Initial BitTree implementation added.

2026-01-09 16:13:16 +01:00
1 parent 45e3455ff3
commit d88ad25476
@@ -5,19 +5,20 @@ using PerfectNumbers.Core.Gpu;

 namespace EvenPerfectBitScanner;

-	internal enum ByDivisorKIncrementMode
-	{
-		Sequential,
-		Pow2Groups,
-		Predictive,
-		Percentile,
-	Additive,
-	TopDown,
-	BitContradiction,
+internal enum ByDivisorKIncrementMode
+{
+    Sequential,
+    Pow2Groups,
+    Predictive,
+    Percentile,
+    Additive,
+    TopDown,
+    BitContradiction,
+    BitTree,
 }

-	internal readonly struct CliArguments
-	{
+internal readonly struct CliArguments
+{
    private const string DefaultCyclesPath = "divisor_cycles.bin";

    internal readonly ulong StartPrime;
@@ -35,19 +36,19 @@ namespace EvenPerfectBitScanner;
    internal readonly bool UseLucas;
    internal readonly bool UseResidue;
    internal readonly bool UseDivisor;
-	internal readonly bool UseByDivisor;
-	internal readonly ByDivisorKIncrementMode ByDivisorKIncrementMode;
-	internal readonly int ByDivisorSpecialRange;
-	internal readonly bool UseGcdFilter;
-	internal readonly bool TestMode;
-	internal readonly bool UseGpuCycles;
-	internal readonly ComputationDevice MersenneDevice;
-	internal readonly ComputationDevice OrderDevice;
-	internal readonly int ScanBatchSize;
-	internal readonly int SliceSize;
-	internal readonly ulong MaxKLimit;
+    internal readonly bool UseByDivisor;
+    internal readonly ByDivisorKIncrementMode ByDivisorKIncrementMode;
+    internal readonly int ByDivisorSpecialRange;
+    internal readonly bool UseGcdFilter;
+    internal readonly bool TestMode;
+    internal readonly bool UseGpuCycles;
+    internal readonly ComputationDevice MersenneDevice;
+    internal readonly ComputationDevice OrderDevice;
+    internal readonly int ScanBatchSize;
+    internal readonly int SliceSize;
+    internal readonly ulong MaxKLimit;
    internal readonly bool MaxKConfigured;
-	internal readonly string FilterFile;
+    internal readonly string FilterFile;
    internal readonly bool ForcePrimeKernelsOnCpu;
    internal readonly ModReductionMode ModReductionMode;
    internal readonly NttBackend NttBackend;
@@ -66,56 +67,56 @@ namespace EvenPerfectBitScanner;
    internal readonly double ZeroFractionConjecture;
    internal readonly int MaxZeroConjecture;
    internal readonly BigInteger MinK;
-	internal readonly int GpuRatio;
+    internal readonly int GpuRatio;

-	private CliArguments(
-            ulong startPrime,
-            bool startPrimeProvided,
-            UInt128 maxPrimeLimit,
-            bool maxPrimeConfigured,
-            int threadCount,
-            int blockSize,
-            int gpuPrimeThreads,
-            int gpuPrimeBatch,
-            GpuKernelType kernelType,
-            bool useOrder,
-            bool showHelp,
-            bool useBitTransform,
-            bool useLucas,
-            bool useResidue,
-            bool useDivisor,
-            bool useByDivisor,
-            bool useGcdFilter,
-            ByDivisorKIncrementMode byDivisorKIncrementMode,
-            int byDivisorSpecialRange,
-            bool testMode,
-            bool useGpuCycles,
-            ComputationDevice mersenneDevice,
-            ComputationDevice orderDevice,
-            int scanBatchSize,
-            int sliceSize,
-            ulong maxKLimit,
-            bool maxKConfigured,
-            string filterFile,
-            bool forcePrimeKernelsOnCpu,
-            ModReductionMode modReductionMode,
-            NttBackend nttBackend,
-            string? resultsDirectory,
-            string? resultsPrefix,
-            int writeBatchSize,
-            string cyclesPath,
-            bool continueCyclesGeneration,
-            int cyclesBatchSize,
-            ulong divisorCyclesSearchLimit,
-            ulong? orderWarmupLimitOverride,
-            string? rleBlacklistPath,
-            ulong rleHardMaxP,
-            bool rleOnlyLast7,
-            double zeroFractionHard,
-            double zeroFractionConjecture,
-            int maxZeroConjecture,
-            BigInteger minK,
-			int gpuRatio)
+    private CliArguments(
+        ulong startPrime,
+        bool startPrimeProvided,
+        UInt128 maxPrimeLimit,
+        bool maxPrimeConfigured,
+        int threadCount,
+        int blockSize,
+        int gpuPrimeThreads,
+        int gpuPrimeBatch,
+        GpuKernelType kernelType,
+        bool useOrder,
+        bool showHelp,
+        bool useBitTransform,
+        bool useLucas,
+        bool useResidue,
+        bool useDivisor,
+        bool useByDivisor,
+        bool useGcdFilter,
+        ByDivisorKIncrementMode byDivisorKIncrementMode,
+        int byDivisorSpecialRange,
+        bool testMode,
+        bool useGpuCycles,
+        ComputationDevice mersenneDevice,
+        ComputationDevice orderDevice,
+        int scanBatchSize,
+        int sliceSize,
+        ulong maxKLimit,
+        bool maxKConfigured,
+        string filterFile,
+        bool forcePrimeKernelsOnCpu,
+        ModReductionMode modReductionMode,
+        NttBackend nttBackend,
+        string? resultsDirectory,
+        string? resultsPrefix,
+        int writeBatchSize,
+        string cyclesPath,
+        bool continueCyclesGeneration,
+        int cyclesBatchSize,
+        ulong divisorCyclesSearchLimit,
+        ulong? orderWarmupLimitOverride,
+        string? rleBlacklistPath,
+        ulong rleHardMaxP,
+        bool rleOnlyLast7,
+        double zeroFractionHard,
+        double zeroFractionConjecture,
+        int maxZeroConjecture,
+        BigInteger minK,
+        int gpuRatio)
    {
        StartPrime = startPrime;
        StartPrimeProvided = startPrimeProvided;
@@ -163,7 +164,7 @@ namespace EvenPerfectBitScanner;
        ZeroFractionConjecture = zeroFractionConjecture;
        MaxZeroConjecture = maxZeroConjecture;
        MinK = minK;
-		GpuRatio = gpuRatio;
+        GpuRatio = gpuRatio;
    }

    internal bool UseFilter => !TestMode && !string.IsNullOrEmpty(FilterFile);
@@ -204,7 +205,7 @@ namespace EvenPerfectBitScanner;
        string? resultsDirectory = null;
        string? resultsPrefix = null;
        int writeBatchSize = CalculationResultHandler.DefaultWriteBatchSize;
-            string cyclesPath = DefaultCyclesPath;
+        string cyclesPath = DefaultCyclesPath;
        bool continueCyclesGeneration = false;
        int cyclesBatchSize = 512;
        ulong divisorCyclesSearchLimit = PerfectNumberConstants.ExtraDivisorCycleSearchLimit;
@@ -216,7 +217,7 @@ namespace EvenPerfectBitScanner;
        double zeroFractionConjecture = -1.0;
        int maxZeroConjecture = -1;
        BigInteger minK = BigInteger.One;
-		int gpuRatio = PerfectNumberConstants.GpuRatio;
+        int gpuRatio = PerfectNumberConstants.GpuRatio;

        foreach (string argument in args)
        {
@@ -279,6 +280,19 @@ namespace EvenPerfectBitScanner;
                continue;
            }

