Merge pull request #510 from cppalliance/More_benchmarks

cppalliance · Apr 26, 2024 · de615c0 · de615c0
2 parents e72a5fd + 4bb8f5b
commit de615c0
Show file tree

Hide file tree

Showing 2 changed files with 543 additions and 90 deletions.
diff --git a/doc/decimal/benchmarks.adoc b/doc/decimal/benchmarks.adoc
@@ -22,34 +22,36 @@ An example on Linux with b2: `../../../b2 cxxstd=20 toolset=gcc-13 define=BOOST_
 
 == Comparisons
 
-The benchmark for comparisons generates a random vector containing 10,000,000 elements and does operations `>`, `>=`, `<`, `<=`, `==`, and `!=` between `vec[i] and vec[i + 1]`.
+The benchmark for comparisons generates a random vector containing 2,000,000 elements and does operations `>`, `>=`, `<`, `<=`, `==`, and `!=` between `vec[i] and vec[i + 1]`.
+This is repeated 5 times to generate stable results.
 
 === M1 macOS Results
 
 Run using a Macbook pro with M1 pro chipset running macOS Sonoma 14.4.1 and homebrew Clang 18.1.4
 
 |===
-| Type | Runtime (ms) | Ratio to `double`
+| Type | Runtime (us) | Ratio to `double`
 | `float`
-| 8
-| 1.333
+| 9032
+| 1.589
 | `double`
-| 6
+| 5684
 | 1.000
 | `decimal32`
-| 380
-| 63.333
+| 285,453
+| 50.2204
 | `decimal64`
-| 408
-| 608.000
+| 352,644
+| 62.042
 | `decimal128`
-| 14641
-| 2440.170
+| 15,355,817
+| 2701.590
 |===
 
 == Basic Operations
 
-The benchmark for these operations generates a random vector containing 10,000,000 elements and does operations `+`, `-`, `*`, `/` between `vec[i] and vec[i + 1]`.
+The benchmark for these operations generates a random vector containing 2,000,000 elements and does operations `+`, `-`, `*`, `/` between `vec[i] and vec[i + 1]`.
+This is repeated 5 times to generate stable results.
 
 === M1 macOS Results
 
@@ -58,83 +60,275 @@ Run using a Macbook pro with M1 pro chipset running macOS Sonoma 14.4.1 and home
 ==== Addition
 
 |===
-| Type | Runtime (ms) | Ratio to `double`
+| Type | Runtime (us) | Ratio to `double`
 | `float`
-| 6
-| 1.500
+| 1641
+| 0.965
 | `double`
-| 4
+| 1708
 | 1.000
 | `decimal32`
-| 361
-| 90.250
+| 378,252
+| 221.459
 | `decimal64`
-| 568
-| 142.000
+| 589,313
+| 345.031
 | `decimal128`
-| 13428
-| 3357.000
+| 13,829,995
+| 8097.190
 |===
 
 ==== Subtraction
 
 |===
-| Type | Runtime (ms) | Ratio to `double`
+| Type | Runtime (us) | Ratio to `double`
 | `float`
-| 3
-| 3.000
+| 3633
+| 2.221
 | `double`
-| 1
+| 1636
 | 1.000
 | `decimal32`
-| 307
-| 307.000
+| 307,765
+| 188.120
 | `decimal64`
-| 465
-| 465.000
+| 461,442
+| 282.055
 | `decimal128`
-| 11444
-| 11444.000
+| 11,449,306
+| 6998.350
 |===
 
 ==== Multiplication
 
 |===
-| Type | Runtime (ms) | Ratio to `double`
+| Type | Runtime (us) | Ratio to `double`
 | `float`
-| 1
-| 0.333
+| 1678
+| 0.523
 | `double`
-| 3
+| 3209
 | 1.000
 | `decimal32`
-| 311
-| 103.667
+| 310,543
+| 96.773
 | `decimal64`
-| 569
-| 189.667
+| 570,938
+| 177.918
 | `decimal128`
-| 9430
-| 3143.330
+| 9,434,297
+| 2939.95
 |===
 
