Use MSVC intrinsics for full multiplication (#5782)

* Use MSVC intrinsics for full multiplication

Signed-off-by: Suyog Verma <suyogverma0057@gmail.com>

* Fix formatting in unit-class_lexer

Signed-off-by: Suyog Verma <suyogverma0057@gmail.com>

* Address review feedback

Signed-off-by: Suyog Verma <suyogverma0057@gmail.com>

---------

Signed-off-by: Suyog Verma <suyogverma0057@gmail.com>

docs/mkdocs/docs/community/roadmap.md

@@ -36,26 +36,13 @@ work items are tracked in the [GitHub milestones](https://github.com/nlohmann/js
 ## API stability
 
 Releases follow [semantic versioning](https://semver.org): a minor or patch release of version 3.x does not break code
-that uses the public API, unless that code opts in to a change with a macro as described [below](#version-40). In
+that uses the public API. In particular, a 3.x release does not:
-particular, a 3.x release does not:
 
-- make breaking changes to the signature of a function: the types or order of its existing parameters, its return type,
+- change the signature of a function (its parameter types, return type, number of parameters, or the const-ness of a
-  its `noexcept` or `constexpr` specifier, or the const-ness of a member function. New parameters may be added if they
+  member function);
-  have a default value;
+- remove or rename a function or class;
-- remove or rename a function or class, or change the template parameters of a public class template;
 - change which exceptions a function throws, or the [exception ids](../home/exceptions.md);
-- change access specifiers, or change or remove existing default arguments. New default arguments may be added;
+- change access specifiers or default arguments.
-- change the JSON type that a valid input parses to, or the text that `dump()` produces for a valid value;
-- accept input that was rejected before, or reject input that was accepted before;
-- change the order in which the keys of an object are iterated. The default type sorts keys, and
-  [`ordered_json`](../api/ordered_json.md) keeps insertion order;
-- change when iterators, pointers, or references are invalidated, or the state of a moved-from `basic_json`;
-- add or remove implicit conversions from `basic_json`;
-- change how `to_json` and `from_json` functions are found, or the behavior of
-  [`adl_serializer`](../api/adl_serializer/index.md);
-- add pure virtual functions to the [`json_sax`](../api/json_sax/index.md) interface;
-- remove, rename, renumber, or add enumerators of `value_t`;
-- remove or rename a documented macro, CMake option, CMake target, or header, or change what a documented macro does.
 
 Exceptions to these rules, for instance when fixing a bug requires changing the exception a function throws, are
 documented in the [release notes](../home/releases.md).
@@ -64,14 +51,13 @@ The following are **not** part of the public API and may change in any release,
 
 - The text of exception messages returned by `what()`. Use the [exception id](../home/exceptions.md) to tell errors
   apart.
-- The ABI, including `sizeof(basic_json)` and the memory layout of its values. The
+- The ABI, including `sizeof(basic_json)` and the memory layout of its values. Recompile your code when you upgrade the
-  [versioned inline namespace](../features/namespace.md) turns mixing versions into a link error.
+  library. The [versioned inline namespace](../features/namespace.md) turns mixing versions into a link error.
-- The hash values returned by `std::hash` for `basic_json`. Numbers that compare equal still hash equally.
 - Everything in namespace `nlohmann::detail`, and macros and type traits that are not documented in the
   [API reference](../api/basic_json/index.md).
 
-Breaking changes are only added behind a macro whose default keeps the 3.x behavior. See [Version 4.0](#version-40) and
+Changes that would break the public API are only added behind a macro whose default keeps the 3.x behavior, see
-the [macro overview](../features/macros.md).
+[Version 4.0](#version-40).
 
 ## Version 4.0
 

include/nlohmann/detail/bit_ops.hpp

@@ -9,12 +9,15 @@
 #pragma once
 
 #include <cstdint> // uint64_t
+#if !defined(__SIZEOF_INT128__) && defined(_MSC_VER) && (defined(_M_X64) || defined(_M_ARM64))
+    #include <intrin0.h> // __umulh, _umul128
+#endif
 
 #include <nlohmann/detail/abi_macros.hpp>
 
 // Portable bit-level helpers for the number and string scanners. They use
-// compiler builtins where available and plain C++ otherwise, so they need no
+// compiler builtins or platform-specific intrinsics where available and plain
-// platform headers and work regardless of byte order.
+// C++ otherwise, so they work regardless of byte order.
 
