Reduce footprint of ProbMap SearchValues

dotnet · May 15, 2024 · c29dd3e · c29dd3e
1 parent 5b68877
commit c29dd3e
Show file tree

Hide file tree

Showing 4 changed files with 16 additions and 8 deletions.
diff --git a/...libraries/System.Private.CoreLib/src/System/SearchValues/ProbabilisticCharSearchValues.cs b/...libraries/System.Private.CoreLib/src/System/SearchValues/ProbabilisticCharSearchValues.cs
@@ -9,15 +9,14 @@ namespace System.Buffers
     internal sealed class ProbabilisticCharSearchValues : SearchValues<char>
     {
         private ProbabilisticMapState _map;
-        private readonly string _values;
 
         public ProbabilisticCharSearchValues(ReadOnlySpan<char> values, int maxInclusive)
         {
-            _values = new string(values);
             _map = new ProbabilisticMapState(values, maxInclusive);
         }
 
-        internal override char[] GetValues() => _values.ToCharArray();
+        internal override char[] GetValues() =>
+            _map.GetValues();
 
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         internal override bool ContainsCore(char value) =>

diff --git a/src/libraries/System.Private.CoreLib/src/System/SearchValues/ProbabilisticMapState.cs b/src/libraries/System.Private.CoreLib/src/System/SearchValues/ProbabilisticMapState.cs
@@ -61,6 +61,16 @@ public unsafe ProbabilisticMapState(ReadOnlySpan<char>* valuesPtr)
             _slowContainsValuesPtr = valuesPtr;
         }
 
+        public char[] GetValues()
+        {
+            Debug.Assert(_hashEntries is not null);
+
+            var unique = new HashSet<char>(_hashEntries);
+            char[] values = new char[unique.Count];
+            unique.CopyTo(values);
+            return values;
+        }
+
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public bool FastContains(char value)
         {

diff --git a/.../System.Private.CoreLib/src/System/SearchValues/ProbabilisticWithAsciiCharSearchValues.cs b/.../System.Private.CoreLib/src/System/SearchValues/ProbabilisticWithAsciiCharSearchValues.cs
@@ -16,7 +16,6 @@ internal sealed class ProbabilisticWithAsciiCharSearchValues<TOptimizations> : S
         private IndexOfAnyAsciiSearcher.AsciiState _asciiState;
         private IndexOfAnyAsciiSearcher.AsciiState _inverseAsciiState;
         private ProbabilisticMapState _map;
-        private readonly string _values;
 
         public ProbabilisticWithAsciiCharSearchValues(ReadOnlySpan<char> values, int maxInclusive)
         {
@@ -26,11 +25,11 @@ public ProbabilisticWithAsciiCharSearchValues(ReadOnlySpan<char> values, int max
             IndexOfAnyAsciiSearcher.ComputeAsciiState(values, out _asciiState);
             _inverseAsciiState = _asciiState.CreateInverse();
 
-            _values = new string(values);
-            _map = new ProbabilisticMapState(_values, maxInclusive);
+            _map = new ProbabilisticMapState(values, maxInclusive);
         }
 
-        internal override char[] GetValues() => _values.ToCharArray();
+        internal override char[] GetValues() =>
+            _map.GetValues();
 
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         internal override bool ContainsCore(char value) =>

diff --git a/src/libraries/System.Private.CoreLib/src/System/SearchValues/SearchValues.cs b/src/libraries/System.Private.CoreLib/src/System/SearchValues/SearchValues.cs
@@ -195,7 +195,7 @@ static bool ShouldUseProbabilisticMap(int valuesLength, int maxInclusive)
 
                 // The probabilistic map is more memory efficient for spare sets, while the bitmap is more efficient for dense sets.
                 int bitmapFootprintBytesEstimate = 64 + (maxInclusive / 8);
-                int probabilisticFootprintBytesEstimate = 128 + (valuesLength * 6);
+                int probabilisticFootprintBytesEstimate = 128 + (valuesLength * 4);
 
                 // The bitmap is a bit faster than the perfect hash checks employed by the probabilistic map.
                 // Sacrifice some memory usage for faster lookups.