 ==== Division
 
 |===
-| Type | Runtime (ms) | Ratio to `double`
+| Type | Runtime (us) | Ratio to `double`
 | `float`
-| 2
-| 0.667
+| 2019
+| 0.565
 | `double`
-| 3
+| 3572
 | 1.000
 | `decimal32`
-| 319
-| 106.333
+| 322,116
+| 90.178
 | `decimal64`
-| 395
-| 131.667
+| 734,173
+| 205.536
 | `decimal128`
-| 14781
-| 4927.000
+| 14,592,284
+| 4085.19
+|===
+
+== Selected Special Functions
+
+The benchmark for these operations generates a random vector containing 2,000,000 elements and does operations `+`, `-`, `*`, `/` between `vec[i] and vec[i + 1]`.
+This is repeated 5 times to generate stable results.
+
+=== M1 macOS Results
+
+Run using a Macbook pro with M1 pro chipset running macOS Sonoma 14.4.1 and homebrew Clang 18.1.4
+
+==== SQRT
+
+|===
+| Type | Runtime (us) | Ratio to `double`
+| `float`
+| 1904
+| 0.565
+| `double`
+| 3746
+| 1.000
+| `decimal32`
+| 5,050,241
+| 1341.72
+| `decimal64`
+| 12,084,821
+| 3210.630
+| `decimal128`
+| 275,779,340
+| 73267.60
+|===
+
+== `<charconv>`
+
+For all the following the results compare against https://github.com/boostorg/charconv[boost.charconv] for 10'000'000 conversions.
+
+=== `from_chars` general
+
+==== M1 macOS Results
+
+Run using a Macbook pro with M1 pro chipset running macOS Sonoma 14.4.1 and homebrew Clang 18.1.4
+
+|===
+| Type | Runtime (us) | Ratio to `double`
+| `float`
+| 235,816
+| 0.953
+| `double`
+| 247,307
+| 1.000
+| `decimal32`
+| 366,682
+| 1.483
+| `decimal64`
+| 485,965
+| 1.965
+// Decimal128 was two orders of magnitude faster. I expect an issue
+//| `decimal128`
+//| 275,779,340
+//| 73267.60
+|===
+
+NOTE: `decimal128` is currently absent due to results showing it is 2 orders of magnitude faster than the others.
+This should not be the case so will be investigated.
+
+=== `from_chars` scientific
+
+==== M1 macOS Results
+
+Run using a Macbook pro with M1 pro chipset running macOS Sonoma 14.4.1 and homebrew Clang 18.1.4
+
+|===
+| Type | Runtime (us) | Ratio to `double`
+| `float`
+| 241,893
+| 0.975
+| `double`
+| 247,975
+| 1.000
+| `decimal32`
+| 358,189
+| 1.444
+| `decimal64`
+| 477,574
+| 1.926
+// Decimal128 was two orders of magnitude faster. I expect an issue
+//| `decimal128`
+//| 275,779,340
+//| 73267.60
+|===
+
+NOTE: `decimal128` is currently absent due to results showing it is 2 orders of magnitude faster than the others.
+This should not be the case so will be investigated.
+
+=== `to_chars` general shortest representation
+
+==== M1 macOS Results
+
+Run using a Macbook pro with M1 pro chipset running macOS Sonoma 14.4.1 and homebrew Clang 18.1.4
+
+|===
+| Type | Runtime (us) | Ratio to `double`
+| `float`
+| 316,300
+| 1.040
+| `double`
+| 304,272
+| 1.000
+| `decimal32`
+| 406,053
+| 1.335
+| `decimal64`
+| 678,451
+| 2.230
+| `decimal128`
+| 6,309,346
+| 20.736
+|===
+
+=== `to_chars` general 6-digits of precision
+
+==== M1 macOS Results
+
+Run using a Macbook pro with M1 pro chipset running macOS Sonoma 14.4.1 and homebrew Clang 18.1.4
+
+|===
+| Type | Runtime (us) | Ratio to `double`
+| `float`
+| 323,867
+| 0.967
+| `double`
+| 334,989
+| 1.000
+| `decimal32`
+| 409,608
+| 1.223
+| `decimal64`
+| 702,339
+| 2.097
+| `decimal128`
+| 6,305,521
+| 18.823
+|===
+
+=== `to_chars` scientific shortest representation
+
+==== M1 macOS Results
+
+Run using a Macbook pro with M1 pro chipset running macOS Sonoma 14.4.1 and homebrew Clang 18.1.4
+
+|===
+| Type | Runtime (us) | Ratio to `double`
+| `float`
+| 286,330
+| 1.011
+| `double`
+| 283,287
+| 1.000
+| `decimal32`
+| 290,117
+| 1.024
+| `decimal64`
+| 499,637
+| 1.764
+| `decimal128`
+| 3,096,910
+| 10.932
+|===
+
+=== `to_chars` scientific 6-digits of precision
+
+==== M1 macOS Results
+
+Run using a Macbook pro with M1 pro chipset running macOS Sonoma 14.4.1 and homebrew Clang 18.1.4
+
+|===
+| Type | Runtime (us) | Ratio to `double`
+| `float`
+| 258,710
+| 0.809
+| `double`
+| 319,676
+| 1.000
+| `decimal32`
+| 292,250
+| 0.914
+| `decimal64`
+| 516,399
+| 1.615
+| `decimal128`
+| 3,108,380
+| 9.724
 |===