Merge pull request #4 from simdutf/scalar_getFirstNonASCII

scalar GetIndexOfFirstNonAsciiByte

Author: Nick-Nuon

benchmark/Benchmark.cs

@@ -15,13 +15,20 @@ namespace SimdUnicodeBenchmarks
     public class Checker
     {
         List<char[]> names;
-        List<bool> results;
-        public static bool RuntimeIsAsciiApproach(ReadOnlySpan<char> s)
-        {
-            // The runtime as of NET 8.0 has a dedicated method for this, but
-            // it is not available prior to that, so let us branch.
+        List<byte[]> AsciiBytes;
+        List<char[]> nonAsciichars;
+        public List<byte[]> nonAsciiBytes; // Declare at the class level
+
+        List<bool> results;
+
+        public static bool RuntimeIsAsciiApproach(ReadOnlySpan<char> s)
+        {
+
+            // The runtime as of NET 8.0 has a dedicated method for this, but
+            // it is not available prior to that, so let us branch.
 #if NET8_0_OR_GREATER
-            return Ascii.IsValid(s);
+            return System.Text.Ascii.IsValid(s);
+
 #else
             foreach (char c in s)
             {
@@ -34,6 +41,8 @@ public static bool RuntimeIsAsciiApproach(ReadOnlySpan<char> s)
             return true;
 #endif
         }
+
+
         public static char[] GetRandomASCIIString(uint n)
         {
             var allowedChars = "abcdefghijkmnopqrstuvwxyzABCDEFGHJKLMNOPQRSTUVWXYZ01234567é89";
@@ -49,23 +58,49 @@ public static char[] GetRandomASCIIString(uint n)
             return chars;
         }
 
+        public static char[] GetRandomNonASCIIString(uint n)
+        {
+            // Chose a few Latin Extended-A and Latin Extended-B characters alongside ASCII chars
+            var allowedChars = "abcdefghijkmnopqrstuvwxyzABCDEFGHJKLMNOPQRSTUVWXYZ01234567é89šžŸũŭůűųŷŹźŻżŽ";
+
+            var chars = new char[n];
+            var rd = new Random(12345); // fixed seed
 
-        [Params(100, 200, 500)]
+            for (var i = 0; i < n; i++)
+            {
+                chars[i] = allowedChars[rd.Next(0, allowedChars.Length)];
+            }
+
+            return chars;
+        }
+
+
+
+        [Params(100, 200, 500, 1000, 2000)]
         public uint N;
 
+
         [GlobalSetup]
         public void Setup()
         {
             names = new List<char[]>();
+            nonAsciiBytes = new List<byte[]>(); // Initialize the list of byte arrays
             results = new List<bool>();
 
             for (int i = 0; i < 100; i++)
             {
                 names.Add(GetRandomASCIIString(N));
+                char[] nonAsciiChars = GetRandomNonASCIIString(N);
+                nonAsciiBytes.Add(Encoding.UTF8.GetBytes(nonAsciiChars));  // Convert to byte array and store
                 results.Add(false);
             }
+
+            AsciiBytes = names
+                .Select(name => System.Text.Encoding.ASCII.GetBytes(name))
+                .ToList();
         }
 
+
         [Benchmark]
         public void FastUnicodeIsAscii()
         {
@@ -98,7 +133,65 @@ public void RuntimeIsAscii()
                 count += 1;
             }
         }
+        [Benchmark]
+        public void Error_GetIndexOfFirstNonAsciiByte()
+        {
+            foreach (byte[] nonAsciiByte in nonAsciiBytes)  // Use nonAsciiBytes directly
+            {
+                unsafe
+                {
+                    fixed (byte* pNonAscii = nonAsciiByte)
+                    {
+                        nuint result = SimdUnicode.Ascii.GetIndexOfFirstNonAsciiByte(pNonAscii, (nuint)nonAsciiByte.Length);
+                    }
+                }
+            }
+        }
+
+        [Benchmark]
+        public void Error_Runtime_GetIndexOfFirstNonAsciiByte()
+        {
+            foreach (byte[] nonAsciiByte in nonAsciiBytes)  // Use nonAsciiBytes directly
+            {
+                unsafe
+                {
+                    fixed (byte* pNonAscii = nonAsciiByte)
+                    {
+                        nuint result = Competition.Ascii.GetIndexOfFirstNonAsciiByte(pNonAscii, (nuint)nonAsciiByte.Length);
+                    }
+                }
+            }
+        }
 
+        [Benchmark]
+        public void allAscii_GetIndexOfFirstNonAsciiByte()
+        {
+            foreach (byte[] Abyte in AsciiBytes)  // Use nonAsciiBytes directly
+            {
+                unsafe
+                {
+                    fixed (byte* pNonAscii = Abyte)
+                    {
+                        nuint result = SimdUnicode.Ascii.GetIndexOfFirstNonAsciiByte(pNonAscii, (nuint)Abyte.Length);
+                    }
+                }
+            }
+        }
+
+        [Benchmark]
+        public void allAscii_Runtime_GetIndexOfFirstNonAsciiByte()
+        {
+            foreach (byte[] Abyte in AsciiBytes)  // Use nonAsciiBytes directly
+            {
+                unsafe
+                {
+                    fixed (byte* pNonAscii = Abyte)
+                    {
+                        nuint result = Competition.Ascii.GetIndexOfFirstNonAsciiByte(pNonAscii, (nuint)Abyte.Length);
+                    }
+                }
+            }
+        }
     }
 
     public class Program