Merge pull request #661 from cppalliance/develop

Author: mborland

.drone.jsonnet

@@ -159,9 +159,44 @@ local windows_pipeline(name, image, environment, arch = "amd64") =
     ),
 
     linux_pipeline(
-        "Linux 22.04 GCC 12 32 ASAN",
+        "Linux 22.04 GCC 12 32 ASAN 03",
         "cppalliance/droneubuntu2204:1",
-        { TOOLSET: 'gcc', COMPILER: 'g++-12', CXXSTD: '03,11,14,17,20,2b', ADDRMD: '32' } + asan,
+        { TOOLSET: 'gcc', COMPILER: 'g++-12', CXXSTD: '03', ADDRMD: '32' } + asan,
+        "g++-12-multilib",
+    ),
+
+    linux_pipeline(
+        "Linux 22.04 GCC 12 32 ASAN 11",
+        "cppalliance/droneubuntu2204:1",
+        { TOOLSET: 'gcc', COMPILER: 'g++-12', CXXSTD: '11', ADDRMD: '32' } + asan,
+        "g++-12-multilib",
+    ),
+
+    linux_pipeline(
+        "Linux 22.04 GCC 12 32 ASAN 14",
+        "cppalliance/droneubuntu2204:1",
+        { TOOLSET: 'gcc', COMPILER: 'g++-12', CXXSTD: '14', ADDRMD: '32' } + asan,
+        "g++-12-multilib",
+    ),
+
+    linux_pipeline(
+        "Linux 22.04 GCC 12 32 ASAN 17",
+        "cppalliance/droneubuntu2204:1",
+        { TOOLSET: 'gcc', COMPILER: 'g++-12', CXXSTD: '17', ADDRMD: '32' } + asan,
+        "g++-12-multilib",
+    ),
+
+    linux_pipeline(
+        "Linux 22.04 GCC 12 32 ASAN 20",
+        "cppalliance/droneubuntu2204:1",
+        { TOOLSET: 'gcc', COMPILER: 'g++-12', CXXSTD: '20', ADDRMD: '32' } + asan,
+        "g++-12-multilib",
+    ),
+
+    linux_pipeline(
+        "Linux 22.04 GCC 12 32 ASAN 2b",
+        "cppalliance/droneubuntu2204:1",
+        { TOOLSET: 'gcc', COMPILER: 'g++-12', CXXSTD: '2b', ADDRMD: '32' } + asan,
         "g++-12-multilib",
     ),
 
@@ -187,16 +222,30 @@ local windows_pipeline(name, image, environment, arch = "amd64") =
     ),
 
     linux_pipeline(
-        "Linux 24.04 GCC 14 32/64",
+        "Linux 24.04 GCC 14 32",
+        "cppalliance/droneubuntu2404:1",
+        { TOOLSET: 'gcc', COMPILER: 'g++-14', CXXSTD: '03,11,14,17,20,23', ADDRMD: '32', CXXFLAGS: "-fexcess-precision=fast" },
+        "g++-14-multilib",
+    ),
+
+    linux_pipeline(
+        "Linux 24.04 GCC 14 64",
+        "cppalliance/droneubuntu2404:1",
+        { TOOLSET: 'gcc', COMPILER: 'g++-14', CXXSTD: '03,11,14,17,20,23', ADDRMD: '64', CXXFLAGS: "-fexcess-precision=fast" },
+        "g++-14-multilib",
+    ),
+
+    linux_pipeline(
+        "Linux 24.04 GCC 14 GNU 32",
         "cppalliance/droneubuntu2404:1",
-        { TOOLSET: 'gcc', COMPILER: 'g++-14', CXXSTD: '03,11,14,17,20,23', ADDRMD: '32,64', CXXFLAGS: "-fexcess-precision=fast" },
+        { TOOLSET: 'gcc', COMPILER: 'g++-14', CXXSTD: '03,11,14,17,20,23', ADDRMD: '32', CXXFLAGS: "-fexcess-precision=fast", CXXSTDDIALECT: "gnu" },
         "g++-14-multilib",
     ),
 
     linux_pipeline(
-        "Linux 24.04 GCC 14 GNU 32/64",
+        "Linux 24.04 GCC 14 GNU 64",
         "cppalliance/droneubuntu2404:1",
-        { TOOLSET: 'gcc', COMPILER: 'g++-14', CXXSTD: '03,11,14,17,20,23', ADDRMD: '32,64', CXXFLAGS: "-fexcess-precision=fast", CXXSTDDIALECT: "gnu" },
+        { TOOLSET: 'gcc', COMPILER: 'g++-14', CXXSTD: '03,11,14,17,20,23', ADDRMD: '64', CXXFLAGS: "-fexcess-precision=fast", CXXSTDDIALECT: "gnu" },
         "g++-14-multilib",
     ),
 