+            if (argument.Equals("--use-order", StringComparison.OrdinalIgnoreCase))
+            {
+                useOrder = true;
+                continue;
+            }
+
+            if (argument.StartsWith("--max-k=", StringComparison.OrdinalIgnoreCase))
+            {
+                maxKLimit = Utf8CliParser.ParseUInt64(argument.AsSpan("--max-k=".Length));
+                maxKConfigured = true;
+                continue;
+            }
+
            if (argument.StartsWith("--mersenne=", StringComparison.OrdinalIgnoreCase))
            {
                ReadOnlySpan<char> value = argument.AsSpan("--mersenne=".Length);
@@ -318,138 +332,46 @@ namespace EvenPerfectBitScanner;

            if (argument.StartsWith("--bydivisor-k-increment=", StringComparison.OrdinalIgnoreCase))
            {
-                string value = argument["--bydivisor-k-increment=".Length..].Trim();
+                ReadOnlySpan<char> value = argument.AsSpan("--bydivisor-k-increment=".Length);
                if (value.Equals("pow2groups", StringComparison.OrdinalIgnoreCase))
                {
                    byDivisorKIncrementMode = ByDivisorKIncrementMode.Pow2Groups;
                }
-                else if (value.Equals("sequential", StringComparison.OrdinalIgnoreCase))
-                {
-                    byDivisorKIncrementMode = ByDivisorKIncrementMode.Sequential;
-                }
                else if (value.Equals("predictive", StringComparison.OrdinalIgnoreCase))
                {
                    byDivisorKIncrementMode = ByDivisorKIncrementMode.Predictive;
                }
-				else if (value.Equals("percentile", StringComparison.OrdinalIgnoreCase))
-				{
-					byDivisorKIncrementMode = ByDivisorKIncrementMode.Percentile;
-				}
-				else if (value.Equals("additive", StringComparison.OrdinalIgnoreCase))
-				{
-					byDivisorKIncrementMode = ByDivisorKIncrementMode.Additive;
-				}
-				else if (value.Equals("topdown", StringComparison.OrdinalIgnoreCase))
-				{
-					byDivisorKIncrementMode = ByDivisorKIncrementMode.TopDown;
-				}
-				else if (value.Equals("bitcontradiction", StringComparison.OrdinalIgnoreCase))
-				{
-					byDivisorKIncrementMode = ByDivisorKIncrementMode.BitContradiction;
-				}
-				else
-				{
-					Console.WriteLine("Invalid value for --bydivisor-k-increment. Use sequential, pow2groups, predictive, percentile, additive, topdown, or bitcontradiction.");
-					return default;
-				}
+                else if (value.Equals("percentile", StringComparison.OrdinalIgnoreCase))
+                {
+                    byDivisorKIncrementMode = ByDivisorKIncrementMode.Percentile;
+                }
+                else if (value.Equals("additive", StringComparison.OrdinalIgnoreCase))
+                {
+                    byDivisorKIncrementMode = ByDivisorKIncrementMode.Additive;
+                }
+                else if (value.Equals("topdown", StringComparison.OrdinalIgnoreCase))
+                {
+                    byDivisorKIncrementMode = ByDivisorKIncrementMode.TopDown;
+                }
+                else if (value.Equals("bitcontradiction", StringComparison.OrdinalIgnoreCase))
+                {
+                    byDivisorKIncrementMode = ByDivisorKIncrementMode.BitContradiction;
+                }
+                else if (value.Equals("bittree", StringComparison.OrdinalIgnoreCase))
+                {
+                    byDivisorKIncrementMode = ByDivisorKIncrementMode.BitTree;
+                }
+                else
+                {
+                    byDivisorKIncrementMode = ByDivisorKIncrementMode.Sequential;
+                }

                continue;
            }

            if (argument.StartsWith("--bydivisor-special-range=", StringComparison.OrdinalIgnoreCase))
            {
-                if (Utf8CliParser.TryParseInt32(argument.AsSpan("--bydivisor-special-range=".Length), out int parsedRange) && parsedRange >= 0)
-                {
-                    byDivisorSpecialRange = parsedRange;
-                }
-                else
-                {
-                    Console.WriteLine("Invalid value for --bydivisor-special-range. Use a non-negative integer.");
-                    return default;
-                }
-
-                continue;
-            }
-
-            if (argument.StartsWith("--divisor-cycles-limit=", StringComparison.OrdinalIgnoreCase))
-            {
-                if (Utf8CliParser.TryParseUInt64(argument.AsSpan("--divisor-cycles-limit=".Length), out ulong parsedLimit))
-                {
-                    divisorCyclesSearchLimit = parsedLimit;
-                }
-
-                continue;
-            }
-
-            if (argument.StartsWith("--max-k=", StringComparison.OrdinalIgnoreCase))
-            {
-                if (Utf8CliParser.TryParseUInt64(argument.AsSpan("--max-k=".Length), out ulong parsedMaxK))
-                {
-                    maxKLimit = parsedMaxK;
-                    maxKConfigured = true;
-                }
-
-                continue;
-            }
-
-            if (argument.StartsWith("--min-k=", StringComparison.OrdinalIgnoreCase))
-            {
-                ReadOnlySpan<char> minKSpan = argument.AsSpan("--min-k=".Length);
-                if (Utf8CliParser.TryParseUInt64(minKSpan, out ulong parsedMinK) && parsedMinK > 0UL)
-                {
-                    minK = parsedMinK;
-                }
-                else
-                {
-                    try
-                    {
-                        BigInteger parsedMinKBig = BigInteger.Parse(minKSpan, CultureInfo.InvariantCulture);
-                        if (parsedMinKBig > BigInteger.Zero)
-                        {
-                            minK = parsedMinKBig;
-                        }
-                        else
-                        {
-                            Console.WriteLine("Invalid value for --min-k.");
-                            return default;
-                        }
-                    }
-                    catch
-                    {
-                        Console.WriteLine("Invalid value for --min-k.");
-                        return default;
-                    }
-                }
-
-                continue;
-            }
-
-            if (argument.StartsWith("--mersenne-device=", StringComparison.OrdinalIgnoreCase))
-            {
-                ReadOnlySpan<char> value = argument.AsSpan("--mersenne-device=".Length);
-                mersenneDevice = ParseDevice(value);
-                continue;
-            }
-
-            if (argument.Equals("--test", StringComparison.OrdinalIgnoreCase))
-            {
-                testMode = true;
-                continue;
-            }
-
-            if (argument.Equals("--use-order", StringComparison.OrdinalIgnoreCase))
-            {
-                useOrder = true;
-                continue;
-            }
-
-            if (argument.StartsWith("--order-warmup-limit=", StringComparison.OrdinalIgnoreCase))
-            {
-                if (Utf8CliParser.TryParseUInt64(argument.AsSpan("--order-warmup-limit=".Length), out ulong parsedLimit))
-                {
-                    orderWarmupLimitOverride = parsedLimit;
-                }
-
+                byDivisorSpecialRange = Utf8CliParser.ParseInt32(argument.AsSpan("--bydivisor-special-range=".Length));
                continue;
            }

@@ -459,138 +381,30 @@ namespace EvenPerfectBitScanner;
                continue;
            }

-            if (argument.StartsWith("--ntt=", StringComparison.OrdinalIgnoreCase))
+            if (argument.Equals("--test-mode", StringComparison.OrdinalIgnoreCase))
            {
-                ReadOnlySpan<char> value = argument.AsSpan("--ntt=".Length);
-                nttBackend = value.Equals("staged", StringComparison.OrdinalIgnoreCase) ? NttBackend.Staged : NttBackend.Reference;
+                testMode = true;
                continue;
            }

-            if (argument.StartsWith("--mod-reduction=", StringComparison.OrdinalIgnoreCase))
+            if (argument.StartsWith("--mersenne-device=", StringComparison.OrdinalIgnoreCase))
            {
-                ReadOnlySpan<char> value = argument.AsSpan("--mod-reduction=".Length);
-                if (value.Equals("mont64", StringComparison.OrdinalIgnoreCase))
-                {
-                    modReductionMode = ModReductionMode.Mont64;
-                }
-                else if (value.Equals("barrett128", StringComparison.OrdinalIgnoreCase))
-                {
-                    modReductionMode = ModReductionMode.Barrett128;
-                }
-                else if (value.Equals("uint128", StringComparison.OrdinalIgnoreCase))
-                {
-                    modReductionMode = ModReductionMode.GpuUInt128;
-                }
-                else
-                {
-                    modReductionMode = ModReductionMode.Auto;
-                }
-
+                mersenneDevice = ParseDevice(argument.AsSpan("--mersenne-device=".Length));
                continue;
            }

-            if (argument.StartsWith("--primes-device=", StringComparison.OrdinalIgnoreCase))
+            if (argument.StartsWith("--primes=", StringComparison.OrdinalIgnoreCase) ||
+                argument.StartsWith("--gpu-kernels=", StringComparison.OrdinalIgnoreCase) ||
+                argument.StartsWith("--accelerator=", StringComparison.OrdinalIgnoreCase))
            {
-                forcePrimeKernelsOnCpu = argument.AsSpan("--primes-device=".Length).Equals("cpu", StringComparison.OrdinalIgnoreCase);
+                ReadOnlySpan<char> value = argument.Contains('=') ? argument.AsSpan(argument.IndexOf('=') + 1) : argument.AsSpan();
+                forcePrimeKernelsOnCpu = value.Equals("cpu", StringComparison.OrdinalIgnoreCase);
                continue;
            }

