diff --git a/EvenPerfectBitScanner.Tests/ProgramTests.cs b/EvenPerfectBitScanner.Tests/ProgramTests.cs
index fecddbf..e4fb578 100644
--- a/EvenPerfectBitScanner.Tests/ProgramTests.cs
+++ b/EvenPerfectBitScanner.Tests/ProgramTests.cs
@@ -540,7 +540,7 @@ public class ProgramTests
     public void Main_displays_help_when_requested()
     {
         var main = typeof(Program).GetMethod("Main", BindingFlags.NonPublic | BindingFlags.Static)!;
-        using var writer = new StringWriter();
+        var writer = new StringWriter();
         TextWriter original = Console.Out;
         Console.SetOut(writer);
 
@@ -554,6 +554,7 @@ public class ProgramTests
         }
 
         string output = writer.ToString();
+        writer.Dispose();
         output.Should().Contain("Usage:");
         output.Should().Contain("--mersenne-device=cpu|gpu");
         output.Should().Contain("--primes-device=cpu|gpu");
diff --git a/EvenPerfectBitScanner/Program.cs b/EvenPerfectBitScanner/Program.cs
index b08f418..75aa570 100644
--- a/EvenPerfectBitScanner/Program.cs
+++ b/EvenPerfectBitScanner/Program.cs
@@ -720,7 +720,21 @@ internal static class Program
 
         if (_useByDivisorMode)
         {
-            RunByDivisorMode(byDivisorCandidates, threadCount, _primeTestLimit);
+            if (_byDivisorTester is null)
+            {
+                throw new InvalidOperationException("By-divisor tester has not been initialized.");
+            }
+
+            MersenneNumberDivisorByDivisorTester.Run(
+                    byDivisorCandidates,
+                    _byDivisorTester,
+                    _byDivisorPreviousResults,
+                    _byDivisorStartPrime,
+                    static () => _lastCompositeP = true,
+                    static () => _lastCompositeP = false,
+                    PrintResult,
+                    threadCount,
+                    _primeTestLimit);
             FlushBuffer();
             StringBuilderPool.Return(_outputBuilder!);
             return;
@@ -861,27 +875,6 @@ internal static class Program
         return candidates;
     }
 
-    private static void RunByDivisorMode(List<ulong> candidates, int threadCount, TimeSpan? primeTestLimit)
-    {
-        if (_byDivisorTester is null)
-        {
-            throw new InvalidOperationException("By-divisor tester has not been initialized.");
-        }
-
-        // TODO: Inline this helper by invoking MersenneNumberDivisorByDivisorTester.Run directly at call sites so
-        // the CLI avoids this extra frame during the hot composite-reporting path.
-        MersenneNumberDivisorByDivisorTester.Run(
-                candidates,
-                _byDivisorTester,
-                _byDivisorPreviousResults,
-                _byDivisorStartPrime,
-                static () => _lastCompositeP = true,
-                static () => _lastCompositeP = false,
-                PrintResult,
-                threadCount,
-                primeTestLimit);
-    }
-
     private static unsafe void LoadResultsFile(string resultsFileName, Action<ulong, bool, bool> lineProcessorAction)
     {
         using FileStream readStream = new(resultsFileName, FileMode.Open, FileAccess.Read, FileShare.ReadWrite);
diff --git a/PerfectNumbers.Core.Tests/Gpu/GpuPrimeLimiterTests.cs b/PerfectNumbers.Core.Tests/Gpu/GpuPrimeLimiterTests.cs
index 3b605cc..62d7ee5 100644
--- a/PerfectNumbers.Core.Tests/Gpu/GpuPrimeLimiterTests.cs
+++ b/PerfectNumbers.Core.Tests/Gpu/GpuPrimeLimiterTests.cs
@@ -18,13 +18,14 @@ public class GpuPrimeLimiterTests
         var startedSecond = new TaskCompletionSource<bool>();
         var sw = new Stopwatch();
 
-        using var first = GpuPrimeWorkLimiter.Acquire();
+        var first = GpuPrimeWorkLimiter.Acquire();
         sw.Start();
 
         var secondTask = Task.Run(() =>
         {
-            using var second = GpuPrimeWorkLimiter.Acquire();
+            var second = GpuPrimeWorkLimiter.Acquire();
             startedSecond.TrySetResult(true);
+            second.Dispose();
         });
 
         // Give secondTask a moment to attempt acquire (it should block).
diff --git a/PerfectNumbers.Core.Tests/Gpu/NttMontgomeryTests.cs b/PerfectNumbers.Core.Tests/Gpu/NttMontgomeryTests.cs
index 57fe816..90937f1 100644
--- a/PerfectNumbers.Core.Tests/Gpu/NttMontgomeryTests.cs
+++ b/PerfectNumbers.Core.Tests/Gpu/NttMontgomeryTests.cs
@@ -34,23 +34,19 @@ public class NttMontgomeryTests
 
         // Device execution
         GpuContextPool.GpuContextLease context = GpuContextPool.Rent();
-        try
-        {
-            var accelerator = context.Accelerator;
-            using var buffer = accelerator.Allocate1D<GpuUInt128>(gpu.Length);
-            buffer.View.CopyFromCPU(ref gpu[0], gpu.Length);
-            NttGpuMath.SquareDevice(accelerator, buffer.View, modulus, primitiveRoot);
-            buffer.View.CopyToCPU(ref gpu[0], gpu.Length);
+        var accelerator = context.Accelerator;
+        var buffer = accelerator.Allocate1D<GpuUInt128>(gpu.Length);
+        buffer.View.CopyFromCPU(ref gpu[0], gpu.Length);
+        NttGpuMath.SquareDevice(accelerator, buffer.View, modulus, primitiveRoot);
+        buffer.View.CopyToCPU(ref gpu[0], gpu.Length);
 
-            for (int i = 0; i < cpu.Length; i++)
-            {
-                gpu[i].Should().Be(cpu[i]);
-            }
-        }
-        finally
+        for (int i = 0; i < cpu.Length; i++)
         {
-            context.Dispose();
+            gpu[i].Should().Be(cpu[i]);
         }
+
+        buffer.Dispose();
+        context.Dispose();
     }
 
