incidentalstoat/blankart
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Merge branch 'blankart-dev' into 32p

Browse source
This commit is contained in:
minenice55 2025-10-17 12:05:34 -04:00
commit 36527b7aa1
2 changed files with 112 additions and 4 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

11
src/blua/luaconf.h
View file

@ -571,12 +571,23 @@
#if defined(LUA_NUMBER_DOUBLE) && !defined(LUA_ANSI) && !defined(__SSE2__) && \
(defined(__i386) || defined (_M_IX86) || defined(__i386__))
/* On a Microsoft compiler, use assembler */
#if defined(_MSC_VER)
#define lua_number2int(i,d) __asm fld d __asm fistp i
#define lua_number2integer(i,n) lua_number2int(i, n)
/* the next trick should work on any Pentium, but sometimes clashes
with a DirectX idiosyncrasy */
#else
union luai_Cast { double l_d; long l_l; };
#define lua_number2int(i,d) \
{ volatile union luai_Cast u; u.l_d = (d) + 6755399441055744.0; (i) = u.l_l; }
#define lua_number2integer(i,n) lua_number2int(i, n)
#endif
/* this option always works, but may be slow */
#else

105
src/libdivide.h
View file

@ -24,6 +24,13 @@
#include <stdlib.h>
#endif
#if defined(_MSC_VER) && (defined(__cplusplus) && (__cplusplus >= 202002L)) || \
(defined(_MSVC_LANG) && (_MSVC_LANG >= 202002L))
#include <limits.h>
#include <type_traits>
#define LIBDIVIDE_VC_CXX20
#endif
#if defined(LIBDIVIDE_SSE2)
#include <emmintrin.h>
#endif
@ -36,6 +43,27 @@
#include <arm_neon.h>
#endif
// Clang-cl prior to Visual Studio 2022 doesn't include __umulh/__mulh intrinsics
#if defined(_MSC_VER) && (!defined(__clang__) || _MSC_VER > 1930) && \
(defined(_M_X64) || defined(_M_ARM64) || defined(_M_HYBRID_X86_ARM64) || defined(_M_ARM64EC))
#define LIBDIVIDE_MULH_INTRINSICS
#endif
#if defined(_MSC_VER)
#if defined(LIBDIVIDE_MULH_INTRINSICS) || !defined(__clang__)
#include <intrin.h>
#endif
#ifndef __clang__
#pragma warning(push)
// 4146: unary minus operator applied to unsigned type, result still unsigned
#pragma warning(disable : 4146)
// 4204: nonstandard extension used : non-constant aggregate initializer
#pragma warning(disable : 4204)
#endif
#define LIBDIVIDE_VC
#endif
#if !defined(__has_builtin)
#define __has_builtin(x) 0
#endif
@ -74,8 +102,12 @@
#endif
#endif
#ifndef LIBDIVIDE_INLINE
#ifdef _MSC_VER
#define LIBDIVIDE_INLINE __forceinline
#else
#define LIBDIVIDE_INLINE inline
#endif
#endif
#if defined(__AVR__)
#define LIBDIVIDE_ERROR(msg)
@ -108,6 +140,15 @@
// Use https://en.cppreference.com/w/cpp/feature_test#cpp_constexpr
#if defined(__cpp_constexpr) && (__cpp_constexpr >= 201304L)
#define LIBDIVIDE_CONSTEXPR constexpr LIBDIVIDE_INLINE
// Supposedly, MSVC might not implement feature test macros right:
// https://stackoverflow.com/questions/49316752/feature-test-macros-not-working-properly-in-visual-c
// so check that _MSVC_LANG corresponds to at least c++14, and _MSC_VER corresponds to at least VS
// 2017 15.0 (for extended constexpr support:
// https://learn.microsoft.com/en-us/cpp/overview/visual-cpp-language-conformance?view=msvc-170)
#elif (defined(_MSC_VER) && _MSC_VER >= 1910) && (defined(_MSVC_LANG) && _MSVC_LANG >= 201402L)
#define LIBDIVIDE_CONSTEXPR constexpr LIBDIVIDE_INLINE
#else
#define LIBDIVIDE_CONSTEXPR LIBDIVIDE_INLINE
#endif
@ -115,6 +156,58 @@
namespace libdivide {
#endif
#if defined(_MSC_VER) && !defined(__clang__)
