Merge branch 'codex/optimize-gpu-memory-usage-in-kernels' of https://github.com/mlemanczyk/even-perfect-numbers-scanner into codex/optimize-gpu-memory-usage-in-kernels

EvenPerfectBitScanner/Program.cs

@@ -419,6 +419,7 @@ internal static class Program
 
 			if (useByDivisor)
 			{
+				int byDivisorPrimeRange = Math.Max(1, _cliArguments.BlockSize);
 				MersenneNumberDivisorByDivisorTester.Run(
 						byDivisorCandidates,
 						_byDivisorTester!,
@@ -427,7 +428,8 @@ internal static class Program
 						static () => _lastCompositeP = true,
 						static () => _lastCompositeP = false,
 						static (candidate, searchedMersenne, detailedCheck, passedAllTests) => CalculationResultHandler.HandleResult(candidate, searchedMersenne, detailedCheck, passedAllTests, _lastCompositeP),
-						threadCount);
+						threadCount,
+						byDivisorPrimeRange);
 						
 				CalculationResultHandler.FlushBuffer();
 				CalculationResultHandler.ReleaseOutputBuffer();

PerfectNumbers.Core.Tests/MersenneNumberDivisorByDivisorTesterTests.cs

@@ -23,7 +23,8 @@ public class MersenneNumberDivisorByDivisorTesterTests
             markComposite: () => compositeMarks++,
             clearComposite: () => clearedMarks++,
             printResult: (prime, searched, detailed, passed) => results.Add((prime, searched, detailed, passed)),
-            threadCount: 1);
+            threadCount: 1,
+            primesPerTask: 2);
 
         tester.ConfiguredMaxPrime.Should().Be(13UL);
         compositeMarks.Should().Be(2);

PerfectNumbers.Core/MersenneNumberDivisorByDivisorTester.cs

@@ -1,4 +1,5 @@
 using System.Buffers;
+using System.Collections.Concurrent;
 using System.Globalization;
 using System.Runtime.InteropServices;
 
@@ -14,7 +15,8 @@ public static class MersenneNumberDivisorByDivisorTester
 			Action markComposite,
 			Action clearComposite,
 			Action<ulong, bool, bool, bool> printResult,
-			int threadCount)
+			int threadCount,
+			int primesPerTask)
 	{
 		if (candidates.Count == 0)
 		{
@@ -233,6 +235,8 @@ public static class MersenneNumberDivisorByDivisorTester
 		//     workerCount = 1;
 		// }
 
+		int chunkSize = primesPerTask <= 0 ? 1 : primesPerTask;
+
 		void ProcessPrime(ulong prime)
 		{
 			Console.WriteLine($"Task started {prime}");
@@ -266,7 +270,20 @@ public static class MersenneNumberDivisorByDivisorTester
 				TaskScheduler = scheduler
 			};
 
-			Parallel.ForEach(filteredPrimes, options, ProcessPrime);
+			int totalCount = filteredPrimes.Count;
+			int partitionSize = chunkSize < 1 ? 1 : chunkSize;
+			if (partitionSize > totalCount)
+			{
+				partitionSize = totalCount;
+			}
+
+			Parallel.ForEach(Partitioner.Create(0, totalCount, partitionSize), options, range =>
+			{
+				for (int index = range.Item1; index < range.Item2; index++)
+				{
+					ProcessPrime(filteredPrimes[index]);
+				}
+			});
 		}
 	}
 }