            if (argument.StartsWith("--order-device=", StringComparison.OrdinalIgnoreCase))
            {
-                ReadOnlySpan<char> value = argument.AsSpan("--order-device=".Length);
-                orderDevice = ParseDevice(value);
-                continue;
-            }
-
-            if (argument.StartsWith("--rle-blacklist=", StringComparison.OrdinalIgnoreCase))
-            {
-                rleBlacklistPath = argument["--rle-blacklist=".Length..];
-                continue;
-            }
-
-            if (argument.StartsWith("--rle-hard-max=", StringComparison.OrdinalIgnoreCase))
-            {
-                if (Utf8CliParser.TryParseUInt64(argument.AsSpan("--rle-hard-max=".Length), out ulong parsedMax))
-                {
-                    rleHardMaxP = parsedMax;
-                }
-
-                continue;
-            }
-
-            if (argument.StartsWith("--rle-only-last7=", StringComparison.OrdinalIgnoreCase))
-            {
-                ReadOnlySpan<char> value = argument.AsSpan("--rle-only-last7=".Length);
-                rleOnlyLast7 = !value.Equals("false", StringComparison.OrdinalIgnoreCase) && !value.Equals("0", StringComparison.OrdinalIgnoreCase);
-                continue;
-            }
-
-            if (argument.StartsWith("--zero-hard=", StringComparison.OrdinalIgnoreCase))
-            {
-                if (Utf8CliParser.TryParseDouble(argument.AsSpan("--zero-hard=".Length), out double zeroFraction))
-                {
-                    zeroFractionHard = zeroFraction;
-                }
-
-                continue;
-            }
-
-            if (argument.StartsWith("--zero-conj=", StringComparison.OrdinalIgnoreCase))
-            {
-                ReadOnlySpan<char> value = argument.AsSpan("--zero-conj=".Length);
-                int colon = value.IndexOf(':');
-                if (colon > 0)
-                {
-                    if (Utf8CliParser.TryParseDouble(value[..colon], out double zeroFrac) &&
-                        Utf8CliParser.TryParseInt32(value[(colon + 1)..], out int maxZero))
-                    {
-                        zeroFractionConjecture = zeroFrac;
-                        maxZeroConjecture = maxZero;
-                    }
-                }
-
-                continue;
-            }
-
-            if (argument.StartsWith("--lucas=", StringComparison.OrdinalIgnoreCase))
-            {
-                mersenneDevice = ParseDevice(argument.AsSpan("--lucas=".Length));
-                continue;
-            }
-
-            if (argument.StartsWith("--primes=", StringComparison.OrdinalIgnoreCase))
-            {
-                forcePrimeKernelsOnCpu = argument.AsSpan("--primes=".Length).Equals("cpu", StringComparison.OrdinalIgnoreCase);
-                continue;
-            }
-
-            if (argument.StartsWith("--gpu-kernels=", StringComparison.OrdinalIgnoreCase))
-            {
-                forcePrimeKernelsOnCpu = argument.AsSpan("--gpu-kernels=".Length).Equals("cpu", StringComparison.OrdinalIgnoreCase);
-                continue;
-            }
-
-            if (argument.StartsWith("--accelerator=", StringComparison.OrdinalIgnoreCase))
-            {
-                forcePrimeKernelsOnCpu = argument.AsSpan("--accelerator=".Length).Equals("cpu", StringComparison.OrdinalIgnoreCase);
-                continue;
-            }
-
-            if (argument.StartsWith("--results-dir=", StringComparison.OrdinalIgnoreCase))
-            {
-                resultsDirectory = argument["--results-dir=".Length..];
-                continue;
-            }
-
-            if (argument.StartsWith("--results-prefix=", StringComparison.OrdinalIgnoreCase))
-            {
-                resultsPrefix = argument["--results-prefix=".Length..];
-                continue;
-            }
-
-            if (argument.StartsWith("--gpu-prime-threads=", StringComparison.OrdinalIgnoreCase))
-            {
-                gpuPrimeThreads = Math.Max(1, Utf8CliParser.ParseInt32(argument.AsSpan("--gpu-prime-threads=".Length)));
+                orderDevice = ParseDevice(argument.AsSpan("--order-device=".Length));
                continue;
            }

@@ -600,6 +414,12 @@ namespace EvenPerfectBitScanner;
                continue;
            }

+            if (argument.StartsWith("--gpu-prime-threads=", StringComparison.OrdinalIgnoreCase))
+            {
+                gpuPrimeThreads = Math.Max(1, Utf8CliParser.ParseInt32(argument.AsSpan("--gpu-prime-threads=".Length)));
+                continue;
+            }
+
            if (argument.StartsWith("--ll-slice=", StringComparison.OrdinalIgnoreCase))
            {
                sliceSize = Math.Max(1, Utf8CliParser.ParseInt32(argument.AsSpan("--ll-slice=".Length)));
@@ -618,6 +438,54 @@ namespace EvenPerfectBitScanner;
                continue;
            }

+            if (argument.StartsWith("--mod-reduction=", StringComparison.OrdinalIgnoreCase))
+            {
+                ReadOnlySpan<char> value = argument.AsSpan("--mod-reduction=".Length);
+                if (value.Equals("uint128", StringComparison.OrdinalIgnoreCase))
+                {
+                    modReductionMode = ModReductionMode.GpuUInt128;
+                }
+                else if (value.Equals("mont64", StringComparison.OrdinalIgnoreCase))
+                {
+                    modReductionMode = ModReductionMode.Mont64;
+                }
+                else if (value.Equals("barrett128", StringComparison.OrdinalIgnoreCase))
+                {
+                    modReductionMode = ModReductionMode.Barrett128;
+                }
+                else
+                {
+                    modReductionMode = ModReductionMode.Auto;
+                }
+
+                continue;
+            }
+
+            if (argument.StartsWith("--ntt=", StringComparison.OrdinalIgnoreCase))
+            {
+                ReadOnlySpan<char> value = argument.AsSpan("--ntt=".Length);
+                nttBackend = value.Equals("staged", StringComparison.OrdinalIgnoreCase) ? NttBackend.Staged : NttBackend.Reference;
+                continue;
+            }
+
+            if (argument.StartsWith("--order-warmup-limit=", StringComparison.OrdinalIgnoreCase))
+            {
+                orderWarmupLimitOverride = Utf8CliParser.ParseUInt64(argument.AsSpan("--order-warmup-limit=".Length));
+                continue;
+            }
+
+            if (argument.StartsWith("--divisor-cycles-limit=", StringComparison.OrdinalIgnoreCase))
+            {
+                divisorCyclesSearchLimit = Utf8CliParser.ParseUInt64(argument.AsSpan("--divisor-cycles-limit=".Length));
+                continue;
+            }
+
+            if (argument.StartsWith("--min-k=", StringComparison.OrdinalIgnoreCase))
+            {
+                minK = BigInteger.Parse(argument.AsSpan("--min-k=".Length), NumberStyles.None, CultureInfo.InvariantCulture);
+                continue;
+            }
+
            if (argument.StartsWith("--filter-p=", StringComparison.OrdinalIgnoreCase))
            {
                filterFile = argument["--filter-p=".Length..];
@@ -651,6 +519,7 @@ namespace EvenPerfectBitScanner;
            if (argument.Equals("--divisor-cycles-continue", StringComparison.OrdinalIgnoreCase))
            {
                continueCyclesGeneration = true;
+                continue;
            }
        }

@@ -659,15 +528,15 @@ namespace EvenPerfectBitScanner;
            startPrimeProvided,
            maxPrimeLimit,
            maxPrimeConfigured,
-                threadCount,
-                blockSize,
-                gpuPrimeThreads,
-                gpuPrimeBatch,
-                kernelType,
-                useOrder,
-                showHelp,
-                useBitTransform,
-                useLucas,
+            threadCount,
+            blockSize,
+            gpuPrimeThreads,
+            gpuPrimeBatch,
+            kernelType,
+            useOrder,
+            showHelp,
+            useBitTransform,
+            useLucas,
            useResidue,
            useDivisor,
            useByDivisor,
@@ -701,7 +570,7 @@ namespace EvenPerfectBitScanner;
            zeroFractionConjecture,
            maxZeroConjecture,
            minK,
-			gpuRatio);
+            gpuRatio);
    }

