ -791,6 +794,8 @@ add_entrypoint_object(
DEPENDS
.common_constants
.explogxf
+ libc.include.errno
+ libc.include.math
libc.src.__support.CPP.bit
libc.src.__support.CPP.optional
libc.src.__support.FPUtil.dyadic_float
@@ -801,10 +806,9 @@ add_entrypoint_object(
libc.src.__support.FPUtil.polyeval
libc.src.__support.FPUtil.rounding_mode
libc.src.__support.FPUtil.triple_double
+ libc.src.__support.integer_literals
libc.src.__support.macros.optimization
- libc.include.errno
libc.src.errno.errno
- libc.include.math
COMPILE_OPTIONS
-O3
)
@@ -1074,12 +1078,13 @@ add_entrypoint_object(
DEPENDS
.common_constants
.log_range_reduction
+ libc.src.__support.FPUtil.double_double
+ libc.src.__support.FPUtil.dyadic_float
libc.src.__support.FPUtil.fenv_impl
libc.src.__support.FPUtil.fp_bits
libc.src.__support.FPUtil.multiply_add
libc.src.__support.FPUtil.polyeval
- libc.src.__support.FPUtil.double_double
- libc.src.__support.FPUtil.dyadic_float
+ libc.src.__support.integer_literals
libc.src.__support.macros.optimization
COMPILE_OPTIONS
-O3
@@ -1110,12 +1115,13 @@ add_entrypoint_object(
../log1p.h
DEPENDS
.common_constants
+ libc.src.__support.FPUtil.double_double
+ libc.src.__support.FPUtil.dyadic_float
libc.src.__support.FPUtil.fenv_impl
libc.src.__support.FPUtil.fp_bits
libc.src.__support.FPUtil.multiply_add
libc.src.__support.FPUtil.polyeval
- libc.src.__support.FPUtil.double_double
- libc.src.__support.FPUtil.dyadic_float
+ libc.src.__support.integer_literals
libc.src.__support.macros.optimization
COMPILE_OPTIONS
-O3
@@ -1148,12 +1154,13 @@ add_entrypoint_object(
DEPENDS
.common_constants
.log_range_reduction
+ libc.src.__support.FPUtil.double_double
+ libc.src.__support.FPUtil.dyadic_float
libc.src.__support.FPUtil.fenv_impl
libc.src.__support.FPUtil.fp_bits
libc.src.__support.FPUtil.multiply_add
libc.src.__support.FPUtil.polyeval
- libc.src.__support.FPUtil.double_double
- libc.src.__support.FPUtil.dyadic_float
+ libc.src.__support.integer_literals
libc.src.__support.macros.optimization
COMPILE_OPTIONS
-O3
@@ -1186,12 +1193,13 @@ add_entrypoint_object(
DEPENDS
.common_constants
.log_range_reduction
+ libc.src.__support.FPUtil.double_double
+ libc.src.__support.FPUtil.dyadic_float
libc.src.__support.FPUtil.fenv_impl
libc.src.__support.FPUtil.fp_bits
libc.src.__support.FPUtil.multiply_add
libc.src.__support.FPUtil.polyeval
- libc.src.__support.FPUtil.double_double
- libc.src.__support.FPUtil.dyadic_float
+ libc.src.__support.integer_literals
libc.src.__support.macros.optimization
COMPILE_OPTIONS
-O3
diff --git a/libc/src/math/generic/exp.cpp b/libc/src/math/generic/exp.cpp
index f23170f8ed4259..42a4491131a04e 100644
--- a/libc/src/math/generic/exp.cpp
+++ b/libc/src/math/generic/exp.cpp
@@ -21,6 +21,7 @@
#include "src/__support/FPUtil/rounding_mode.h"
#include "src/__support/FPUtil/triple_double.h"
#include "src/__support/common.h"
+#include "src/__support/integer_literals.h"
#include "src/__support/macros/optimization.h" // LIBC_UNLIKELY
#include <errno.h>
@@ -31,6 +32,7 @@ using fputil::DoubleDouble;
using fputil::TripleDouble;
using Float128 = typename fputil::DyadicFloat<128>;
using Sign = fputil::Sign;
+using LIBC_NAMESPACE::operator""_u128;
// log2(e)
constexpr double LOG2_E = 0x1.71547652b82fep+0;
@@ -97,21 +99,15 @@ DoubleDouble poly_approx_dd(const DoubleDouble &dx) {
// For |dx| < 2^-13 + 2^-30:
// | output - exp(dx) | < 2^-126.