     [Fact]
diff --git a/PerfectNumbers.Core.Tests/GpuWorkLimiterTests.cs b/PerfectNumbers.Core.Tests/GpuWorkLimiterTests.cs
index 7e51ad4..feefce1 100644
--- a/PerfectNumbers.Core.Tests/GpuWorkLimiterTests.cs
+++ b/PerfectNumbers.Core.Tests/GpuWorkLimiterTests.cs
@@ -17,13 +17,14 @@ public class GpuWorkLimiterTests
         var startedSecond = new TaskCompletionSource<bool>();
         var sw = new Stopwatch();
 
-        using var first = GpuWorkLimiter.Acquire();
+        var first = GpuWorkLimiter.Acquire();
         sw.Start();
 
         var secondTask = Task.Run(() =>
         {
-            using var second = GpuWorkLimiter.Acquire();
+            var second = GpuWorkLimiter.Acquire();
             startedSecond.TrySetResult(true);
+            second.Dispose();
         });
 
         await Task.Delay(50);
diff --git a/PerfectNumbers.Core.Tests/MersenneDivisorCyclesGenerateGpuTests.cs b/PerfectNumbers.Core.Tests/MersenneDivisorCyclesGenerateGpuTests.cs
index af63152..ddc21a3 100644
--- a/PerfectNumbers.Core.Tests/MersenneDivisorCyclesGenerateGpuTests.cs
+++ b/PerfectNumbers.Core.Tests/MersenneDivisorCyclesGenerateGpuTests.cs
@@ -14,8 +14,8 @@ public class MersenneDivisorCyclesGenerateGpuTests
         {
             MersenneDivisorCycles.GenerateGpu(path, maxDivisor: 50UL, batchSize: 16, skipCount: 0, nextPosition: 0);
             var pairs = new List<(ulong divisor, ulong cycle)>();
-            using var stream = new FileStream(path, FileMode.Open, FileAccess.Read);
-            using var reader = new BinaryReader(stream);
+            var stream = new FileStream(path, FileMode.Open, FileAccess.Read);
+            var reader = new BinaryReader(stream);
             while (stream.Position < stream.Length)
             {
                 ulong d = reader.ReadUInt64();
@@ -41,6 +41,9 @@ public class MersenneDivisorCyclesGenerateGpuTests
                 MontgomeryDivisorData divisorData = MontgomeryDivisorData.FromModulus(pair.divisor);
                 pair.cycle.Should().Be(MersenneDivisorCycles.CalculateCycleLength(pair.divisor, divisorData));
             }
+
+            reader.Dispose();
+            stream.Dispose();
         }
         finally
         {
@@ -61,8 +64,8 @@ public class MersenneDivisorCyclesGenerateGpuTests
             MersenneDivisorCycles.Generate(path, maxDivisor: 50UL, threads: 4);
 
             var pairs = new List<(ulong divisor, ulong cycle)>();
-            using var stream = new FileStream(path, FileMode.Open, FileAccess.Read);
-            using var reader = new BinaryReader(stream);
+            var stream = new FileStream(path, FileMode.Open, FileAccess.Read);
+            var reader = new BinaryReader(stream);
             while (stream.Position < stream.Length)
             {
                 ulong divisor = reader.ReadUInt64();
@@ -95,6 +98,9 @@ public class MersenneDivisorCyclesGenerateGpuTests
 
             var (_, completeCount) = MersenneDivisorCycles.FindLast(path);
             completeCount.Should().Be(expected.Count);
+
+            reader.Dispose();
+            stream.Dispose();
         }
         finally
         {
diff --git a/PerfectNumbers.Core.Tests/MersenneNumberDivisorCpuTesterTests.cs b/PerfectNumbers.Core.Tests/MersenneNumberDivisorCpuTesterTests.cs
index 4b01a18..609a27d 100644
--- a/PerfectNumbers.Core.Tests/MersenneNumberDivisorCpuTesterTests.cs
+++ b/PerfectNumbers.Core.Tests/MersenneNumberDivisorCpuTesterTests.cs
@@ -46,7 +46,7 @@ public class MersenneNumberDivisorCpuTesterTests
         var tester = new MersenneNumberDivisorByDivisorCpuTester();
         tester.ConfigureFromMaxPrime(197UL);
 
-        using var session = tester.CreateDivisorSession();
+        var session = tester.CreateDivisorSession();
         ulong[] exponents = { 59UL, 61UL, 67UL, 71UL, 73UL, 79UL, 117_652UL, 333_380UL };
         byte[] hits = new byte[exponents.Length];
 
@@ -74,6 +74,8 @@ public class MersenneNumberDivisorCpuTesterTests
         }
 
         hits.Should().Equal(expected2);
+
+        session.Dispose();
     }
 
     [Fact]
@@ -83,7 +85,7 @@ public class MersenneNumberDivisorCpuTesterTests
         var tester = new MersenneNumberDivisorByDivisorCpuTester();
         tester.ConfigureFromMaxPrime(197UL);
 
-        using var session = tester.CreateDivisorSession();
+        var session = tester.CreateDivisorSession();
         MontgomeryDivisorData divisorData = MontgomeryDivisorData.FromModulus(LargeDivisor1);
         ulong cycle = MersenneDivisorCycles.CalculateCycleLength(LargeDivisor1, divisorData);
 
@@ -98,6 +100,8 @@ public class MersenneNumberDivisorCpuTesterTests
         }
 
         hits.Should().Equal(expected);
+
+        session.Dispose();
     }
 
     [Fact]
@@ -107,7 +111,7 @@ public class MersenneNumberDivisorCpuTesterTests
         var tester = new MersenneNumberDivisorByDivisorCpuTester();
         tester.ConfigureFromMaxPrime(197UL);
 
-        using var session = tester.CreateDivisorSession();
+        var session = tester.CreateDivisorSession();
         MontgomeryDivisorData divisorData = MontgomeryDivisorData.FromModulus(LargeDivisor2);
         ulong cycle = MersenneDivisorCycles.CalculateCycleLength(LargeDivisor2, divisorData);
 