    internal static void PrintHelp()
@@ -715,14 +584,14 @@ namespace EvenPerfectBitScanner;
        Console.WriteLine("  --increment=bit|add    exponent increment method (default add)");
        Console.WriteLine("  --threads=<value>      number of worker threads");
        Console.WriteLine("  --mersenne=pow2mod|incremental|lucas|residue|divisor|bydivisor  Mersenne test method");
-		Console.WriteLine("  --bydivisor-k-increment=sequential|pow2groups|predictive|percentile|additive|topdown|bitcontradiction  k iteration strategy for --mersenne=bydivisor (default sequential)");
+        Console.WriteLine("  --bydivisor-k-increment=sequential|pow2groups|predictive|percentile|additive|topdown|bitcontradiction|bittree  k iteration strategy for --mersenne=bydivisor (default sequential)");
        Console.WriteLine("  --bydivisor-special-range=<n>  widen pow2groups special values to +/- n percent (default 0 = points)");
        Console.WriteLine("  --max-k=<value> max k for Mersenne tests (q = 2*p*k + 1), including --mersenne=bydivisor");
        Console.WriteLine("  --mersenne-device=cpu|gpu|hybrid  Device for Mersenne method (default gpu)");
        Console.WriteLine("  --primes-device=cpu|gpu device for prime-scan kernels (default gpu)");
-        Console.WriteLine("  --gpu-prime-batch=<n>   batch size for GPU primality sieve (default 262144)");
-        Console.WriteLine($"  --gpu-ratio=<n>      	ratio of GPU operations vs CPU in hybrid device mode (default {PerfectNumberConstants.GpuRatio})");
+        Console.WriteLine($"  --gpu-ratio=<n>          ratio of GPU operations vs CPU in hybrid device mode (default {PerfectNumberConstants.GpuRatio})");
        Console.WriteLine("  --gpu-prime-threads=<value>  max concurrent GPU prime checks (default 1)");
+        Console.WriteLine("  --gpu-prime-batch=<n>    batch size for GPU primality sieve (default 262144)");
        Console.WriteLine("  --gpu-scan-batch=<value>  GPU q-scan batch size (default 2_097_152)");
        Console.WriteLine("  --order-device=cpu|gpu|hybrid  Device for order computations (default gpu)");
        Console.WriteLine("  --ll-slice=<value>     Lucas–Lehmer iterations per slice (default 32)");
@@ -32,6 +32,9 @@ using ByDivisorGpuOrderTopDown = PerfectNumbers.Core.Cpu.MersenneNumberDivisorBy
 using ByDivisorCpuOrderBitContradiction = PerfectNumbers.Core.Cpu.MersenneNumberDivisorByDivisorCpuTesterWithForBitContradictionDivisorSetForCpuOrder;
 using ByDivisorHybridOrderBitContradiction = PerfectNumbers.Core.Cpu.MersenneNumberDivisorByDivisorCpuTesterWithForBitContradictionDivisorSetForHybridOrder;
 using ByDivisorGpuOrderBitContradiction = PerfectNumbers.Core.Cpu.MersenneNumberDivisorByDivisorCpuTesterWithForBitContradictionDivisorSetForGpuOrder;
+using ByDivisorCpuOrderBitTree = PerfectNumbers.Core.Cpu.MersenneNumberDivisorByDivisorCpuTesterWithForBitTreeDivisorSetForCpuOrder;
+using ByDivisorHybridOrderBitTree = PerfectNumbers.Core.Cpu.MersenneNumberDivisorByDivisorCpuTesterWithForBitTreeDivisorSetForHybridOrder;
+using ByDivisorGpuOrderBitTree = PerfectNumbers.Core.Cpu.MersenneNumberDivisorByDivisorCpuTesterWithForBitTreeDivisorSetForGpuOrder;
 using System.Globalization;

 namespace EvenPerfectBitScanner;
@@ -64,6 +67,9 @@ private static ThreadLocal<object>? MersenneTesters;
 	private static ByDivisorCpuOrderBitContradiction _byDivisorCpuOrderTesterBitContradiction;
 	private static ByDivisorHybridOrderBitContradiction _byDivisorHybridOrderTesterBitContradiction;
 	private static ByDivisorGpuOrderBitContradiction _byDivisorGpuOrderTesterBitContradiction;
+	private static ByDivisorCpuOrderBitTree _byDivisorCpuOrderTesterBitTree;
+	private static ByDivisorHybridOrderBitTree _byDivisorHybridOrderTesterBitTree;
+	private static ByDivisorGpuOrderBitTree _byDivisorGpuOrderTesterBitTree;
 	private static Dictionary<ulong, (bool DetailedCheck, bool PassedAllTests)>? _byDivisorPreviousResults;
 	private static string? _byDivisorModelPath;
 	private static bool _byDivisorModeUsesModel;
@@ -335,6 +341,12 @@ private static ThreadLocal<object>? MersenneTesters;
 						case ByDivisorKIncrementMode.TopDown:
 							_byDivisorCpuOrderTesterTopDown = new ByDivisorCpuOrderTopDown();
 							break;
+						case ByDivisorKIncrementMode.BitContradiction:
+							_byDivisorCpuOrderTesterBitContradiction = new ByDivisorCpuOrderBitContradiction();
+							break;
+						case ByDivisorKIncrementMode.BitTree:
+							_byDivisorCpuOrderTesterBitTree = new ByDivisorCpuOrderBitTree();
+							break;
 						default:
 							_byDivisorCpuOrderTester = new MersenneNumberDivisorByDivisorCpuTesterWithCpuOrder();
 							break;
@@ -359,6 +371,12 @@ private static ThreadLocal<object>? MersenneTesters;
 						case ByDivisorKIncrementMode.TopDown:
 							_byDivisorHybridOrderTesterTopDown = new ByDivisorHybridOrderTopDown();
 							break;
+						case ByDivisorKIncrementMode.BitContradiction:
+							_byDivisorHybridOrderTesterBitContradiction = new ByDivisorHybridOrderBitContradiction();
+							break;
+						case ByDivisorKIncrementMode.BitTree:
+							_byDivisorHybridOrderTesterBitTree = new ByDivisorHybridOrderBitTree();
+							break;
 						default:
 							_byDivisorHybridOrderTester = new MersenneNumberDivisorByDivisorCpuTesterWithHybridOrder();
 							break;
@@ -383,6 +401,12 @@ private static ThreadLocal<object>? MersenneTesters;
 						case ByDivisorKIncrementMode.TopDown:
 							_byDivisorGpuOrderTesterTopDown = new ByDivisorGpuOrderTopDown();
 							break;
+						case ByDivisorKIncrementMode.BitContradiction:
+							_byDivisorGpuOrderTesterBitContradiction = new ByDivisorGpuOrderBitContradiction();
+							break;
+						case ByDivisorKIncrementMode.BitTree:
+							_byDivisorGpuOrderTesterBitTree = new ByDivisorGpuOrderBitTree();
+							break;
 						default:
 							_byDivisorGpuOrderTester = new MersenneNumberDivisorByDivisorCpuTesterWithGpuOrder();
 							break;
@@ -545,6 +569,7 @@ private static ThreadLocal<object>? MersenneTesters;
 					ByDivisorKIncrementMode.Additive => "additive.kstate.faster",
 					ByDivisorKIncrementMode.TopDown => "topdown.kstate.faster",
 					ByDivisorKIncrementMode.BitContradiction => "bitcontradiction.kstate.faster",
+					ByDivisorKIncrementMode.BitTree => "bittree.kstate.faster",
 					_ => "sequential.kstate.faster",
 				};
 				string lmdbPath = Path.Combine(checksDir, Path.ChangeExtension(kRepoFile, ".lmdb"));
@@ -664,7 +689,7 @@ private static ThreadLocal<object>? MersenneTesters;
 			CalculationResultHandler.InitializeOutputBuffer();

 			// Limit GPU concurrency only for prime checks (LL/NTT & GPU order scans).
-			GpuPrimeWorkLimiter.SetLimit(gpuPrimeThreads);
+			GpuPrimeWorkLimiter.SetLimit(gpuPrimeThreads * PerfectNumberConstants.ThreadsByAccelerator);
 			// Configure batch size for GPU primality sieve
 			PrimeTester.GpuBatchSize = gpuPrimeBatch;