Float128 poly_approx_f128(const Float128 &dx) {
- using MType = typename Float128::MantissaType;
-
constexpr Float128 COEFFS_128[]{
- {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
+ {Sign::POS, -127, 0x80000000'00000000'00000000'00000000_u128}, // 1.0
+ {Sign::POS, -127, 0x80000000'00000000'00000000'00000000_u128}, // 1.0
+ {Sign::POS, -128, 0x80000000'00000000'00000000'00000000_u128}, // 0.5
+ {Sign::POS, -130, 0xaaaaaaaa'aaaaaaaa'aaaaaaaa'aaaaaaab_u128}, // 1/6
+ {Sign::POS, -132, 0xaaaaaaaa'aaaaaaaa'aaaaaaaa'aaaaaaab_u128}, // 1/24
+ {Sign::POS, -134, 0x88888888'88888888'88888888'88888889_u128}, // 1/120
+ {Sign::POS, -137, 0xb60b60b6'0b60b60b'60b60b60'b60b60b6_u128}, // 1/720
+ {Sign::POS, -140, 0xd00d00d0'0d00d00d'00d00d00'd00d00d0_u128}, // 1/5040
};
Float128 p = fputil::polyeval(dx, COEFFS_128[0], COEFFS_128[1], COEFFS_128[2],
diff --git a/libc/src/math/generic/exp10.cpp b/libc/src/math/generic/exp10.cpp
index 6b40f5561845d8..72ece669765688 100644
--- a/libc/src/math/generic/exp10.cpp
+++ b/libc/src/math/generic/exp10.cpp
@@ -21,6 +21,7 @@
#include "src/__support/FPUtil/rounding_mode.h"
#include "src/__support/FPUtil/triple_double.h"
#include "src/__support/common.h"
+#include "src/__support/integer_literals.h"
#include "src/__support/macros/optimization.h" // LIBC_UNLIKELY
#include <errno.h>
@@ -31,6 +32,7 @@ using fputil::DoubleDouble;
using fputil::TripleDouble;
using Float128 = typename fputil::DyadicFloat<128>;
using Sign = fputil::Sign;
+using LIBC_NAMESPACE::operator""_u128;
// log2(10)
constexpr double LOG2_10 = 0x1.a934f0979a371p+1;
@@ -99,17 +101,15 @@ DoubleDouble poly_approx_dd(const DoubleDouble &dx) {
// For |dx| < 2^-14:
// | output - 10^dx | < 1.5 * 2^-124.
Float128 poly_approx_f128(const Float128 &dx) {
- using MType = typename Float128::MantissaType;
-
constexpr Float128 COEFFS_128[]{
- {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})},
+ {Sign::POS, -127, 0x80000000'00000000'00000000'00000000_u128}, // 1.0
+ {Sign::POS, -126, 0x935d8ddd'aaa8ac16'ea56d62b'82d30a2d_u128},
+ {Sign::POS, -126, 0xa9a92639'e753443a'80a99ce7'5f4d5bdb_u128},
+ {Sign::POS, -126, 0x82382c8e'f1652304'6a4f9d7d'bf6c9635_u128},
+ {Sign::POS, -124, 0x12bd7609'fd98c44c'34578701'9216c7af_u128},
+ {Sign::POS, -127, 0x450a7ff4'7535d889'cc41ed7e'0d27aee5_u128},
+ {Sign::POS, -130, 0xd3f6b844'702d636b'8326bb91'a6e7601d_u128},
+ {Sign::POS, -130, 0x45b937f0'd05bb1cd'fa7b46df'314112a9_u128},
};
Float128 p = fputil::polyeval(dx, COEFFS_128[0], COEFFS_128[1], COEFFS_128[2],
diff --git a/libc/src/math/generic/exp2.cpp b/libc/src/math/generic/exp2.cpp
index 01e66d1ae00f70..83f545eb116bd3 100644
--- a/libc/src/math/generic/exp2.cpp
+++ b/libc/src/math/generic/exp2.cpp
@@ -21,6 +21,7 @@
#include "src/__support/FPUtil/rounding_mode.h"
#include "src/__support/FPUtil/triple_double.h"
#include "src/__support/common.h"
+#include "src/__support/integer_literals.h"
#include "src/__support/macros/optimization.h" // LIBC_UNLIKELY
#include <errno.h>
@@ -31,6 +32,7 @@ using fputil::DoubleDouble;
using fputil::TripleDouble;
using Float128 = typename fputil::DyadicFloat<128>;
using Sign = fputil::Sign;
+using LIBC_NAMESPACE::operator""_u128;
// Error bounds:
// Errors when using double precision.