@@ -122,5 +126,7 @@ public class MersenneNumberDivisorCpuTesterTests
         }
 
         hits.Should().Equal(expected);
+
+        session.Dispose();
     }
 }
diff --git a/PerfectNumbers.Core.Tests/MersenneNumberDivisorGpuTesterTests.cs b/PerfectNumbers.Core.Tests/MersenneNumberDivisorGpuTesterTests.cs
index 3ec73c6..7b32931 100644
--- a/PerfectNumbers.Core.Tests/MersenneNumberDivisorGpuTesterTests.cs
+++ b/PerfectNumbers.Core.Tests/MersenneNumberDivisorGpuTesterTests.cs
@@ -157,7 +157,7 @@ public class MersenneNumberDivisorGpuTesterTests
         var tester = new MersenneNumberDivisorByDivisorGpuTester();
         tester.ConfigureFromMaxPrime(197UL);
 
-        using var session = tester.CreateDivisorSession();
+        var session = tester.CreateDivisorSession();
         ulong[] exponents = { 59UL, 61UL, 67UL, 71UL, 73UL, 79UL, 117_652UL, 333_380UL };
         byte[] hits = new byte[exponents.Length];
 
@@ -185,6 +185,8 @@ public class MersenneNumberDivisorGpuTesterTests
         }
 
         hits.Should().Equal(expected2);
+
+        session.Dispose();
     }
 
     [Fact]
@@ -198,7 +200,7 @@ public class MersenneNumberDivisorGpuTesterTests
 
         tester.ConfigureFromMaxPrime(197UL);
 
-        using var session = tester.CreateDivisorSession();
+        var session = tester.CreateDivisorSession();
         ulong[] exponents = { 59UL, 61UL, 67UL, 71UL, 73UL, 79UL, 117_652UL };
         byte[] hits = new byte[exponents.Length];
 
@@ -213,6 +215,8 @@ public class MersenneNumberDivisorGpuTesterTests
         }
 
         hits.Should().Equal(expected);
+
+        session.Dispose();
     }
 
     [Fact]
@@ -225,7 +229,7 @@ public class MersenneNumberDivisorGpuTesterTests
         };
         tester.ConfigureFromMaxPrime(197UL);
 
-        using var session = tester.CreateDivisorSession();
+        var session = tester.CreateDivisorSession();
         MontgomeryDivisorData divisorData = MontgomeryDivisorData.FromModulus(4_000_169UL);
         ulong cycle = MersenneDivisorCycles.CalculateCycleLength(4_000_169UL, divisorData);
 
@@ -252,6 +256,8 @@ public class MersenneNumberDivisorGpuTesterTests
 
         expected.Should().Equal(direct);
         hits.Should().Equal(direct);
+
+        session.Dispose();
     }
 
     [Fact]
@@ -264,7 +270,7 @@ public class MersenneNumberDivisorGpuTesterTests
         };
         tester.ConfigureFromMaxPrime(197UL);
 
-        using var session = tester.CreateDivisorSession();
+        var session = tester.CreateDivisorSession();
         MontgomeryDivisorData divisorData = MontgomeryDivisorData.FromModulus(4_000_561UL);
         ulong cycle = MersenneDivisorCycles.CalculateCycleLength(4_000_561UL, divisorData);
 
@@ -291,6 +297,8 @@ public class MersenneNumberDivisorGpuTesterTests
 
         expected.Should().Equal(direct);
         hits.Should().Equal(direct);
+
+        session.Dispose();
     }
 
     [Fact]
@@ -300,7 +308,7 @@ public class MersenneNumberDivisorGpuTesterTests
         var tester = new MersenneNumberDivisorByDivisorGpuTester();
         tester.ConfigureFromMaxPrime(197UL);
 
-        using var session = tester.CreateDivisorSession();
+        var session = tester.CreateDivisorSession();
         ulong[] exponents = { 59UL, 67UL, 117_652UL };
         byte[] hits = new byte[exponents.Length];
 
@@ -316,6 +324,8 @@ public class MersenneNumberDivisorGpuTesterTests
         }
 
         hits.Should().Equal(expected);
+
+        session.Dispose();
     }
 
     [Fact]
@@ -329,7 +339,7 @@ public class MersenneNumberDivisorGpuTesterTests
 
         tester.ConfigureFromMaxPrime(197UL);
 
-        using var session = tester.CreateDivisorSession();
+        var session = tester.CreateDivisorSession();
         ulong[] exponents = { 59UL, 61UL, 67UL, 71UL, 73UL, 79UL, 117_652UL };
         byte[] hits = new byte[exponents.Length];
 
@@ -344,6 +354,8 @@ public class MersenneNumberDivisorGpuTesterTests
         }
 
         hits.Should().Equal(expected);
+
+        session.Dispose();
     }
 }
 
diff --git a/PerfectNumbers.Core/Gpu/GpuKernelPool.cs b/PerfectNumbers.Core/Gpu/GpuKernelPool.cs
index f65d521..d921377 100644
--- a/PerfectNumbers.Core/Gpu/GpuKernelPool.cs
+++ b/PerfectNumbers.Core/Gpu/GpuKernelPool.cs
@@ -76,10 +76,10 @@ public readonly struct ResiduePrimeViews(
     public readonly ArrayView1D<ulong, Stride1D.Dense> LastSevenPow2 = lastSevenPow2;
 }
 
-public ref struct GpuKernelLease(IDisposable limiter, GpuContextLease gpu, KernelContainer kernels)
+public ref struct GpuKernelLease(GpuPrimeWorkLimiter.Lease limiter, GpuContextLease gpu, KernelContainer kernels)
 {
     private bool disposedValue;
-    private IDisposable? _limiter = limiter;
+    private GpuPrimeWorkLimiter.Lease? _limiter = limiter;
     private GpuContextLease? _gpu = gpu;
     private KernelContainer _kernels = kernels;
     private AcceleratorStream? _stream;
diff --git a/PerfectNumbers.Core/Gpu/GpuPrimeWorkLimiter.cs b/PerfectNumbers.Core/Gpu/GpuPrimeWorkLimiter.cs
index 9602b0c..78ee601 100644
--- a/PerfectNumbers.Core/Gpu/GpuPrimeWorkLimiter.cs
+++ b/PerfectNumbers.Core/Gpu/GpuPrimeWorkLimiter.cs
@@ -6,13 +6,13 @@ public static class GpuPrimeWorkLimiter
     private static readonly object _lock = new();
     private static int _currentLimit = 1;
 