@@ -751,6 +776,19 @@ private static ThreadLocal<object>? MersenneTesters;
 								threadCount,
 								byDivisorPrimeRange);
 							break;
+						case ByDivisorKIncrementMode.BitTree:
+							Console.WriteLine("Using by-divisor bit-tree increment (CPU order).");
+							MersenneNumberDivisorByDivisorTester.Run(
+								byDivisorCandidates,
+								ref _byDivisorCpuOrderTesterBitTree,
+								_byDivisorPreviousResults,
+								_byDivisorStartPrime,
+								static () => _lastCompositeP = true,
+								static () => _lastCompositeP = false,
+								HandleByDivisorResult,
+								threadCount,
+								byDivisorPrimeRange);
+							break;
 						default:
 							MersenneNumberDivisorByDivisorTester.Run(
 								byDivisorCandidates,
@@ -834,6 +872,19 @@ private static ThreadLocal<object>? MersenneTesters;
 								threadCount,
 								byDivisorPrimeRange);
 							break;
+						case ByDivisorKIncrementMode.BitTree:
+							Console.WriteLine("Using by-divisor bit-tree increment (Hybrid order).");
+							MersenneNumberDivisorByDivisorTester.Run(
+								byDivisorCandidates,
+								ref _byDivisorHybridOrderTesterBitTree,
+								_byDivisorPreviousResults,
+								_byDivisorStartPrime,
+								static () => _lastCompositeP = true,
+								static () => _lastCompositeP = false,
+								HandleByDivisorResult,
+								threadCount,
+								byDivisorPrimeRange);
+							break;
 						default:
 							MersenneNumberDivisorByDivisorTester.Run(
 								byDivisorCandidates,
@@ -917,6 +968,19 @@ private static ThreadLocal<object>? MersenneTesters;
 								threadCount,
 								byDivisorPrimeRange);
 							break;
+						case ByDivisorKIncrementMode.BitTree:
+							Console.WriteLine("Using by-divisor bit-tree increment (GPU order).");
+							MersenneNumberDivisorByDivisorTester.Run(
+								byDivisorCandidates,
+								ref _byDivisorGpuOrderTesterBitTree,
+								_byDivisorPreviousResults,
+								_byDivisorStartPrime,
+								static () => _lastCompositeP = true,
+								static () => _lastCompositeP = false,
+								HandleByDivisorResult,
+								threadCount,
+								byDivisorPrimeRange);
+							break;
 						default:
 							MersenneNumberDivisorByDivisorTester.Run(
 								byDivisorCandidates,
@@ -1327,6 +1391,7 @@ private static ThreadLocal<object>? MersenneTesters;
 						ByDivisorKIncrementMode.Additive => _byDivisorCpuOrderTesterAdditive.IsPrime(p, out detailedCheck, out _),
 						ByDivisorKIncrementMode.TopDown => _byDivisorCpuOrderTesterTopDown.IsPrime(p, out detailedCheck, out _),
 						ByDivisorKIncrementMode.BitContradiction => _byDivisorCpuOrderTesterBitContradiction.IsPrime(p, out detailedCheck, out _),
+						ByDivisorKIncrementMode.BitTree => _byDivisorCpuOrderTesterBitTree.IsPrime(p, out detailedCheck, out _),
 						_ => _byDivisorCpuOrderTester.IsPrime(p, out detailedCheck, out _),
 					};
 				}
@@ -1341,6 +1406,7 @@ private static ThreadLocal<object>? MersenneTesters;
 						ByDivisorKIncrementMode.Additive => _byDivisorHybridOrderTesterAdditive.IsPrime(p, out detailedCheck, out _),
 						ByDivisorKIncrementMode.TopDown => _byDivisorHybridOrderTesterTopDown.IsPrime(p, out detailedCheck, out _),
 						ByDivisorKIncrementMode.BitContradiction => _byDivisorHybridOrderTesterBitContradiction.IsPrime(p, out detailedCheck, out _),
+						ByDivisorKIncrementMode.BitTree => _byDivisorHybridOrderTesterBitTree.IsPrime(p, out detailedCheck, out _),
 						_ => _byDivisorHybridOrderTester.IsPrime(p, out detailedCheck, out _),
 					};
 				}
@@ -1353,6 +1419,7 @@ private static ThreadLocal<object>? MersenneTesters;
 					ByDivisorKIncrementMode.Additive => _byDivisorGpuOrderTesterAdditive.IsPrime(p, out detailedCheck, out _),
 					ByDivisorKIncrementMode.TopDown => _byDivisorGpuOrderTesterTopDown.IsPrime(p, out detailedCheck, out _),
 					ByDivisorKIncrementMode.BitContradiction => _byDivisorGpuOrderTesterBitContradiction.IsPrime(p, out detailedCheck, out _),
+					ByDivisorKIncrementMode.BitTree => _byDivisorGpuOrderTesterBitTree.IsPrime(p, out detailedCheck, out _),
 					_ => _byDivisorGpuOrderTester.IsPrime(p, out detailedCheck, out _),
 				};
 			}
@@ -0,0 +1,43 @@
+using System.Numerics;
+using FluentAssertions;
+using PerfectNumbers.Core.Cpu;
+using Xunit;
+
+namespace PerfectNumbers.Core.Tests;
+
+public class BitTreeSolverTests
+{
+    [Fact]
+    public void BitTree_FindsKnownDivisor()
+    {
+        // 2^11-1 = 2047 = 23 * 89, so k=1 yields divisor 23.
+        var tester = new MersenneNumberDivisorByDivisorCpuTesterWithForBitTreeDivisorSetForCpuOrder
+        {
+            DivisorLimit = new BigInteger(1000),
+            MinK = BigInteger.One,
+        };
+
+        bool isPrime = tester.IsPrime(11, out bool exhausted, out BigInteger divisor);
+
+        isPrime.Should().BeFalse();
+        exhausted.Should().BeTrue();
+        divisor.Should().Be(23);
+    }
+
+    [Fact]
+    public void BitTree_ReportsPrimeWhenNoDivisorFound()
+    {
+        // 2^5-1 = 31 is prime.
+        var tester = new MersenneNumberDivisorByDivisorCpuTesterWithForBitTreeDivisorSetForCpuOrder
+        {
+            DivisorLimit = new BigInteger(10_000),
+            MinK = BigInteger.One,
+        };
+
+        bool isPrime = tester.IsPrime(5, out bool exhausted, out BigInteger divisor);
+
+        isPrime.Should().BeTrue();
+        exhausted.Should().BeTrue();
+        divisor.Should().Be(BigInteger.Zero);
+    }
+}
@@ -27,6 +27,10 @@ using TesterGpu = PerfectNumbers.Core.Cpu.MersenneNumberDivisorByDivisorCpuTeste
 using TesterCpu = PerfectNumbers.Core.Cpu.MersenneNumberDivisorByDivisorCpuTesterWithForBitContradictionDivisorSetForCpuOrder;
 using TesterHybrid = PerfectNumbers.Core.Cpu.MersenneNumberDivisorByDivisorCpuTesterWithForBitContradictionDivisorSetForHybridOrder;
 using TesterGpu = PerfectNumbers.Core.Cpu.MersenneNumberDivisorByDivisorCpuTesterWithForBitContradictionDivisorSetForGpuOrder;
+#elif DivisorSet_BitTree
+using TesterCpu = PerfectNumbers.Core.Cpu.MersenneNumberDivisorByDivisorCpuTesterWithForBitTreeDivisorSetForCpuOrder;
+using TesterHybrid = PerfectNumbers.Core.Cpu.MersenneNumberDivisorByDivisorCpuTesterWithForBitTreeDivisorSetForHybridOrder;
+using TesterGpu = PerfectNumbers.Core.Cpu.MersenneNumberDivisorByDivisorCpuTesterWithForBitTreeDivisorSetForGpuOrder;
 #else
 using TesterCpu = PerfectNumbers.Core.Cpu.MersenneNumberDivisorByDivisorCpuTesterWithForSequentialDivisorSetForCpuOrder;
 using TesterHybrid = PerfectNumbers.Core.Cpu.MersenneNumberDivisorByDivisorCpuTesterWithForSequentialDivisorSetForHybridOrder;
@@ -91,6 +95,8 @@ internal struct MersenneNumberDivisorByDivisorPrimeScanSessionWithTemplate
 			extension = ".topdown";
 #elif DivisorSet_BitContradiction
 			extension = ".bitcontradiction";
+#elif DivisorSet_BitTree
+			extension = ".bittree";
 #endif
 			string stateFile = prime.ToString(CultureInfo.InvariantCulture) + extension;

@@ -149,7 +149,7 @@ public sealed class PrimeOrderCalculatorAccelerator

 	private sealed class CalculatorKernels(Accelerator accelerator)
 	{
-		public readonly Action<AcceleratorStream, Index1D, ulong, CalculateOrderKernelConfig, ulong, ulong, ulong, ulong, ulong, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, int, OrderKernelBuffers> CalculateOrderKernelLauncher = KernelUtil.GetKernel(accelerator.LoadAutoGroupedStreamKernel<Index1D, ulong, CalculateOrderKernelConfig, ulong, ulong, ulong, ulong, ulong, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, int, OrderKernelBuffers>(PrimeOrderKernels.CalculateOrderKernel)).CreateLauncherDelegate<Action<AcceleratorStream, Index1D, ulong, CalculateOrderKernelConfig, ulong, ulong, ulong, ulong, ulong, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, int, OrderKernelBuffers>>();
+		// public readonly Action<AcceleratorStream, Index1D, ulong, CalculateOrderKernelConfig, ulong, ulong, ulong, ulong, ulong, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, int, OrderKernelBuffers> CalculateOrderKernelLauncher = KernelUtil.GetKernel(accelerator.LoadAutoGroupedStreamKernel<Index1D, ulong, CalculateOrderKernelConfig, ulong, ulong, ulong, ulong, ulong, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, int, OrderKernelBuffers>(PrimeOrderKernels.CalculateOrderKernel)).CreateLauncherDelegate<Action<AcceleratorStream, Index1D, ulong, CalculateOrderKernelConfig, ulong, ulong, ulong, ulong, ulong, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, int, OrderKernelBuffers>>();

 		public readonly Kernel CheckFactorsKernel = KernelUtil.GetKernel(accelerator.LoadStreamKernel<int, ulong, ArrayView1D<KeyValuePair<ulong, int>, Stride1D.Dense>, ulong, ulong, ulong, ulong, ulong, ArrayView1D<ulong, Stride1D.Dense>>(PrimeOrderKernels.CheckFactorsKernel));