 NLOHMANN_JSON_NAMESPACE_BEGIN
 namespace detail
@@ -52,6 +55,12 @@ inline uint128_parts full_multiplication(std::uint64_t a, std::uint64_t b) noexc
     __extension__ using uint128 = unsigned __int128;
     const uint128 r = static_cast<uint128>(a) * b;
     return {static_cast<std::uint64_t>(r), static_cast<std::uint64_t>(r >> 64u)};
+#elif defined(_MSC_VER) && defined(_M_X64)
+    std::uint64_t high = 0;
+    const std::uint64_t low = _umul128(a, b, &high);
+    return {low, high};
+#elif defined(_MSC_VER) && defined(_M_ARM64)
+    return {a * b, __umulh(a, b)};
 #else
     const std::uint64_t a_lo = a & 0xFFFFFFFFu;
     const std::uint64_t a_hi = a >> 32u;

single_include/nlohmann/json.hpp

@@ -8791,13 +8791,16 @@ NLOHMANN_JSON_NAMESPACE_END
 
 
 #include <cstdint> // uint64_t
+#if !defined(__SIZEOF_INT128__) && defined(_MSC_VER) && (defined(_M_X64) || defined(_M_ARM64))
+    #include <intrin0.h> // __umulh, _umul128
+#endif
 
 // #include <nlohmann/detail/abi_macros.hpp>
 
 
 // Portable bit-level helpers for the number and string scanners. They use
-// compiler builtins where available and plain C++ otherwise, so they need no
+// compiler builtins or platform-specific intrinsics where available and plain
-// platform headers and work regardless of byte order.
+// C++ otherwise, so they work regardless of byte order.
 
 NLOHMANN_JSON_NAMESPACE_BEGIN
 namespace detail
@@ -8835,6 +8838,12 @@ inline uint128_parts full_multiplication(std::uint64_t a, std::uint64_t b) noexc
     __extension__ using uint128 = unsigned __int128;
     const uint128 r = static_cast<uint128>(a) * b;
     return {static_cast<std::uint64_t>(r), static_cast<std::uint64_t>(r >> 64u)};
+#elif defined(_MSC_VER) && defined(_M_X64)
+    std::uint64_t high = 0;
+    const std::uint64_t low = _umul128(a, b, &high);
+    return {low, high};
+#elif defined(_MSC_VER) && defined(_M_ARM64)
+    return {a * b, __umulh(a, b)};
 #else
     const std::uint64_t a_lo = a & 0xFFFFFFFFu;
     const std::uint64_t a_hi = a >> 32u;

tests/src/unit-class_lexer.cpp

@@ -17,6 +17,7 @@ using nlohmann::json;
 #include <cstdint> // uint32_t, uint64_t
 #include <cstdlib> // strtod
 #include <cstring> // memcpy
+#include <limits> // numeric_limits
 #include <sstream> // stringstream
 #include <string> // string
 #include <utility> // pair
@@ -891,8 +892,39 @@ TEST_CASE("Eisel-Lemire float conversion")
 
     SECTION("128-bit products and leading zeros")
     {
+        const auto check_product = [](std::uint64_t a, std::uint64_t b)
+        {
+            const auto product = nlohmann::detail::full_multiplication(a, b);
+            CHECK(big_from(product.high, product.low) == big_mul(big_from(0, a), big_from(0, b)));
+        };
+
+        const std::uint64_t max = (std::numeric_limits<std::uint64_t>::max)();
+        const std::array<std::pair<std::uint64_t, std::uint64_t>, 13> edge_cases =
+        {
+            {
+                {0, 0},
+                {0, 1},
+                {1, 1},
+                {1, max},
+                {0xFFFFFFFFu, 0x100000000u},
+                {0x100000000u, 0x100000000u},
+                {0x100000001u, 0x100000001u},
+                {max, max},
+                {max, 2},
+                {0xFFFFFFFF00000000u, 0x100000001u},
+                {0x100000001u, 0xFFFFFFFF00000000u},
+                {max, 1},
+                {2, max},
+            }
+        };
+
+        for (const auto& test : edge_cases)
+        {
+            check_product(test.first, test.second);
+        }
+
         // whichever implementation the compiler gets (with or without a
-        // 128-bit integer type or a builtin)
+        // 128-bit integer type or a builtin / intrinsic)
         std::uint64_t state = 42;
         for (int i = 0; i < 10000; ++i)
         {
@@ -901,8 +933,7 @@ TEST_CASE("Eisel-Lemire float conversion")
             state ^= state << 17u;
             const std::uint64_t a = state;
             const std::uint64_t b = (state * 0x9E3779B97F4A7C15u) >> (i % 64);
-            const auto product = nlohmann::detail::full_multiplication(a, b);
+            check_product(a, b);
-            CHECK(big_from(product.high, product.low) == big_mul(big_from(0, a), big_from(0, b)));
 
             const int k = i % 64;
             const std::uint64_t x = (std::uint64_t{1} << k) | (a & ((std::uint64_t{1} << k) - 1));