-    public static IDisposable Acquire()
+    public static Lease Acquire()
     {
         // TODO: Inline the pooled limiter guard from the GpuLimiterThroughputBenchmarks so prime-sieve
         // GPU jobs stop allocating Releaser instances on every acquisition.
         var sem = _semaphore;
         sem.Wait();
-        return new Releaser(sem);
+        return new Lease(sem);
     }
 
     public static void SetLimit(int value)
@@ -36,12 +36,12 @@ public static class GpuPrimeWorkLimiter
         }
     }
 
-    private sealed class Releaser : IDisposable
+    public sealed class Lease
     {
         private readonly SemaphoreSlim _sem;
         private bool _disposed;
 
-        public Releaser(SemaphoreSlim sem)
+        public Lease(SemaphoreSlim sem)
         {
             _sem = sem;
         }
diff --git a/PerfectNumbers.Core/Gpu/GpuWorkLimiter.cs b/PerfectNumbers.Core/Gpu/GpuWorkLimiter.cs
index a27b11c..557b317 100644
--- a/PerfectNumbers.Core/Gpu/GpuWorkLimiter.cs
+++ b/PerfectNumbers.Core/Gpu/GpuWorkLimiter.cs
@@ -6,14 +6,14 @@ public static class GpuWorkLimiter
     private static readonly object _lock = new();
     private static int _currentLimit = 1;
 
-    public static IDisposable Acquire()
+    public static Lease Acquire()
     {
         // TODO: Replace the per-call Releaser allocation with the pooled struct-based guard from the
         // GpuLimiterThroughputBenchmarks so limiter acquisition does not allocate when we enter the
         // GPU scanning hot path.
         var sem = _semaphore;
         sem.Wait();
-        return new Releaser(sem);
+        return new Lease(sem);
     }
 
     public static void SetLimit(int value)
@@ -40,12 +40,12 @@ public static class GpuWorkLimiter
         }
     }
 
-    private sealed class Releaser : IDisposable
+    public sealed class Lease
     {
         private readonly SemaphoreSlim _sem;
         private bool _disposed;
 
-        public Releaser(SemaphoreSlim sem)
+        public Lease(SemaphoreSlim sem)
         {
             _sem = sem;
         }
diff --git a/PerfectNumbers.Core/Gpu/MersenneNumberDivisorByDivisorGpuTester.cs b/PerfectNumbers.Core/Gpu/MersenneNumberDivisorByDivisorGpuTester.cs
index f0f7e65..dbef3b1 100644
--- a/PerfectNumbers.Core/Gpu/MersenneNumberDivisorByDivisorGpuTester.cs
+++ b/PerfectNumbers.Core/Gpu/MersenneNumberDivisorByDivisorGpuTester.cs
@@ -1,6 +1,7 @@
 using System;
 using System.Buffers;
 using System.Collections.Concurrent;
+using System.Collections.Generic;
 using System.Runtime.CompilerServices;
 using System.Runtime.InteropServices;
 using System.Numerics;
@@ -122,6 +123,7 @@ public sealed class MersenneNumberDivisorByDivisorGpuTester : IMersenneNumberDiv
             resources.CycleCandidates,
             resources.CycleLengths,
             resources.CycleCapacity,
+            resources,
             out lastProcessed,
             out coveredRange,
             out processedCount
@@ -186,6 +188,7 @@ public sealed class MersenneNumberDivisorByDivisorGpuTester : IMersenneNumberDiv
         ulong[] cycleCandidates,
         ulong[] cycleLengths,
         int cycleCapacity,
+        BatchResources resources,
         out ulong lastProcessed,
         out bool coveredRange,
         out ulong processedCount)
@@ -340,10 +343,10 @@ public sealed class MersenneNumberDivisorByDivisorGpuTester : IMersenneNumberDiv
 
                 gapView.CopyFromCPU(ref MemoryMarshal.GetReference(gapSpan), chunkCount);
 
-                ComputeRemaindersOnGpu(chunkCount, remainder10, 10, gapView, deltaView, remainder10View, remainderDeltaKernel, remainderScanKernel);
-                ComputeRemaindersOnGpu(chunkCount, remainder8, 8, gapView, deltaView, remainder8View, remainderDeltaKernel, remainderScanKernel);
-                ComputeRemaindersOnGpu(chunkCount, remainder5, 5, gapView, deltaView, remainder5View, remainderDeltaKernel, remainderScanKernel);
-                ComputeRemaindersOnGpu(chunkCount, remainder3, 3, gapView, deltaView, remainder3View, remainderDeltaKernel, remainderScanKernel);
+                ComputeRemaindersOnGpu(chunkCount, remainder10, 10, gapView, deltaView, remainder10View, remainderDeltaKernel, remainderScanKernel, resources.GetRemainderScanConfig);
+                ComputeRemaindersOnGpu(chunkCount, remainder8, 8, gapView, deltaView, remainder8View, remainderDeltaKernel, remainderScanKernel, resources.GetRemainderScanConfig);
+                ComputeRemaindersOnGpu(chunkCount, remainder5, 5, gapView, deltaView, remainder5View, remainderDeltaKernel, remainderScanKernel, resources.GetRemainderScanConfig);
+                ComputeRemaindersOnGpu(chunkCount, remainder3, 3, gapView, deltaView, remainder3View, remainderDeltaKernel, remainderScanKernel, resources.GetRemainderScanConfig);
 
                 BuildCandidateMaskOnGpu(chunkCount, remainder10View, remainder8View, remainder5View, remainder3View, lastIsSevenFlag, maskView, maskSpan, candidateMaskKernel);
 