@@ -177,7 +177,7 @@ public sealed class PrimeOrderCalculatorAccelerator
 		public readonly Action<AcceleratorStream, Index1D, ulong, ArrayView<ulong>, int, ulong, ArrayView<ulong>> SpecialMaxKernelLauncher = KernelUtil.GetKernel(accelerator.LoadAutoGroupedStreamKernel<Index1D, ulong, ArrayView<ulong>, int, ulong, ArrayView<ulong>>(PrimeOrderGpuHeuristics.EvaluateSpecialMaxCandidatesKernel)).CreateLauncherDelegate<Action<AcceleratorStream, Index1D, ulong, ArrayView<ulong>, int, ulong, ArrayView<ulong>>>();
 	}

-	public readonly Action<AcceleratorStream, Index1D, ulong, CalculateOrderKernelConfig, ulong, ulong, ulong, ulong, ulong, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, int, OrderKernelBuffers> CalculateOrderKernelLauncher;
+	// public readonly Action<AcceleratorStream, Index1D, ulong, CalculateOrderKernelConfig, ulong, ulong, ulong, ulong, ulong, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, int, OrderKernelBuffers> CalculateOrderKernelLauncher;
 	public readonly Kernel CheckFactorsKernel;
 	public readonly Action<AcceleratorStream, Index1D, ArrayView1D<ulong, Stride1D.Dense>, ulong, ulong, ulong, ulong, ulong, ArrayView1D<ulong, Stride1D.Dense>> ConvertToStandardKernelLauncher;
 	public readonly Action<AcceleratorStream, Index1D, ArrayView<int>, ulong, ulong, ArrayView<ulong>, ArrayView<ulong>> HeuristicCombinedTrialDivisionKernelLauncher;
@@ -421,9 +421,9 @@ public sealed class PrimeOrderCalculatorAccelerator
 		DevicePrimesPow2LastThree = lastDigitSharedTables.DevicePrimesPow2LastThree.View;
 		DevicePrimesPow2LastNine = lastDigitSharedTables.DevicePrimesPow2LastNine.View;

-		DivisorTables = HeuristicCombinedGpuTables
-			.GetStaticTables(acceleratorIndex)
-			.CreateViews();
+		// DivisorTables = HeuristicCombinedGpuTables
+		// 	.GetStaticTables(acceleratorIndex)
+		// 	.CreateViews();

 		var smallPrimeFactorTables = SmallPrimeFactorGpuTables.GetStaticTables(acceleratorIndex);

@@ -436,7 +436,7 @@ public sealed class PrimeOrderCalculatorAccelerator

 		var kernels = _kernels[acceleratorIndex];

-		CalculateOrderKernelLauncher = kernels.CalculateOrderKernelLauncher;
+		// CalculateOrderKernelLauncher = kernels.CalculateOrderKernelLauncher;
 		CheckFactorsKernel = kernels.CheckFactorsKernel;
 		ConvertToStandardKernelLauncher = kernels.ConvertToStandardKernelLauncher;
 		HeuristicCombinedTrialDivisionKernelLauncher = kernels.HeuristicCombinedTrialDivisionKernelLauncher;
@@ -272,27 +272,27 @@ internal static partial class PrimeOrderGpuHeuristics
 		var kernelConfig = new CalculateOrderKernelConfig(previousValue, hasPrevious, limit, config.MaxPowChecks, config.StrictMode);
 		ref var buffers = ref gpu.CalculateOrderKernelBuffers;

-		var kernelLauncher = gpu.CalculateOrderKernelLauncher;
+		// var kernelLauncher = gpu.CalculateOrderKernelLauncher;

 		AcceleratorStream stream = AcceleratorStreamPool.Rent(acceleratorIndex);
 		// TODO: Remove the cleaning after the order kernel is modified to always set the result.
 		resultBufferView.MemSetToZero(stream);
 		statusBufferView.MemSetToZero(stream);

-		kernelLauncher(
-			stream,
-			1,
-			prime,
-			kernelConfig,
-			divisorData.Modulus,
-			divisorData.NPrime,
-			divisorData.MontgomeryOne,
-			divisorData.MontgomeryTwo,
-			divisorData.MontgomeryTwoSquared,
-			smallPrimesView,
-			smallSquaresView,
-			(int)smallPrimesView.Length,
-			buffers);
+		// kernelLauncher(
+		// 	stream,
+		// 	1,
+		// 	prime,
+		// 	kernelConfig,
+		// 	divisorData.Modulus,
+		// 	divisorData.NPrime,
+		// 	divisorData.MontgomeryOne,
+		// 	divisorData.MontgomeryTwo,
+		// 	divisorData.MontgomeryTwoSquared,
+		// 	smallPrimesView,
+		// 	smallSquaresView,
+		// 	(int)smallPrimesView.Length,
+		// 	buffers);


 		byte status = 0;
@@ -21,6 +21,9 @@ using MersenneNumberDivisorByDivisorCpuTesterWithAdditiveGpuOrder = PerfectNumbe
 using MersenneNumberDivisorByDivisorCpuTesterWithBitContradictionCpuOrder = PerfectNumbers.Core.Cpu.MersenneNumberDivisorByDivisorCpuTesterWithForBitContradictionDivisorSetForCpuOrder;
 using MersenneNumberDivisorByDivisorCpuTesterWithBitContradictionHybridOrder = PerfectNumbers.Core.Cpu.MersenneNumberDivisorByDivisorCpuTesterWithForBitContradictionDivisorSetForHybridOrder;
 using MersenneNumberDivisorByDivisorCpuTesterWithBitContradictionGpuOrder = PerfectNumbers.Core.Cpu.MersenneNumberDivisorByDivisorCpuTesterWithForBitContradictionDivisorSetForGpuOrder;
+using MersenneNumberDivisorByDivisorCpuTesterWithBitTreeCpuOrder = PerfectNumbers.Core.Cpu.MersenneNumberDivisorByDivisorCpuTesterWithForBitTreeDivisorSetForCpuOrder;
+using MersenneNumberDivisorByDivisorCpuTesterWithBitTreeHybridOrder = PerfectNumbers.Core.Cpu.MersenneNumberDivisorByDivisorCpuTesterWithForBitTreeDivisorSetForHybridOrder;
+using MersenneNumberDivisorByDivisorCpuTesterWithBitTreeGpuOrder = PerfectNumbers.Core.Cpu.MersenneNumberDivisorByDivisorCpuTesterWithForBitTreeDivisorSetForGpuOrder;

 namespace PerfectNumbers.Core;

@@ -88,6 +91,37 @@ public static class MersenneNumberDivisorByDivisorTester
 				Action<ulong, bool, bool, bool, BigInteger> pr) =>
 				RunBitContradictionCpuOrderFiltered(filteredPrimes, prototype, workerCount, partitionSize, mc, cc, pr));

+	[MethodImpl(MethodImplOptions.AggressiveInlining | MethodImplOptions.AggressiveOptimization)]
+	public static void Run(
+		List<ulong> candidates,
+		ref MersenneNumberDivisorByDivisorCpuTesterWithBitTreeCpuOrder tester,
+		Dictionary<ulong, (bool DetailedCheck, bool PassedAllTests)>? previousResults,
+		ulong startPrime,
+		Action markComposite,
+		Action clearComposite,
+		Action<ulong, bool, bool, bool, BigInteger> printResult,
+		int threadCount,
+		int primesPerTask)
+		=> RunStructTester(
+			candidates,
+			ref tester,
+			previousResults,
+			startPrime,
+			markComposite,
+			clearComposite,
+			printResult,
+			threadCount,
+			primesPerTask,
+			static (
+				List<ulong> filteredPrimes,
+				in MersenneNumberDivisorByDivisorCpuTesterWithBitTreeCpuOrder prototype,
+				int workerCount,
+				int partitionSize,
+				Action mc,
+				Action cc,
+				Action<ulong, bool, bool, bool, BigInteger> pr) =>
+				RunBitTreeCpuOrderFiltered(filteredPrimes, prototype, workerCount, partitionSize, mc, cc, pr));
+
 	[MethodImpl(MethodImplOptions.AggressiveInlining | MethodImplOptions.AggressiveOptimization)]
 	public static void Run(
 		List<ulong> candidates,
@@ -243,6 +277,37 @@ public static class MersenneNumberDivisorByDivisorTester
 				Action<ulong, bool, bool, bool, BigInteger> pr) =>
 				RunBitContradictionHybridOrderFiltered(filteredPrimes, prototype, workerCount, partitionSize, mc, cc, pr));