#if defined(LIBDIVIDE_VC_CXX20)
static LIBDIVIDE_CONSTEXPR int __builtin_clz(unsigned x) {
if (std::is_constant_evaluated()) {
for (int i = 0; i < sizeof(x) * CHAR_BIT; ++i) {
if (x >> (sizeof(x) * CHAR_BIT - 1 - i)) return i;
}
return sizeof(x) * CHAR_BIT;
}
#else
static LIBDIVIDE_INLINE int __builtin_clz(unsigned x) {
#endif
#if defined(_M_ARM) || defined(_M_ARM64) || defined(_M_HYBRID_X86_ARM64) || defined(_M_ARM64EC)
return (int)_CountLeadingZeros(x);
#elif defined(__AVX2__) || defined(__LZCNT__)
return (int)_lzcnt_u32(x);
#else
unsigned long r;
_BitScanReverse(&r, x);
return (int)(r ^ 31);
#endif
}
#if defined(LIBDIVIDE_VC_CXX20)
static LIBDIVIDE_CONSTEXPR int __builtin_clzll(unsigned long long x) {
if (std::is_constant_evaluated()) {
for (int i = 0; i < sizeof(x) * CHAR_BIT; ++i) {
if (x >> (sizeof(x) * CHAR_BIT - 1 - i)) return i;
}
return sizeof(x) * CHAR_BIT;
}
#else
static LIBDIVIDE_INLINE int __builtin_clzll(unsigned long long x) {
#endif
#if defined(_M_ARM) || defined(_M_ARM64) || defined(_M_HYBRID_X86_ARM64) || defined(_M_ARM64EC)
return (int)_CountLeadingZeros64(x);
#elif defined(_WIN64)
#if defined(__AVX2__) || defined(__LZCNT__)
return (int)_lzcnt_u64(x);
#else
unsigned long r;
_BitScanReverse64(&r, x);
return (int)(r ^ 63);
#endif
#else
int l = __builtin_clz((unsigned)x) + 32;
int h = __builtin_clz((unsigned)(x >> 32));
return !!((unsigned)(x >> 32)) ? h : l;
#endif
}
#endif // defined(_MSC_VER) && !defined(__clang__)
// pack divider structs to prevent compilers from padding.
// This reduces memory usage by up to 43% when using a large
// array of libdivide dividers and improves performance
@ -376,7 +469,7 @@ static LIBDIVIDE_INLINE int16_t libdivide_count_leading_zeros16(uint16_t val) {
// Fast way to count leading zeros
// On the AVR 8-bit architecture __builtin_clz() works on a int16_t.
return __builtin_clz(val);
#elif defined(__GNUC__) || __has_builtin(__builtin_clz)
#elif defined(__GNUC__) || __has_builtin(__builtin_clz) || defined(_MSC_VER)
// Fast way to count leading zeros
return (int16_t)(__builtin_clz(val) - 16);
#else
@ -399,7 +492,7 @@ static LIBDIVIDE_INLINE int32_t libdivide_count_leading_zeros32(uint32_t val) {
#if defined(__AVR__)
// Fast way to count leading zeros
return __builtin_clzl(val);
#elif defined(__GNUC__) || __has_builtin(__builtin_clz)
#elif defined(__GNUC__) || __has_builtin(__builtin_clz) || defined(_MSC_VER)
// Fast way to count leading zeros
return __builtin_clz(val);
#else
@ -419,7 +512,7 @@ static LIBDIVIDE_INLINE int32_t libdivide_count_leading_zeros32(uint32_t val) {
}
static LIBDIVIDE_INLINE int32_t libdivide_count_leading_zeros64(uint64_t val) {
#if defined(__GNUC__) || __has_builtin(__builtin_clzll)
#if defined(__GNUC__) || __has_builtin(__builtin_clzll) || defined(_MSC_VER)
// Fast way to count leading zeros
return __builtin_clzll(val);
#else
@ -3265,7 +3358,7 @@ LIBDIVIDE_INLINE typename NeonVecFor<T>::type operator/=(
}
#endif
#if __cplusplus >= 201103L
#if __cplusplus >= 201103L || (defined(_MSC_VER) && _MSC_VER >= 1900)
// libdivide::branchfree_divider<T>
template <typename T>
using branchfree_divider = divider<T, BRANCHFREE>;
@ -3275,4 +3368,8 @@ using branchfree_divider = divider<T, BRANCHFREE>;
#endif // __cplusplus
#if defined(_MSC_VER) && !defined(__clang__)
#pragma warning(pop)
#endif
#endif // LIBDIVIDE_H