@@ -461,7 +464,8 @@ public sealed class MersenneNumberDivisorByDivisorGpuTester : IMersenneNumberDiv
         ArrayView1D<byte, Stride1D.Dense> deltaView,
         ArrayView1D<byte, Stride1D.Dense> remainderView,
         Action<Index1D, ArrayView<ulong>, ArrayView<byte>, byte> deltaKernel,
-        Action<KernelConfig, Index1D, ArrayView<byte>, ArrayView<byte>, byte, byte> scanKernel)
+        Action<KernelConfig, Index1D, ArrayView<byte>, ArrayView<byte>, byte, byte> scanKernel,
+        Func<int, KernelConfig> configProvider)
     {
         // GPU batches always contain at least one divisor gap, so the empty-count guard remains commented
         // to eliminate redundant branching.
@@ -471,10 +475,7 @@ public sealed class MersenneNumberDivisorByDivisorGpuTester : IMersenneNumberDiv
         // }
 
         deltaKernel(count, gapView, deltaView, modulus);
-        SharedMemoryConfig sharedMemoryConfig = SharedMemoryConfig.RequestDynamic<int>(count);
-        Index1D gridDim = new Index1D(1);
-        Index1D groupDim = new Index1D(count);
-        KernelConfig kernelConfig = new KernelConfig(gridDim, groupDim, sharedMemoryConfig);
+        KernelConfig kernelConfig = configProvider(count);
 
         scanKernel(kernelConfig, new Index1D(count), deltaView, remainderView, baseRemainder, modulus);
     }
@@ -844,7 +845,7 @@ public sealed class MersenneNumberDivisorByDivisorGpuTester : IMersenneNumberDiv
 
     private void ReturnBatchResources(Accelerator accelerator, BatchResources resources) => _resourcePools.GetOrAdd(accelerator, static _ => []).Add(resources);
 
-    private sealed class BatchResources : IDisposable
+    private sealed class BatchResources
     {
         internal BatchResources(MersenneNumberDivisorByDivisorGpuTester owner, Accelerator accelerator, int capacity)
         {
@@ -906,10 +907,24 @@ public sealed class MersenneNumberDivisorByDivisorGpuTester : IMersenneNumberDiv
         internal readonly Action<KernelConfig, Index1D, ArrayView<byte>, ArrayView<byte>, byte, byte> RemainderScanKernel;
         internal readonly Action<Index1D, ArrayView<byte>, ArrayView<byte>, ArrayView<byte>, ArrayView<byte>, byte, ArrayView<byte>> CandidateMaskKernel;
 
+        private readonly Dictionary<int, KernelConfig> _remainderScanConfigs = new();
+
         internal readonly int Capacity;
 
         internal readonly int CycleCapacity;
 
+        internal KernelConfig GetRemainderScanConfig(int count)
+        {
+            if (!_remainderScanConfigs.TryGetValue(count, out KernelConfig config))
+            {
+                SharedMemoryConfig sharedMemoryConfig = SharedMemoryConfig.RequestDynamic<int>(count);
+                config = new KernelConfig(new Index1D(1), new Index1D(count), sharedMemoryConfig);
+                _remainderScanConfigs[count] = config;
+            }
+
+            return config;
+        }
+
         public void Dispose()
         {
             DivisorsBuffer.Dispose();
@@ -948,7 +963,7 @@ public sealed class MersenneNumberDivisorByDivisorGpuTester : IMersenneNumberDiv
         return ComputeAllowedMaxDivisor(prime, _divisorLimit);
     }
 
-    public DivisorScanSession CreateDivisorSession()
+    public IMersenneNumberDivisorByDivisorTester.IDivisorScanSession CreateDivisorSession()
     {
         // Divisor sessions are handed out only after configuration, so this guard would never trigger on the
         // production path.
@@ -965,8 +980,6 @@ public sealed class MersenneNumberDivisorByDivisorGpuTester : IMersenneNumberDiv
 
         return new DivisorScanSession(this);
     }
-
-    IMersenneNumberDivisorByDivisorTester.IDivisorScanSession IMersenneNumberDivisorByDivisorTester.CreateDivisorSession() => CreateDivisorSession();
 }
 
 
diff --git a/PerfectNumbers.Core/Gpu/MersenneNumberDivisorCandidateGpuEvaluator.cs b/PerfectNumbers.Core/Gpu/MersenneNumberDivisorCandidateGpuEvaluator.cs
index 4177fc7..b3a4aa3 100644
--- a/PerfectNumbers.Core/Gpu/MersenneNumberDivisorCandidateGpuEvaluator.cs
+++ b/PerfectNumbers.Core/Gpu/MersenneNumberDivisorCandidateGpuEvaluator.cs
@@ -6,7 +6,7 @@ using ILGPU.Runtime;
 
 namespace PerfectNumbers.Core.Gpu;
 
-internal sealed class MersenneNumberDivisorCandidateGpuEvaluator : IDisposable
+internal sealed class MersenneNumberDivisorCandidateGpuEvaluator
 {
     private const int RemainderCount = 6;
 
diff --git a/PerfectNumbers.Core/Gpu/MersenneNumberDivisorMontgomeryGpuBuilder.cs b/PerfectNumbers.Core/Gpu/MersenneNumberDivisorMontgomeryGpuBuilder.cs
index 6b8a163..1f71d19 100644
--- a/PerfectNumbers.Core/Gpu/MersenneNumberDivisorMontgomeryGpuBuilder.cs
+++ b/PerfectNumbers.Core/Gpu/MersenneNumberDivisorMontgomeryGpuBuilder.cs
@@ -7,7 +7,7 @@ using ILGPU.Runtime;
 
 namespace PerfectNumbers.Core.Gpu;
 