+	[MethodImpl(MethodImplOptions.AggressiveInlining | MethodImplOptions.AggressiveOptimization)]
+	public static void Run(
+		List<ulong> candidates,
+		ref MersenneNumberDivisorByDivisorCpuTesterWithBitTreeHybridOrder tester,
+		Dictionary<ulong, (bool DetailedCheck, bool PassedAllTests)>? previousResults,
+		ulong startPrime,
+		Action markComposite,
+		Action clearComposite,
+		Action<ulong, bool, bool, bool, BigInteger> printResult,
+		int threadCount,
+		int primesPerTask)
+		=> RunStructTester(
+			candidates,
+			ref tester,
+			previousResults,
+			startPrime,
+			markComposite,
+			clearComposite,
+			printResult,
+			threadCount,
+			primesPerTask,
+			static (
+				List<ulong> filteredPrimes,
+				in MersenneNumberDivisorByDivisorCpuTesterWithBitTreeHybridOrder prototype,
+				int workerCount,
+				int partitionSize,
+				Action mc,
+				Action cc,
+				Action<ulong, bool, bool, bool, BigInteger> pr) =>
+				RunBitTreeHybridOrderFiltered(filteredPrimes, prototype, workerCount, partitionSize, mc, cc, pr));
+
 	[MethodImpl(MethodImplOptions.AggressiveInlining | MethodImplOptions.AggressiveOptimization)]
 	public static void Run(
 		List<ulong> candidates,
@@ -398,6 +463,37 @@ public static class MersenneNumberDivisorByDivisorTester
 				Action<ulong, bool, bool, bool, BigInteger> pr) =>
 				RunBitContradictionGpuOrderFiltered(filteredPrimes, prototype, workerCount, partitionSize, mc, cc, pr));

+	[MethodImpl(MethodImplOptions.AggressiveInlining | MethodImplOptions.AggressiveOptimization)]
+	public static void Run(
+		List<ulong> candidates,
+		ref MersenneNumberDivisorByDivisorCpuTesterWithBitTreeGpuOrder tester,
+		Dictionary<ulong, (bool DetailedCheck, bool PassedAllTests)>? previousResults,
+		ulong startPrime,
+		Action markComposite,
+		Action clearComposite,
+		Action<ulong, bool, bool, bool, BigInteger> printResult,
+		int threadCount,
+		int primesPerTask)
+		=> RunStructTester(
+			candidates,
+			ref tester,
+			previousResults,
+			startPrime,
+			markComposite,
+			clearComposite,
+			printResult,
+			threadCount,
+			primesPerTask,
+			static (
+				List<ulong> filteredPrimes,
+				in MersenneNumberDivisorByDivisorCpuTesterWithBitTreeGpuOrder prototype,
+				int workerCount,
+				int partitionSize,
+				Action mc,
+				Action cc,
+				Action<ulong, bool, bool, bool, BigInteger> pr) =>
+				RunBitTreeGpuOrderFiltered(filteredPrimes, prototype, workerCount, partitionSize, mc, cc, pr));
+
 	[MethodImpl(MethodImplOptions.AggressiveInlining | MethodImplOptions.AggressiveOptimization)]
 	public static void Run(
 		List<ulong> candidates,
@@ -2575,6 +2671,150 @@ public static class MersenneNumberDivisorByDivisorTester
 			}));
 	}

