Console logging extended

EvenPerfectBitScanner/Program.cs

@@ -473,7 +473,7 @@ internal static class Program
 		if (!useDivisor && !useByDivisor)
 		{
 			MersenneTesters = new ThreadLocal<MersenneNumberTester>(() =>
-			{
+{
 	var tester = new MersenneNumberTester(
 						useIncremental: !useLucas,
 						useOrderCache: false,
@@ -491,7 +491,7 @@ internal static class Program
 	}
 
 	return tester;
-			}, trackAllValues: true);
+}, trackAllValues: true);
 		}
 		else if (useDivisor)
 		{
@@ -616,7 +616,7 @@ internal static class Program
 			{
 				filter.AddRange(localFilter[..count]);
 			}
-                        filterList = [..filter];
+			filterList = [.. filter];
 		}
 
 		if (useByDivisor)
@@ -642,6 +642,8 @@ internal static class Program
 		// Configure batch size for GPU primality sieve
 		PrimeTester.GpuBatchSize = gpuPrimeBatch;
 
+		Console.WriteLine("Starting scan...");
+
 		if (_useByDivisorMode)
 		{
 			RunByDivisorMode(filterList, divisorCyclesSearchLimit);
@@ -662,7 +664,6 @@ internal static class Program
 			}
 		}
 
-		Console.WriteLine("Starting scan...");
 		_state = ((long)currentP << 3) | (long)remainder;
 		Task[] tasks = new Task[threadCount];
 
@@ -740,7 +741,6 @@ internal static class Program
 			return;
 		}
 
-                primes.Sort();
 		List<ByDivisorPrimeState> states = new(primes.Count);
 
 		_byDivisorPrecheckOnly = true;

PerfectNumbers.Core/Gpu/MersenneNumberDivisorByDivisorGpuTester.cs

@@ -17,9 +17,12 @@ public sealed class MersenneNumberDivisorByDivisorGpuTester
 	private ulong _lastStatusDivisor;
 
 	private readonly ConcurrentDictionary<Accelerator, Action<Index1D, ulong, ArrayView<ulong>, ArrayView<byte>>> _kernelCache = new();
+	private readonly ConcurrentDictionary<Accelerator, Action<Index1D, ulong, ArrayView<ulong>, ArrayView<byte>>> _kernelByPrimeCache = new();
 
 	private Action<Index1D, ulong, ArrayView<ulong>, ArrayView<byte>> GetKernel(Accelerator accelerator) =>
 					_kernelCache.GetOrAdd(accelerator, acc => acc.LoadAutoGroupedStreamKernel<Index1D, ulong, ArrayView<ulong>, ArrayView<byte>>(CheckKernel));
+	private Action<Index1D, ulong, ArrayView<ulong>, ArrayView<byte>> GetKernelByPrime(Accelerator accelerator) =>
+					_kernelByPrimeCache.GetOrAdd(accelerator, acc => acc.LoadAutoGroupedStreamKernel<Index1D, ulong, ArrayView<ulong>, ArrayView<byte>>(CheckKernelByPrime));
 
 	public int GpuBatchSize
 	{
@@ -110,6 +113,7 @@ public sealed class MersenneNumberDivisorByDivisorGpuTester
 					divisors[batchSize++] = current;
 
 					ulong next = current + 2UL;
+					ReportStatus(lastProcessed);
 					if (next <= current)
 					{
 						reachedEndInBatch = true;
@@ -153,6 +157,7 @@ public sealed class MersenneNumberDivisorByDivisorGpuTester
 					}
 				}
 
+				ReportStatus(lastProcessed);
 				if (composite)
 				{
 					break;
@@ -232,6 +237,79 @@ public sealed class MersenneNumberDivisorByDivisorGpuTester
 		hits[index] = pow == 1UL ? (byte)1 : (byte)0;
 	}
 