-internal sealed class MersenneNumberDivisorMontgomeryGpuBuilder : IDisposable
+internal sealed class MersenneNumberDivisorMontgomeryGpuBuilder
 {
     private readonly GpuContextPool.GpuContextLease _lease;
     private readonly Accelerator _accelerator;
diff --git a/PerfectNumbers.Core/Gpu/MersenneNumberDivisorRemainderGpuStepper.cs b/PerfectNumbers.Core/Gpu/MersenneNumberDivisorRemainderGpuStepper.cs
index 721a519..9a9b26e 100644
--- a/PerfectNumbers.Core/Gpu/MersenneNumberDivisorRemainderGpuStepper.cs
+++ b/PerfectNumbers.Core/Gpu/MersenneNumberDivisorRemainderGpuStepper.cs
@@ -6,7 +6,7 @@ using ILGPU.Runtime;
 
 namespace PerfectNumbers.Core.Gpu;
 
-internal sealed class MersenneNumberDivisorRemainderGpuStepper : IDisposable
+internal sealed class MersenneNumberDivisorRemainderGpuStepper
 {
     private const int RemainderCount = 6;
 
diff --git a/PerfectNumbers.Core/Gpu/MersenneNumberDivisorResidueGpuEvaluator.cs b/PerfectNumbers.Core/Gpu/MersenneNumberDivisorResidueGpuEvaluator.cs
index 72c60ad..4af10b6 100644
--- a/PerfectNumbers.Core/Gpu/MersenneNumberDivisorResidueGpuEvaluator.cs
+++ b/PerfectNumbers.Core/Gpu/MersenneNumberDivisorResidueGpuEvaluator.cs
@@ -7,7 +7,7 @@ using ILGPU.Runtime;
 
 namespace PerfectNumbers.Core.Gpu;
 
-internal sealed class MersenneNumberDivisorResidueGpuEvaluator : IDisposable
+internal sealed class MersenneNumberDivisorResidueGpuEvaluator
 {
     private readonly GpuContextPool.GpuContextLease _lease;
     private readonly Accelerator _accelerator;
diff --git a/PerfectNumbers.Core/Gpu/MersenneNumberLucasLehmerGpuTester.cs b/PerfectNumbers.Core/Gpu/MersenneNumberLucasLehmerGpuTester.cs
index 76a137e..377bbd3 100644
--- a/PerfectNumbers.Core/Gpu/MersenneNumberLucasLehmerGpuTester.cs
+++ b/PerfectNumbers.Core/Gpu/MersenneNumberLucasLehmerGpuTester.cs
@@ -327,8 +327,9 @@ public class MersenneNumberLucasLehmerGpuTester
             // TODO: Replace StreamWriter with the pooled TextFileWriter pipeline so persisting NTT
             // parameters reuses the zero-allocation buffered writes highlighted in the scanner I/O
             // benchmarks instead of allocating a new encoder per append.
-            using var writer = new StreamWriter(ParameterFilePath, append: true);
+            var writer = new StreamWriter(ParameterFilePath, append: true);
             writer.WriteLine($"{length} {modulus.High} {modulus.Low} {primitiveRoot.High} {primitiveRoot.Low}");
+            writer.Dispose();
         }
     }
 
@@ -338,7 +339,7 @@ public class MersenneNumberLucasLehmerGpuTester
         ulong foundRoot = 0UL;
         bool success = false;
         var processorCount = Environment.ProcessorCount;
-        using var cts = new System.Threading.CancellationTokenSource();
+        var cts = new System.Threading.CancellationTokenSource();
         object sync = new();
 
         // TODO: Move this Parallel.For to the shared low-overhead work scheduler once the NTT parameter
@@ -381,11 +382,13 @@ public class MersenneNumberLucasLehmerGpuTester
         {
             modulus = new GpuUInt128(0UL, foundCandidate);
             primitiveRoot = new GpuUInt128(0UL, foundRoot);
+            cts.Dispose();
             return true;
         }
 
         modulus = new GpuUInt128(0UL, 18446744069414584321UL);
         primitiveRoot = new GpuUInt128(0UL, 7UL);
+        cts.Dispose();
         return false;
     }
 
diff --git a/PerfectNumbers.Core/Gpu/MontgomeryOddPowerGpu.cs b/PerfectNumbers.Core/Gpu/MontgomeryOddPowerGpu.cs
index d1a9efb..d6f016c 100644
--- a/PerfectNumbers.Core/Gpu/MontgomeryOddPowerGpu.cs
+++ b/PerfectNumbers.Core/Gpu/MontgomeryOddPowerGpu.cs
@@ -64,31 +64,25 @@ internal static class MontgomeryOddPowerGpu
         var accelerator = lease.Accelerator;
         var stream = lease.Stream;
 
-        using var divisorBuffer = accelerator.Allocate1D<MontgomeryDivisorData>(divisorCount);
-        using var countBuffer = accelerator.Allocate1D<int>(divisorCount);
-        using var resultBuffer = accelerator.Allocate1D<ulong>(divisorCount * stride);
+        var divisorBuffer = accelerator.Allocate1D<MontgomeryDivisorData>(divisorCount);
+        var countBuffer = accelerator.Allocate1D<int>(divisorCount);
+        var resultBuffer = accelerator.Allocate1D<ulong>(divisorCount * stride);
 
-        try
-        {
-            divisorBuffer.View.CopyFromCPU(ref MemoryMarshal.GetReference(divisors), divisorCount);
-            countBuffer.View.CopyFromCPU(ref MemoryMarshal.GetReference(oddPowerCounts), divisorCount);
-            resultBuffer.MemSetToZero();
+        divisorBuffer.View.CopyFromCPU(ref MemoryMarshal.GetReference(divisors), divisorCount);
+        countBuffer.View.CopyFromCPU(ref MemoryMarshal.GetReference(oddPowerCounts), divisorCount);
+        resultBuffer.MemSetToZero();
 
-            var kernel = lease.MontgomeryOddPowerKernel;
-            kernel(stream, divisorCount, divisorBuffer.View, countBuffer.View, resultBuffer.View, stride);
-            stream.Synchronize();
+        var kernel = lease.MontgomeryOddPowerKernel;
+        kernel(stream, divisorCount, divisorBuffer.View, countBuffer.View, resultBuffer.View, stride);
+        stream.Synchronize();
 