+	private static void RunBitTreeCpuOrderFiltered(
+		List<ulong> filteredPrimes,
+		in MersenneNumberDivisorByDivisorCpuTesterWithBitTreeCpuOrder prototype,
+		int workerCount,
+		int partitionSize,
+		Action markComposite,
+		Action clearComposite,
+		Action<ulong, bool, bool, bool, BigInteger> printResult)
+	{
+		MersenneNumberDivisorByDivisorCpuTesterWithBitTreeCpuOrder prototypeCopy = prototype;
+		int totalCount = filteredPrimes.Count;
+		if (partitionSize > totalCount)
+		{
+			partitionSize = totalCount;
+		}
+
+		if (workerCount <= 1 || totalCount <= partitionSize)
+		{
+			var session = new Cpu.MersenneNumberDivisorByDivisorPrimeScanSessionWithForBitTreeDivisorSetForCpuOrder(prototypeCopy, markComposite, clearComposite, printResult);
+			for (int i = 0; i < totalCount; i++)
+			{
+				session.ProcessPrime(filteredPrimes[i]);
+			}
+
+			return;
+		}
+
+		var options = new ParallelOptions
+		{
+			MaxDegreeOfParallelism = workerCount,
+		};
+
+		Parallel.For(
+			0,
+			(int)Math.Ceiling((double)totalCount / partitionSize),
+			options,
+			new Action<int>(partitionIndex =>
+			{
+				int startIndex = partitionIndex * partitionSize;
+				int count = Math.Min(partitionSize, totalCount - startIndex);
+				var session = new Cpu.MersenneNumberDivisorByDivisorPrimeScanSessionWithForBitTreeDivisorSetForCpuOrder(prototypeCopy, markComposite, clearComposite, printResult);
+				for (int i = 0; i < count; i++)
+				{
+					session.ProcessPrime(filteredPrimes[startIndex + i]);
+				}
+			}));
+	}
+
+	private static void RunBitTreeHybridOrderFiltered(
+		List<ulong> filteredPrimes,
+		in MersenneNumberDivisorByDivisorCpuTesterWithBitTreeHybridOrder prototype,
+		int workerCount,
+		int partitionSize,
+		Action markComposite,
+		Action clearComposite,
+		Action<ulong, bool, bool, bool, BigInteger> printResult)
+	{
+		MersenneNumberDivisorByDivisorCpuTesterWithBitTreeHybridOrder prototypeCopy = prototype;
+		int totalCount = filteredPrimes.Count;
+		if (partitionSize > totalCount)
+		{
+			partitionSize = totalCount;
+		}
+
+		if (workerCount <= 1 || totalCount <= partitionSize)
+		{
+			var session = new Cpu.MersenneNumberDivisorByDivisorPrimeScanSessionWithForBitTreeDivisorSetForHybridOrder(prototypeCopy, markComposite, clearComposite, printResult);
+			for (int i = 0; i < totalCount; i++)
+			{
+				session.ProcessPrime(filteredPrimes[i]);
+			}
+
+			return;
+		}
+
+		var options = new ParallelOptions
+		{
+			MaxDegreeOfParallelism = workerCount,
+		};
+
+		Parallel.For(
+			0,
+			(int)Math.Ceiling((double)totalCount / partitionSize),
+			options,
+			new Action<int>(partitionIndex =>
+			{
+				int startIndex = partitionIndex * partitionSize;
+				int count = Math.Min(partitionSize, totalCount - startIndex);
+				var session = new Cpu.MersenneNumberDivisorByDivisorPrimeScanSessionWithForBitTreeDivisorSetForHybridOrder(prototypeCopy, markComposite, clearComposite, printResult);
+				for (int i = 0; i < count; i++)
+				{
+					session.ProcessPrime(filteredPrimes[startIndex + i]);
+				}
+			}));
+	}
+
+	private static void RunBitTreeGpuOrderFiltered(
+		List<ulong> filteredPrimes,
+		in MersenneNumberDivisorByDivisorCpuTesterWithBitTreeGpuOrder prototype,
+		int workerCount,
+		int partitionSize,
+		Action markComposite,
+		Action clearComposite,
+		Action<ulong, bool, bool, bool, BigInteger> printResult)
+	{
+		MersenneNumberDivisorByDivisorCpuTesterWithBitTreeGpuOrder prototypeCopy = prototype;
+		int totalCount = filteredPrimes.Count;
+		if (partitionSize > totalCount)
+		{
+			partitionSize = totalCount;
+		}
+
+		if (workerCount <= 1 || totalCount <= partitionSize)
+		{
+			var session = new Cpu.MersenneNumberDivisorByDivisorPrimeScanSessionWithForBitTreeDivisorSetForGpuOrder(prototypeCopy, markComposite, clearComposite, printResult);
+			for (int i = 0; i < totalCount; i++)
+			{
+				session.ProcessPrime(filteredPrimes[i]);
+			}
+
+			return;
+		}
+
+		var options = new ParallelOptions
+		{
+			MaxDegreeOfParallelism = workerCount,
+		};
+
+		Parallel.For(
+			0,
+			(int)Math.Ceiling((double)totalCount / partitionSize),
+			options,
+			new Action<int>(partitionIndex =>
+			{
+				int startIndex = partitionIndex * partitionSize;
+				int count = Math.Min(partitionSize, totalCount - startIndex);
+				var session = new Cpu.MersenneNumberDivisorByDivisorPrimeScanSessionWithForBitTreeDivisorSetForGpuOrder(prototypeCopy, markComposite, clearComposite, printResult);
+				for (int i = 0; i < count; i++)
+				{
+					session.ProcessPrime(filteredPrimes[startIndex + i]);
+				}
+			}));
+	}
+
 	private static void RunBitContradictionHybridOrderFiltered(
 		List<ulong> filteredPrimes,
 		in MersenneNumberDivisorByDivisorCpuTesterWithBitContradictionHybridOrder prototype,
@@ -0,0 +1,129 @@
+using System;
+using System.Collections.Generic;
+using System.Globalization;
+using System.IO;
+using System.Numerics;
+using System.Text;
+using FASTER.core;
+
+namespace PerfectNumbers.Core.Persistence;
+
+/// <summary>
+/// Stores BitTree frontier batches using Microsoft.FASTER.Core.
+/// Each batch is a byte[] payload containing a sequence of serialized states.
+/// Maintains a simple head/tail cursor stored inside the KV store.
+/// </summary>
+public sealed class BitTreeFrontierRepository : IDisposable
+{
+    private readonly FasterKV<string, byte[]> _store;
+    private readonly IDevice _logDevice;
+    private readonly IDevice _objectLogDevice;
+    private long _headIndex;
+    private long _tailIndex;
+    private const string HeadKey = "__head__";
+    private const string TailKey = "__tail__";
+
+    public string DatabasePath { get; }
+
+    public static BitTreeFrontierRepository Open(string directory, string repositoryFileName = "bittree.frontier.faster")
+    {
+        Directory.CreateDirectory(directory);
+        string dbPath = Path.Combine(directory, repositoryFileName);
+        return new BitTreeFrontierRepository(dbPath);
+    }
+
+    private BitTreeFrontierRepository(string path)
+    {
+        DatabasePath = path;
+        _logDevice = Devices.CreateLogDevice(path + ".log");
+        _objectLogDevice = Devices.CreateLogDevice(path + ".obj.log");
+        _store = new FasterKV<string, byte[]>(
+            1L << 20,
+            new LogSettings
+            {
+                LogDevice = _logDevice,
+                ObjectLogDevice = _objectLogDevice,
+            });
+
+        LoadCursors();
+    }
+
+    public void AppendBatch(ReadOnlySpan<byte> batch)
+    {
+        using var session = _store.For(new SimpleFunctions<string, byte[]>()).NewSession<SimpleFunctions<string, byte[]>>();
+        string key = BuildChunkKey(_tailIndex);
+        session.Upsert(key, batch.ToArray());
+        _tailIndex++;
+        PersistCursors(session);
+        session.CompletePending(true);
+		_store.Log.FlushAndEvict(true);
+    }
+
+    public bool TryDequeueBatch(int maxCount, out List<byte[]> batches)
+    {
+        batches = new List<byte[]>();
+        using var session = _store.For(new SimpleFunctions<string, byte[]>()).NewSession<SimpleFunctions<string, byte[]>>();
+
+        while (_headIndex < _tailIndex && batches.Count < maxCount)
+        {
+            string key = BuildChunkKey(_headIndex);
+            if (session.Read(key, out byte[]? stored).Found && stored is not null)
+            {
+                batches.Add(stored);
+                session.Delete(key);
+            }
+
+            _headIndex++;
+        }
+
+
+        if (batches.Count > 0)
+        {
+            PersistCursors(session);
+            session.CompletePending(true);
+			_store.Log.FlushAndEvict(true);
+            return true;
+        }
+
+        session.CompletePending(true);
+		_store.Log.FlushAndEvict(true);
+        return false;
+    }
+
+    public void Dispose()
+    {
+        _store.Dispose();
+        _logDevice.Dispose();
+        _objectLogDevice.Dispose();
+    }
+
+    private void LoadCursors()
+    {
+        using var session = _store.For(new SimpleFunctions<string, byte[]>()).NewSession<SimpleFunctions<string, byte[]>>();
+        _headIndex = ReadCursor(session, HeadKey);
+        _tailIndex = ReadCursor(session, TailKey);
+        session.CompletePending(true);
+    }
+
+    private static long ReadCursor(ClientSession<string, byte[], byte[], byte[], Empty, SimpleFunctions<string, byte[]>> session, string key)
+    {
+        if (session.Read(key, out byte[]? stored).Found && stored is not null)
+        {
+            string text = Encoding.UTF8.GetString(stored);
+            if (long.TryParse(text, NumberStyles.Integer, CultureInfo.InvariantCulture, out long value))
+            {
+                return value;
+            }
+        }
+
+        return 0;
+    }
+
+    private void PersistCursors(ClientSession<string, byte[], byte[], byte[], Empty, SimpleFunctions<string, byte[]>> session)
+    {
+        session.Upsert(HeadKey, Encoding.UTF8.GetBytes(_headIndex.ToString(CultureInfo.InvariantCulture)));
+        session.Upsert(TailKey, Encoding.UTF8.GetBytes(_tailIndex.ToString(CultureInfo.InvariantCulture)));
+    }
+
+    private static string BuildChunkKey(long index) => $"chunk-{index}";
+}
@@ -0,0 +1,52 @@
+# Bit-Tree Divisor Scan
+
+Plan and notes for the `DivisorSet.BitTree` / `--bydivisor-k-increment=bittree` mode.
+
+## Goal
+- Divisor search via joint reconstruction of unknown bits in both the divisor \(q\) (only LSB=1 is known initially) and the cofactor \(a\), matching column-wise carries in \(q \cdot a = 2^p - 1\).
+- Explore the full bit tree (all admissible \(q,a\) combinations) with early pruning from parity reachability, carry budgets, and length limits; no recursion (explicit stack/queue).
+- Report primes \(p\) whose entire search space exhausts with no successful branch first.
+
+## Operational assumptions
+- \(p\) is odd; we target prime-like divisors with \(q \equiv 1 \pmod{2p}\) but start from bit patterns (q LSB = 1) and let the tree explore.
+- Work in joint (a,q) space, not tied to a single \(k\). All admissible \(q\) consistent with \(2pk+1\) should be reachable through the bit tree.
+- Iterative DFS/BFS (explicit stack/queue), no recursion. Process columns streaming (no full 138M+ bit buffers).
+
+## Implementation steps
+- [x] Add `BitTree` mode to CLI (`--bydivisor-k-increment`) and state repositories/files; ensure distinct state extensions.
+- [x] Wire mode through CPU/Hybrid/GPU paths with `.bittree` state files.
+- [x] Implement core bit-tree solver (frontier/BFS variant):
+  - Joint state: column index, carry, partial `q` (LSB=1), partial `a`, bounds on `a` length (≤ p - bitlen(q)+1), checks against `allowedMax`.
+  - Branching: choose next `q` bit (0/1), derive `a` bit from parity+carry; prune on length overflow, parity unreachable, carry cannot collapse, or `q` beyond allowedMax.
+  - Termination: success iff at/beyond column `p` with carry=0 and no further contributions from `(a,q)`; otherwise contradiction.
+- [x] Integrate BitTree solver into `CheckDivisorsBitTree`; legacy per-`k` loop removed for this mode.
+- [ ] Logging: emit Decided/Divides and contradiction reasons; throw if solver cannot decide.
+- [ ] Tests:
+  - Unit tests on small \(p\) for success/failure paths.
+  - Integration test selecting `DivisorSet.BitTree` to ensure composites are detected and primes pass.
+- [ ] Docs/help: mention the new mode in help text and README references.
+
+## Variants considered
+
+### Variant A (implemented baseline)
+- Explicit stack-based DFS over `(a,q)` bits with reachability/pruning on parity and carry budgets.
+- Processes columns sequentially without recursion; suited to small/medium `p` in current form.
+
+### Variant B (frontier/windowed exploration — in progress)
+- Implemented: frontier/BFS over branches with sliding window (suffix) storage for `(a,q)` and per-branch carry, bounded by `allowedMax` and `p`; window slides forward to cap memory and drop obsolete history.
+- Planned enhancement: tune window size/adaptive policy and add optional frontier compression (baseline/delta) for `p ≥ 138M` and many threads.
+
+### Variant C (frontier with shared baseline/deltas — planned)
+- For each processed column maintain a shared baseline state (e.g., minimal carry/bit sums across the frontier).
+- Each branch stores only its delta relative to the baseline (carry delta and the minimal differing `(a,q)` suffix), shrinking per-branch memory.
+- When compressing the frontier after a window, pick the baseline that minimises aggregate delta size (e.g., minimal sum of carries/bit-counts across surviving branches).
+- Objective: reduce memory further for highly parallel runs (many threads) while preserving full branch coverage.
+
+## Current status
+- Variant A solver is integrated and invoked from `DivisorSet.BitTree` (branch scan path); legacy per-`k` loop is disabled for this mode.
+- Variant B is not implemented yet; design notes above.
+
+## Next steps
+- Implement frontier/windowed solver (Variant B) and add mode toggle if needed.
+- Prototype baseline/delta compression for the frontier (Variant C) and measure memory savings under parallel load.
+- Expand tests to cover both variants on small `p` and ensure integration keeps reporting decisions for all candidates.