@@ -358,16 +407,6 @@ local windows_pipeline(name, image, environment, arch = "amd64") =
         ["deb http://apt.llvm.org/jammy/ llvm-toolchain-jammy-18 main"],
     ),
 
-    macos_pipeline(
-        "MacOS 10.15 Xcode 12.2 UBSAN",
-        { TOOLSET: 'clang', COMPILER: 'clang++', CXXSTD: '03,11,14,17,2a' } + ubsan,
-    ),
-
-    macos_pipeline(
-        "MacOS 10.15 Xcode 12.2 ASAN",
-        { TOOLSET: 'clang', COMPILER: 'clang++', CXXSTD: '03,11,14,17,2a' } + asan,
-    ),
-
     macos_pipeline(
         "MacOS 12.4 Xcode 13.4.1 UBSAN",
         { TOOLSET: 'clang', COMPILER: 'clang++', CXXSTD: '03,11,14,17,20,2b' } + ubsan,

.github/workflows/ci.yml

@@ -248,7 +248,7 @@ jobs:
             cxxstd: "03,11,14,17,20,2b"
             os: macos-14
 
-    timeout-minutes: 120
+    timeout-minutes: 180
     runs-on: ${{matrix.os}}
     container: ${{matrix.container}}
 
@@ -456,20 +456,30 @@ jobs:
         include:
           - toolset: msvc-14.2
             cxxstd: "14,17,20,latest"
-            addrmd: 32,64
+            addrmd: "32"
             os: windows-2019
-          - toolset: msvc-14.3
+          # B2 does not work with MSVC 17.10. Once it's updated we can re-enable these tests
+          # Still covered in drone
+          #- toolset: msvc-14.3
+          #  cxxstd: "14,17,20,latest"
+          #  addrmd: "32"
+          #  os: windows-2022
+          - toolset: msvc-14.2
             cxxstd: "14,17,20,latest"
-            addrmd: 32,64
-            os: windows-2022
-          - toolset: clang-win
-            cxxstd: "14,17,latest"
-            addrmd: "32"
-            os: windows-2022
-          - toolset: clang-win
-            cxxstd: "14,17,latest"
             addrmd: "64"
-            os: windows-2022
+            os: windows-2019
+          #- toolset: msvc-14.3
+          #  cxxstd: "14,17,20,latest"
+          #  addrmd: "64"
+          #  os: windows-2022
+          #- toolset: clang-win
+          #  cxxstd: "14,17,latest"
+          #  addrmd: "32"
+          #  os: windows-2022
+          #- toolset: clang-win
+          #  cxxstd: "14,17,latest"
+          #  addrmd: "64"
+          #  os: windows-2022
           - toolset: gcc
             cxxstd: "03,11,14,17,2a"
             addrmd: "64"

doc/decimal.adoc

@@ -20,6 +20,11 @@ include::decimal/generic_decimal.adoc[]
 include::decimal/decimal32.adoc[]
 include::decimal/decimal64.adoc[]
 include::decimal/decimal128.adoc[]
+include::decimal/fast_types.adoc[]
+include::decimal/decimal32_fast.adoc[]
+include::decimal/decimal64_fast.adoc[]
+include::decimal/decimal128_fast.adoc[]
+include::decimal/conversions.adoc[]
 include::decimal/literals.adoc[]
 include::decimal/numbers.adoc[]
 include::decimal/cmath.adoc[]

doc/decimal/benchmarks.adoc

@@ -32,20 +32,29 @@ Run using a Macbook pro with M1 pro chipset running macOS Sonoma 14.4.1 and home
 |===
 | Type | Runtime (us) | Ratio to `double`
 | `float`
-| 8764
-| 1.577
+| 8587
+| 1.376
 | `double`
-| 5559
+| 6240
 | 1.000
 | `decimal32`
-| 276,124
-| 49.672
+| 275,597
+| 44.166
 | `decimal64`
-| 355,999
-| 64.760
+| 296,929
+| 47.587
 | `decimal128`
-| 989,028
-| 177.915
+| 821,847
+| 131.706
+| `decimal32_fast`
+| 99,664
+| 15.972
+| `decimal64_fast`
+| 102,132
+| 16.367
+| `decimal128_fast`
+| 146,302
+| 23.446
 |===
 