+	public sealed class DivisorScanSession : IDisposable
+	{
+		private readonly MersenneNumberDivisorByDivisorGpuTester _owner;
+		private readonly GpuContextPool.GpuContextLease _lease;
+		private readonly Accelerator _accelerator;
+		private readonly Action<Index1D, ulong, ArrayView<ulong>, ArrayView<byte>> _kernel;
+		private readonly MemoryBuffer1D<ulong, Stride1D.Dense> _primesBuffer;
+		private readonly MemoryBuffer1D<byte, Stride1D.Dense> _hitsBuffer;
+		private readonly ulong[] _primesHost;
+		private readonly byte[] _hitsHost;
+		private bool _disposed;
+
+		internal DivisorScanSession(MersenneNumberDivisorByDivisorGpuTester owner)
+		{
+			_owner = owner;
+			_lease = GpuContextPool.RentPreferred(preferCpu: false);
+			_accelerator = _lease.Accelerator;
+			_kernel = owner.GetKernelByPrime(_accelerator);
+			_primesBuffer = _accelerator.Allocate1D<ulong>(owner._gpuBatchSize);
+			_hitsBuffer = _accelerator.Allocate1D<byte>(owner._gpuBatchSize);
+			_primesHost = ArrayPool<ulong>.Shared.Rent(owner._gpuBatchSize);
+			_hitsHost = ArrayPool<byte>.Shared.Rent(owner._gpuBatchSize);
+		}
+
+		public void CheckDivisor(ulong divisor, ReadOnlySpan<ulong> primes, Span<byte> hits)
+		{
+			if (_disposed)
+			{
+				throw new ObjectDisposedException(nameof(DivisorScanSession));
+			}
+
+			if (primes.Length == 0)
+			{
+				return;
+			}
+
+			int offset = 0;
+			while (offset < primes.Length)
+			{
+				int batchSize = Math.Min(_owner._gpuBatchSize, primes.Length - offset);
+				Span<ulong> primeSpan = _primesHost.AsSpan(0, batchSize);
+				primes.Slice(offset, batchSize).CopyTo(primeSpan);
+
+				var primesView = _primesBuffer.View.SubView(0, batchSize);
+				var hitsView = _hitsBuffer.View.SubView(0, batchSize);
+
+				primesView.CopyFromCPU(ref MemoryMarshal.GetReference(primeSpan), batchSize);
+				_kernel(batchSize, divisor, primesView, hitsView);
+				_accelerator.Synchronize();
+
+				Span<byte> hitSpan = _hitsHost.AsSpan(0, batchSize);
+				hitsView.CopyToCPU(ref MemoryMarshal.GetReference(hitSpan), batchSize);
+				hitSpan.CopyTo(hits.Slice(offset, batchSize));
+				offset += batchSize;
+			}
+		}
+
+		public void Dispose()
+		{
+			if (_disposed)
+			{
+				return;
+			}
+
+			_disposed = true;
+			ArrayPool<ulong>.Shared.Return(_primesHost, clearArray: true);
+			ArrayPool<byte>.Shared.Return(_hitsHost, clearArray: true);
+			_primesBuffer.Dispose();
+			_hitsBuffer.Dispose();
+			_lease.Dispose();
+		}
+	}
+
 	private static ulong Pow2Mod(ulong exponent, ulong modulus)
 	{
 		if (modulus <= 1UL)
@@ -293,5 +371,60 @@ public sealed class MersenneNumberDivisorByDivisorGpuTester
 
 		return result;
 	}
+
+	private static void CheckKernelByPrime(Index1D index, ulong divisor, ArrayView<ulong> primes, ArrayView<byte> hits)
+	{
+		ulong prime = primes[index];
+		if (divisor <= 1UL || (divisor & 1UL) == 0UL)
+		{
+			hits[index] = 0;
+			return;
+		}
+
+		ulong pow = Pow2Mod(prime, divisor);
+		hits[index] = pow == 1UL ? (byte)1 : (byte)0;
+	}
+
+	public ulong DivisorLimit
+	{
+		get
+		{
+			lock (_sync)
+			{
+				if (!_isConfigured)
+				{
+					throw new InvalidOperationException("ConfigureFromMaxPrime must be called before using the tester.");
+				}
+
+				return _divisorLimit;
+			}
+		}
+	}
+
+	public ulong GetAllowedMaxDivisor(ulong prime)
+	{
+		lock (_sync)
+		{
+			if (!_isConfigured)
+			{
+				throw new InvalidOperationException("ConfigureFromMaxPrime must be called before using the tester.");
+			}
+
+			return ComputeAllowedMaxDivisor(prime);
+		}
+	}
+
+	public DivisorScanSession CreateDivisorSession()
+	{
+		lock (_sync)
+		{
+			if (!_isConfigured)
+			{
+				throw new InvalidOperationException("ConfigureFromMaxPrime must be called before using the tester.");
+			}
+
+			return new DivisorScanSession(this);
+		}
+	}
 }