Adjust dedup table bits

src/Combination.StringPools/StringPool.cs

@@ -6,6 +6,7 @@ namespace Combination.StringPools;
 public static class StringPool
 {
     private const int DefaultPageSize = 16384; // Size of a pool memory page in bytes
+    private const int DefaultDeduplicationTableBits = 10; // Size of a pool memory page in bytes
 
 
     /// <summary>
@@ -20,7 +21,18 @@ public static class StringPool
     /// <param name="pageSize">The size of the memory pages to allocate (allocation increment quantum).</param>
     /// <param name="initialPageCount">The initial number of pages to allocate.</param>
     /// <returns>The newly create string pool.</returns>
-    public static IUtf8StringPool Utf8(int pageSize, int initialPageCount) => new Utf8StringPool(pageSize, initialPageCount, false);
+    public static IUtf8StringPool Utf8(int pageSize, int initialPageCount) => Utf8(pageSize, initialPageCount, DefaultDeduplicationTableBits);
+
+    /// <summary>
+    /// A Utf8 string pool that does not deduplicate strings, created with specified page size and initial page count.
+    /// </summary>
+    /// <param name="pageSize">The size of the memory pages to allocate (allocation increment quantum).</param>
+    /// <param name="initialPageCount">The initial number of pages to allocate.</param>
+    /// <param name="deduplicationTableBits">Number of bits in the deduplication table (size will be 2^bits).</param>
+    /// <returns>The newly create string pool.</returns>
+    public static IUtf8StringPool Utf8(int pageSize, int initialPageCount, int deduplicationTableBits) =>
+        new Utf8StringPool(pageSize, initialPageCount, false, deduplicationTableBits);
+
 
     /// <summary>
     /// A Utf8 string pool that de-duplicates equal strings, created with default page size and initial page count.
@@ -34,7 +46,18 @@ public static class StringPool
     /// <param name="pageSize">The size of the memory pages to allocate (allocation increment quantum).</param>
     /// <param name="initialPageCount">The initial number of pages to allocate.</param>
     /// <returns>The newly create string pool.</returns>
-    public static IUtf8DeduplicatedStringPool DeduplicatedUtf8(int pageSize, int initialPageCount) => new Utf8StringPool(pageSize, initialPageCount, true);
+    public static IUtf8DeduplicatedStringPool DeduplicatedUtf8(int pageSize, int initialPageCount) =>
+        DeduplicatedUtf8(pageSize, initialPageCount, DefaultDeduplicationTableBits);
+
+    /// <summary>
+    /// A Utf8 string pool that de-duplicates equal strings, created with specified page size and initial page count.
+    /// </summary>
+    /// <param name="pageSize">The size of the memory pages to allocate (allocation increment quantum).</param>
+    /// <param name="initialPageCount">The initial number of pages to allocate.</param>
+    /// <param name="deduplicationTableBits">Number of bits in the deduplication table (size will be 2^bits).</param>
+    /// <returns>The newly create string pool.</returns>
+    public static IUtf8DeduplicatedStringPool DeduplicatedUtf8(int pageSize, int initialPageCount, int deduplicationTableBits) =>
+        new Utf8StringPool(pageSize, initialPageCount, true, deduplicationTableBits);
 
     /// <summary>
     /// The total number of bytes allocated in all string pools.

src/Combination.StringPools/Utf8StringPool.cs

@@ -11,7 +11,6 @@ private const int
         PoolIndexBits =
             24; // Number of bits to use for pool index in handle (more bits = more pools, but less strings per pool)
 
-    private const int DeduplicationTableBits = 10; // Number of bits to use for deduplication table (2^bits entries)
     private static readonly List<Utf8StringPool?> Pools = new();
     internal static long totalAllocatedBytes, totalUsedBytes, totalAddedBytes;
 
@@ -21,28 +20,35 @@ private const int
 
     private readonly List<ulong>?[]? deduplicationTable;
     private readonly DisposeLock disposeLock = new();
+    private readonly int deduplicationTableBits = 10; // Number of bits to use for deduplication table (2^bits entries)
     private readonly int pageSize;
 
     private readonly object writeLock = new();
 
     // ReSharper disable once MemberCanBePrivate.Global
-    public Utf8StringPool(int pageSize, int initialPageCount, bool deduplicateStrings)
+    public Utf8StringPool(int pageSize, int initialPageCount, bool deduplicateStrings, int deduplicationTableBits)
     {
         if (pageSize < 16)
         {
             // We need at least 16 bytes to store the length of the string and an actual string
             throw new ArgumentOutOfRangeException(nameof(pageSize));
         }
 
+        if (deduplicationTableBits < 2 || deduplicationTableBits > 24)
+        {
+            throw new ArgumentOutOfRangeException(nameof(deduplicationTableBits));
+        }
+
         if (initialPageCount < 0)
         {
             throw new ArgumentOutOfRangeException(nameof(initialPageCount));
         }
 
+        this.deduplicationTableBits = deduplicationTableBits;
         this.pageSize = pageSize;
         if (deduplicateStrings)
         {
-            deduplicationTable = new List<ulong>?[1 << DeduplicationTableBits];
+            deduplicationTable = new List<ulong>?[1 << deduplicationTableBits];
         }
 
         bool didAlloc;
@@ -202,7 +208,7 @@ private bool TryDeduplicate(int stringHash, string value, out ulong offset)
                 return false;
             }
 
-            var tableIndex = stringHash & ((1 << DeduplicationTableBits) - 1);
+            var tableIndex = stringHash & ((1 << deduplicationTableBits) - 1);
             var table = deduplicationTable[tableIndex];
             if (table is null)
             {
@@ -233,7 +239,7 @@ private void AddToDeduplicationTable(int stringHash, ulong handle)
             return;
         }
 
-        var tableIndex = stringHash & ((1 << DeduplicationTableBits) - 1);
+        var tableIndex = stringHash & ((1 << deduplicationTableBits) - 1);
         var table = deduplicationTable[tableIndex] ?? (deduplicationTable[tableIndex] = new List<ulong>());
 
         table.Add(handle);