 == Basic Operations
@@ -62,83 +71,119 @@ Run using a Macbook pro with M1 pro chipset running macOS Sonoma 14.4.1 and home
 |===
 | Type | Runtime (us) | Ratio to `double`
 | `float`
-| 2113
-| 0.739
+| 2705
+| 0.859
 | `double`
-| 2860
+| 3148
 | 1.000
 | `decimal32`
-| 353,836
-| 123.719
+| 351,505
+| 111.660
 | `decimal64`
-| 409,098
-| 143.041
+| 359,425
+| 114.176
 | `decimal128`
-| 2,418,039
-| 845.468
+| 1,446,674
+| 459.553
+| `decimal32_fast`
+| 146,873
+| 46.656
+| `decimal64_fast`
+| 139,294
+| 44.248
+| `decimal128_fast`
+| 707,308
+| 224.685
 |===
 
 ==== Subtraction
 
 |===
 | Type | Runtime (us) | Ratio to `double`
 | `float`
-| 1782
-| 1.061
+| 3339
+| 2.014
 | `double`
-| 1680
+| 1658
 | 1.000
 | `decimal32`
-| 293,927
-| 174.957
+| 267,646
+| 161.427
 | `decimal64`
-| 329,425
-| 196.086
+| 303,589
+| 183.106
 | `decimal128`
-| 1,527,261
-| 909.084
+| 954,211
+| 575.519
+| `decimal32_fast`
+| 147,112
+| 88.729
+| `decimal64_fast`
+| 145,606
+| 87.820
+| `decimal128_fast`
+| 394,538
+| 2387.960
 |===
 
 ==== Multiplication
 
 |===
 | Type | Runtime (us) | Ratio to `double`
 | `float`
-| 1691
-| 0.979
+| 1646
+| 0.957
 | `double`
-| 1728
+| 1720
 | 1.000
 | `decimal32`
-| 309,117
-| 178.887
+| 313,219
+| 182.104
 | `decimal64`
-| 408,010
-| 236.117
+| 583,818
+| 339.429
 | `decimal128`
-| 2,506,105
-| 1450.292
+| 1,881,936
+| 1094.149
+| `decimal32_fast`
+| 86,093
+| 50.054
+| `decimal64_fast`
+| 333,582
+| 193.943
+| `decimal128_fast`
+| 1,269,429
+| 738.040
 |===
 
 ==== Division
 
 |===
 | Type | Runtime (us) | Ratio to `double`
 | `float`
-| 2058
-| 0.846
+| 2120
+| 0.547
 | `double`
-| 2434
+| 3874
 | 1.000
 | `decimal32`
-| 304,852
-| 125.247
+| 307,337
+| 79.333
 | `decimal64`
-| 519,990
-| 213.636
+| 447,910
+| 115.620
 | `decimal128`
-| 3,534,909
-| 1452.304
+| 2,544,798
+| 656.892
+| `decimal32_fast`
+| 105,796
+| 27.309
+| `decimal64_fast`
+| 291,671
+| 75.289
+| `decimal128_fast`
+| 302,003
+| 77.956
 |===
 
 == Selected Special Functions

doc/decimal/cmath.adoc

@@ -458,3 +458,14 @@ constexpr boost::decimal::detail::uint128 frexpd128(decimal128 num, int* expptr)
 
 This function is very similar to https://en.cppreference.com/w/cpp/numeric/math/frexp[frexp], but returns the significand and an integral power of 10 since the `FLT_RADIX` of this type is 10.
 The significand is normalized to the number of digits of precision the type has (e.g. for decimal32 it is [1'000'000, 9'999'999]).
+
+=== trunc_to
+
+[source, c++]
+----
+template <typename Decimal>
+constexpr Decimal trunc_to(Decimal val, int precision = 0);
+----
+
+The function returns the decimal type with number of fractional digits equal to the value of precision.
+`trunc_to` is similar to https://en.cppreference.com/w/cpp/numeric/math/trunc[trunc], and with the default precision argument of 0 it is identical.

doc/decimal/config.adoc

@@ -25,3 +25,6 @@ constexpr decimal64 half {5, -1};
 std::complex<decimal64> test_val {half, half};
 const auto res = std::acos(test_val);
 ----
+
+- `BOOST_DECIMAL_FAST_MATH` performs optimizations similar to that of the `-ffast-math` compiler flag such as removing all checks for non-finite values.
+This flag increases the performance of the basis operations (e.g. add, sub, mul, div, and comparisons) by up to 20%.