-            resultBuffer.View.CopyToCPU(ref MemoryMarshal.GetReference(results), divisorCount * stride);
-            return true;
-        }
-        catch
-        {
-            return false;
-        }
-        finally
-        {
-            execution.Dispose();
-            lease.Dispose();
-        }
+        resultBuffer.View.CopyToCPU(ref MemoryMarshal.GetReference(results), divisorCount * stride);
+
+        resultBuffer.Dispose();
+        countBuffer.Dispose();
+        divisorBuffer.Dispose();
+        execution.Dispose();
+        lease.Dispose();
+        return true;
     }
 }
diff --git a/PerfectNumbers.Core/Gpu/NttGpuMath.cs b/PerfectNumbers.Core/Gpu/NttGpuMath.cs
index 821a6b0..fc4b8d2 100644
--- a/PerfectNumbers.Core/Gpu/NttGpuMath.cs
+++ b/PerfectNumbers.Core/Gpu/NttGpuMath.cs
@@ -47,7 +47,7 @@ public static class NttGpuMath
 
     private static readonly ConcurrentDictionary<Accelerator, Action<Index1D, ArrayView<GpuUInt128>, int>> BitReverseKernelCache = new(); // TODO: Precompute bit-reversal kernels during boot so we can remove the concurrent lookup overhead flagged in the benchmarks.
 
-    private sealed class SquareCacheEntry : IDisposable
+    private sealed class SquareCacheEntry
     {
         public MemoryBuffer1D<GpuUInt128, Stride1D.Dense> Buffer { get; }
         public GpuUInt128 Root { get; }
diff --git a/PerfectNumbers.Core/Gpu/PrimeOrderGpuHeuristics.cs b/PerfectNumbers.Core/Gpu/PrimeOrderGpuHeuristics.cs
index 2e066c1..084473c 100644
--- a/PerfectNumbers.Core/Gpu/PrimeOrderGpuHeuristics.cs
+++ b/PerfectNumbers.Core/Gpu/PrimeOrderGpuHeuristics.cs
@@ -1611,7 +1611,7 @@ internal static partial class PrimeOrderGpuHeuristics
     private static ulong CalculateByDoublingKernel(ulong prime)
     {
         ulong order = 1UL;
-        ulong value = 2UL % prime;
+        ulong value = 2UL; // Candidate primes exceed two, so the modulo would always return two.
 
         while (value != 1UL)
         {
diff --git a/PerfectNumbers.Core/IMersenneNumberDivisorByDivisorTester.cs b/PerfectNumbers.Core/IMersenneNumberDivisorByDivisorTester.cs
index d452476..f02cf4f 100644
--- a/PerfectNumbers.Core/IMersenneNumberDivisorByDivisorTester.cs
+++ b/PerfectNumbers.Core/IMersenneNumberDivisorByDivisorTester.cs
@@ -18,8 +18,10 @@ public interface IMersenneNumberDivisorByDivisorTester
 
     void PrepareCandidates(ReadOnlySpan<ulong> primes, Span<ulong> allowedMaxValues);
 
-    public interface IDivisorScanSession : IDisposable
+    public interface IDivisorScanSession
     {
         void CheckDivisor(ulong divisor, in MontgomeryDivisorData divisorData, ulong divisorCycle, ReadOnlySpan<ulong> primes, Span<byte> hits);
+
+        void Dispose();
     }
 }
diff --git a/PerfectNumbers.Core/MersenneDivisorCycles.cs b/PerfectNumbers.Core/MersenneDivisorCycles.cs
index 819c8de..cc1aa29 100644
--- a/PerfectNumbers.Core/MersenneDivisorCycles.cs
+++ b/PerfectNumbers.Core/MersenneDivisorCycles.cs
@@ -196,7 +196,7 @@ public class MersenneDivisorCycles
     public static IEnumerable<(ulong divisor, ulong cycleLength)> EnumerateStream(Stream compressor)
     {
         // Binary pairs: (ulong divisor, ulong cycle)
-        using var reader = new BinaryReader(compressor, Encoding.UTF8, leaveOpen: true);
+        var reader = new BinaryReader(compressor, Encoding.UTF8, leaveOpen: true);
         while (true)
         {
             ulong d, c;
@@ -207,6 +207,7 @@ public class MersenneDivisorCycles
             }
             catch (EndOfStreamException)
             {
+                reader.Dispose();
                 yield break;
             }
 
@@ -829,7 +830,7 @@ public class MersenneDivisorCycles
         // Start fresh to avoid mixing formats
         outputStream.SetLength(0L);
         outputStream.Position = 0L;
-        using var writer = new BinaryWriter(outputStream, Encoding.UTF8, leaveOpen: false);
+        var writer = new BinaryWriter(outputStream, Encoding.UTF8, leaveOpen: false);
         for (var taskIndex = 0; taskIndex < threads; taskIndex++)
         {
             ulong threadStart = start + (ulong)taskIndex * blockSize;
@@ -911,6 +912,7 @@ public class MersenneDivisorCycles
         }
 
         Task.WaitAll(tasks);
+        writer.Dispose();
     }
 
     public static void GenerateGpu(string path, ulong maxDivisor, int batchSize = 1_000_000, long skipCount = 0L, long nextPosition = 0L)
@@ -947,7 +949,7 @@ public class MersenneDivisorCycles
             outputStream.Position = 0L;
         }
 
-        using var writer = new BinaryWriter(outputStream, Encoding.UTF8, leaveOpen: false);
+        var writer = new BinaryWriter(outputStream, Encoding.UTF8, leaveOpen: false);
         try
         {
             while (skipCount > 0L && start <= maxDivisor)
@@ -1022,6 +1024,7 @@ public class MersenneDivisorCycles
             pool.Return(outCycles, clearArray: false);
         }
 