@@ -88,17 +90,15 @@ DoubleDouble poly_approx_dd(const DoubleDouble &dx) {
// For |dx| < 2^-13 + 2^-30:
// | output - exp(dx) | < 2^-126.
Float128 poly_approx_f128(const Float128 &dx) {
- using MType = typename Float128::MantissaType;
-
constexpr Float128 COEFFS_128[]{
- {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})},
+ {Sign::POS, -127, 0x80000000'00000000'00000000'00000000_u128}, // 1.0
+ {Sign::POS, -128, 0xb17217f7'd1cf79ab'c9e3b398'03f2f6af_u128},
+ {Sign::POS, -128, 0x3d7f7bff'058b1d50'de2d60dd'9c9a1d9f_u128},
+ {Sign::POS, -132, 0xe35846b8'2505fc59'9d3b15d9'e7fb6897_u128},
+ {Sign::POS, -134, 0x9d955b7d'd273b94e'184462f6'bcd2b9e7_u128},
+ {Sign::POS, -137, 0xaec3ff3c'53398883'39ea1bb9'64c51a89_u128},
+ {Sign::POS, -138, 0x2861225f'345c396a'842c5341'8fa8ae61_u128},
+ {Sign::POS, -144, 0xffe5fe2d'109a319d'7abeb5ab'd5ad2079_u128},
};
Float128 p = fputil::polyeval(dx, COEFFS_128[0], COEFFS_128[1], COEFFS_128[2],
diff --git a/libc/src/math/generic/expm1.cpp b/libc/src/math/generic/expm1.cpp
index c1fb80309d7b46..9f14a8c2068ec1 100644
--- a/libc/src/math/generic/expm1.cpp
+++ b/libc/src/math/generic/expm1.cpp
@@ -22,6 +22,7 @@
#include "src/__support/FPUtil/rounding_mode.h"
#include "src/__support/FPUtil/triple_double.h"
#include "src/__support/common.h"
+#include "src/__support/integer_literals.h"
#include "src/__support/macros/optimization.h" // LIBC_UNLIKELY
#include <errno.h>
@@ -39,6 +40,7 @@ using fputil::DoubleDouble;
using fputil::TripleDouble;
using Float128 = typename fputil::DyadicFloat<128>;
using Sign = fputil::Sign;
+using LIBC_NAMESPACE::operator""_u128;
// log2(e)
constexpr double LOG2_E = 0x1.71547652b82fep+0;
@@ -107,20 +109,14 @@ DoubleDouble poly_approx_dd(const DoubleDouble &dx) {
// For |dx| < 2^-13 + 2^-30:
// | output - exp(dx) | < 2^-126.
Float128 poly_approx_f128(const Float128 &dx) {
- using MType = typename Float128::MantissaType;
-
constexpr Float128 COEFFS_128[]{
- {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
+ {Sign::POS, -127, 0x80000000'00000000'00000000'00000000_u128}, // 1.0
+ {Sign::POS, -128, 0x80000000'00000000'00000000'00000000_u128}, // 0.5
+ {Sign::POS, -130, 0xaaaaaaaa'aaaaaaaa'aaaaaaaa'aaaaaaab_u128}, // 1/6
+ {Sign::POS, -132, 0xaaaaaaaa'aaaaaaaa'aaaaaaaa'aaaaaaab_u128}, // 1/24
+ {Sign::POS, -134, 0x88888888'88888888'88888888'88888889_u128}, // 1/120
+ {Sign::POS, -137, 0xb60b60b6'0b60b60b'60b60b60'b60b60b6_u128}, // 1/720
+ {Sign::POS, -140, 0xd00d00d0'0d00d00d'00d00d00'd00d00d0_u128}, // 1/5040
};
Float128 p = fputil::polyeval(dx, COEFFS_128[0], COEFFS_128[1], COEFFS_128[2],
@@ -146,7 +142,6 @@ std::ostream &operator<<(std::ostream &OS, const DoubleDouble &r) {
// TODO(lntue): investigate triple-double precision implementation for this
// step.
Float128 expm1_f128...
[truncated]
|
lntue
approved these changes
Feb 16, 2024
re-relanding 🤞 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.