perf(strings): establish Utf8ToAsciiConverter baseline benchmarks

2025-12-12 23:11:24 +00:00
parent 475010148b
commit 610976c41c
3 changed files with 159 additions and 0 deletions
--- a/docs/benchmarks/utf8-converter-baseline-2025-11-27.md
+++ b/docs/benchmarks/utf8-converter-baseline-2025-11-27.md
@@ -0,0 +1,44 @@
+# Utf8ToAsciiConverter Baseline Benchmarks
+
+**Date:** 2025-11-27
+**Implementation:** Original 3,631-line switch statement
+**Runtime:** .NET 10.0
+
+## Results
+
+```
+BenchmarkDotNet v0.15.6, Linux Ubuntu 25.10 (Questing Quokka)
+Intel Xeon CPU 2.80GHz, 1 CPU, 16 logical and 8 physical cores
+.NET SDK 10.0.100
+  [Host]     : .NET 10.0.0 (10.0.0, 10.0.25.52411), X64 RyuJIT x86-64-v4
+  DefaultJob : .NET 10.0.0 (10.0.0, 10.0.25.52411), X64 RyuJIT x86-64-v4
+```
+
+| Method                 | Mean            | Error         | StdDev        | Rank | Gen0     | Gen1     | Gen2     | Allocated |
+|----------------------- |----------------:|--------------:|--------------:|-----:|---------:|---------:|---------:|----------:|
+| Tiny_Ascii             |        82.81 ns |      0.402 ns |      0.314 ns |    2 |   0.0027 |        - |        - |      48 B |
+| Tiny_Mixed             |        71.05 ns |      0.225 ns |      0.176 ns |    1 |   0.0027 |        - |        - |      48 B |
+| Small_Ascii            |       695.75 ns |      4.394 ns |      3.669 ns |    3 |   0.0124 |        - |        - |     224 B |
+| Small_Mixed            |       686.54 ns |      8.868 ns |      8.295 ns |    3 |   0.0124 |        - |        - |     224 B |
+| Medium_Ascii           |     5,994.68 ns |     32.905 ns |     30.779 ns |    4 |   0.4730 |        - |        - |    8240 B |
+| Medium_Mixed           |     7,116.65 ns |     27.489 ns |     22.955 ns |    5 |   0.4730 |        - |        - |    8264 B |
+| Large_Ascii            |   593,733.29 ns |  2,040.378 ns |  1,703.808 ns |    7 | 249.0234 | 249.0234 | 249.0234 |  819332 B |
+| Large_Mixed            | 1,066,297.43 ns |  8,507.650 ns |  7,958.061 ns |    8 | 248.0469 | 248.0469 | 248.0469 |  823523 B |
+| Large_WorstCase        | 2,148,169.56 ns | 16,455.374 ns | 15,392.367 ns |    9 | 246.0938 | 246.0938 | 246.0938 | 1024125 B |
+| CharArray_Medium_Mixed |     7,357.24 ns |     59.719 ns |     55.861 ns |    6 |   0.5951 |   0.0076 |        - |   10336 B |
+
+## Notes
+
+- Baseline before SIMD refactor
+- Used as comparison target for Task 7
+- Original implementation uses 3,631-line switch statement for character mappings
+- All benchmarks allocate new strings on every call
+- Large_WorstCase (Cyrillic text) is the slowest at ~2.1ms for 100KB
+
+## Key Observations
+
+1. **Pure ASCII performance**: 82.81 ns for 10 characters, 593 µs for 100KB
+2. **Mixed content performance**: 71.05 ns for 10 characters, 1.07 ms for 100KB
+3. **Worst case (Cyrillic)**: 2.15 ms for 100KB (2x slower than mixed)
+4. **Memory allocation**: Linear with input size, plus overhead for output string
+5. **GC pressure**: Significant Gen0/Gen1/Gen2 collections on large inputs
--- a/tests/Umbraco.Tests.Benchmarks/BenchmarkTextGenerator.cs
+++ b/tests/Umbraco.Tests.Benchmarks/BenchmarkTextGenerator.cs
@@ -0,0 +1,63 @@
+using System.Text;
+
+namespace Umbraco.Tests.Benchmarks;
+
+public static class BenchmarkTextGenerator
+{
+    private const int Seed = 42;
+
+    private static readonly char[] AsciiAlphaNum =
+        "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789".ToCharArray();
+
+    private static readonly char[] AsciiPunctuation =
+        " .,;:!?-_'\"()".ToCharArray();
+
+    private static readonly char[] LatinAccented =
+        "àáâãäåæèéêëìíîïñòóôõöøùúûüýÿÀÁÂÃÄÅÆÈÉÊËÌÍÎÏÑÒÓÔÕÖØÙÚÛÜÝŸœŒßðÐþÞ".ToCharArray();
+
+    private static readonly char[] Cyrillic =
+        "АБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЫЬЭЮЯабвгдеёжзийклмнопрстуфхцчшщъыьэюя".ToCharArray();
+
+    private static readonly char[] Symbols =
+        "©®™€£¥°±×÷§¶†‡•".ToCharArray();
+
+    private static readonly char[] WorstCaseCyrillic =
+        "ЩЮЯЖЧШщюяжчш".ToCharArray();
+
+    public static string GeneratePureAscii(int length) =>
+        GenerateFromCharset(length, AsciiAlphaNum);
+
+    public static string GenerateMixed(int length)
+    {
+        var random = new Random(Seed);
+        var sb = new StringBuilder(length);
+
+        for (int i = 0; i < length; i++)
+        {
+            var roll = random.Next(100);
+            var charset = roll switch
+            {
+                < 70 => AsciiAlphaNum,
+                < 85 => AsciiPunctuation,
+                < 95 => LatinAccented,
+                < 99 => Cyrillic,
+                _ => Symbols
+            };
+            sb.Append(charset[random.Next(charset.Length)]);
+        }
+
+        return sb.ToString();
+    }
+
+    public static string GenerateWorstCase(int length) =>
+        GenerateFromCharset(length, WorstCaseCyrillic);
+
+    private static string GenerateFromCharset(int length, char[] charset)
+    {
+        var random = new Random(Seed);
+        var sb = new StringBuilder(length);
+        for (int i = 0; i < length; i++)
+            sb.Append(charset[random.Next(charset.Length)]);
+        return sb.ToString();
+    }
+}
--- a/tests/Umbraco.Tests.Benchmarks/Utf8ToAsciiConverterBaselineBenchmarks.cs
+++ b/tests/Umbraco.Tests.Benchmarks/Utf8ToAsciiConverterBaselineBenchmarks.cs
@@ -0,0 +1,52 @@
+using BenchmarkDotNet.Attributes;
+using BenchmarkDotNet.Columns;
+using BenchmarkDotNet.Jobs;
+using Umbraco.Cms.Core.Strings;
+
+namespace Umbraco.Tests.Benchmarks;
+
+[MemoryDiagnoser]
+[RankColumn]
+[StatisticalTestColumn]
+public class Utf8ToAsciiConverterBaselineBenchmarks
+{
+    private static readonly string TinyAscii = BenchmarkTextGenerator.GeneratePureAscii(10);
+    private static readonly string TinyMixed = BenchmarkTextGenerator.GenerateMixed(10);
+    private static readonly string SmallAscii = BenchmarkTextGenerator.GeneratePureAscii(100);
+    private static readonly string SmallMixed = BenchmarkTextGenerator.GenerateMixed(100);
+    private static readonly string MediumAscii = BenchmarkTextGenerator.GeneratePureAscii(1024);
+    private static readonly string MediumMixed = BenchmarkTextGenerator.GenerateMixed(1024);
+    private static readonly string LargeAscii = BenchmarkTextGenerator.GeneratePureAscii(100 * 1024);
+    private static readonly string LargeMixed = BenchmarkTextGenerator.GenerateMixed(100 * 1024);
+    private static readonly string LargeWorstCase = BenchmarkTextGenerator.GenerateWorstCase(100 * 1024);
+
+    [Benchmark]
+    public string Tiny_Ascii() => Utf8ToAsciiConverter.ToAsciiString(TinyAscii);
+
+    [Benchmark]
+    public string Tiny_Mixed() => Utf8ToAsciiConverter.ToAsciiString(TinyMixed);
+
+    [Benchmark]
+    public string Small_Ascii() => Utf8ToAsciiConverter.ToAsciiString(SmallAscii);
+
+    [Benchmark]
+    public string Small_Mixed() => Utf8ToAsciiConverter.ToAsciiString(SmallMixed);
+
+    [Benchmark]
+    public string Medium_Ascii() => Utf8ToAsciiConverter.ToAsciiString(MediumAscii);
+
+    [Benchmark]
+    public string Medium_Mixed() => Utf8ToAsciiConverter.ToAsciiString(MediumMixed);
+
+    [Benchmark]
+    public string Large_Ascii() => Utf8ToAsciiConverter.ToAsciiString(LargeAscii);
+
+    [Benchmark]
+    public string Large_Mixed() => Utf8ToAsciiConverter.ToAsciiString(LargeMixed);
+
+    [Benchmark]
+    public string Large_WorstCase() => Utf8ToAsciiConverter.ToAsciiString(LargeWorstCase);
+
+    [Benchmark]
+    public char[] CharArray_Medium_Mixed() => Utf8ToAsciiConverter.ToAsciiCharArray(MediumMixed);
+}