+        writer.Dispose();
         stream.Dispose();
         lease.Dispose();
 
diff --git a/PerfectNumbers.Core/PrimeFactorPartialResult.cs b/PerfectNumbers.Core/PrimeFactorPartialResult.cs
index bdb8d0d..74cc020 100644
--- a/PerfectNumbers.Core/PrimeFactorPartialResult.cs
+++ b/PerfectNumbers.Core/PrimeFactorPartialResult.cs
@@ -3,7 +3,7 @@ using System.Buffers;
 
 namespace PerfectNumbers.Core
 {
-    internal sealed class PrimeFactorPartialResult : IDisposable
+    internal sealed class PrimeFactorPartialResult
     {
         [ThreadStatic]
         private static PrimeFactorPartialResult? s_poolHead;
diff --git a/PerfectNumbers.Core/PrimeTester.cs b/PerfectNumbers.Core/PrimeTester.cs
index ad53846..c3cf08a 100644
--- a/PerfectNumbers.Core/PrimeTester.cs
+++ b/PerfectNumbers.Core/PrimeTester.cs
@@ -176,7 +176,7 @@ public sealed class PrimeTester(bool useInternal = false)
             DevicePrimes.View.CopyFromCPU(primes);
         }
 
-        internal sealed class ScratchBuffers : IDisposable
+        internal sealed class ScratchBuffers
         {
             public MemoryBuffer1D<ulong, Stride1D.Dense> Input { get; private set; }
             public MemoryBuffer1D<byte, Stride1D.Dense> Output { get; private set; }
diff --git a/PerfectNumbers.Core/TextFileWriter.cs b/PerfectNumbers.Core/TextFileWriter.cs
index 39a8f83..dda50f6 100644
--- a/PerfectNumbers.Core/TextFileWriter.cs
+++ b/PerfectNumbers.Core/TextFileWriter.cs
@@ -3,7 +3,7 @@ namespace PerfectNumbers.Core;
 /// <summary>
 /// Provides thread-safe file writing for both text and binary data.
 /// </summary>
-public sealed class TextFileWriter : IDisposable
+public sealed class TextFileWriter
 {
     private readonly FileStream _stream;
     private StreamWriter? _textWriter;
diff --git a/PerfectNumbers.Core/ULongExtensions.cs b/PerfectNumbers.Core/ULongExtensions.cs
index cf45857..3c84598 100644
--- a/PerfectNumbers.Core/ULongExtensions.cs
+++ b/PerfectNumbers.Core/ULongExtensions.cs
@@ -530,7 +530,9 @@ public static class ULongExtensions
         ulong modulus = divisor.Modulus;
         if (exponent == 0UL)
         {
-            return keepMontgomery ? divisor.MontgomeryOne : 1UL % modulus;
+            return keepMontgomery
+                ? divisor.MontgomeryOne
+                : 1UL; // Montgomery moduli stay above one on the scanning path, so the modulo would always yield one.
         }
 
         if (exponent <= Pow2WindowFallbackThreshold)
@@ -622,7 +624,9 @@ public static class ULongExtensions
         ulong modulus = divisor.Modulus;
         if (exponent == 0UL)
         {
-            return keepMontgomery ? divisor.MontgomeryOne : 1UL % modulus;
+            return keepMontgomery
+                ? divisor.MontgomeryOne
+                : 1UL; // Montgomery moduli stay above one on the scanning path, so the modulo would always yield one.
         }
 
         if (exponent <= Pow2WindowFallbackThreshold)
@@ -769,7 +773,7 @@ public static class ULongExtensions
             return;
         }
 
-        ulong baseValue = 2UL % modulus;
+        ulong baseValue = 2UL; // Moduli arrive as odd primes greater than two, so the modulo would always return two.
         destination[0] = baseValue;
         if (oddPowerCount == 1)
         {
diff --git a/apply_patch.cs b/apply_patch.cs
index 71b34f6..e22f034 100644
--- a/apply_patch.cs
+++ b/apply_patch.cs
@@ -335,10 +335,13 @@ if (patchResults.Any(static r => !r.Success))
 
 static List<PatchFile> ParsePatchFromFile(string patchFilePath)
 {
-    using var stream = File.OpenRead(patchFilePath);
-    using var reader = new StreamReader(stream, Encoding.UTF8, detectEncodingFromByteOrderMarks: true, bufferSize: 16_384, leaveOpen: false);
-    using var lineReader = new PatchLineReader(reader);
+    var stream = File.OpenRead(patchFilePath);
+    var reader = new StreamReader(stream, Encoding.UTF8, detectEncodingFromByteOrderMarks: true, bufferSize: 16_384, leaveOpen: false);
+    var lineReader = new PatchLineReader(reader);
     var files = ParsePatch(lineReader);
+    lineReader.Dispose();
+    reader.Dispose();
+    stream.Dispose();
     return files;
 }
 
@@ -500,7 +503,7 @@ static void FinalizePatchFile(
     files.Add(new PatchFile(originalPath, modifiedPath, hunks ?? new List<PatchHunk>(), type));
 }
 
-private sealed class PatchLineReader : IDisposable
+private sealed class PatchLineReader
 {
     private const int ReadBufferSize = 4096;
     private const int InitialLineBufferSize = 256;
diff --git a/tools/GpuNumericTypeProbe.csx b/tools/GpuNumericTypeProbe.csx
index fe3d284..d2ca4b3 100644
--- a/tools/GpuNumericTypeProbe.csx
+++ b/tools/GpuNumericTypeProbe.csx
@@ -54,10 +54,11 @@ static ProbeResult ProbeBigInteger(Accelerator accelerator)
     try
     {
         var kernel = accelerator.LoadAutoGroupedStreamKernel<Index1D, ArrayView1D<int, Stride1D.Dense>>(BigIntegerKernel);
-        using var buffer = accelerator.Allocate1D<int>(1);
+        var buffer = accelerator.Allocate1D<int>(1);
         var extent = new Index1D((int)buffer.Length);
         kernel(extent, buffer.View);
         accelerator.Synchronize();
+        buffer.Dispose();
         return ProbeResult.Success(
             "System.Numerics.BigInteger",
             "Addition, subtraction, multiplication, division, and modulo compile on the CPU accelerator.");
@@ -73,9 +74,10 @@ static ProbeResult ProbeEInteger(Accelerator accelerator)
     try
     {
         var kernel = accelerator.LoadAutoGroupedStreamKernel<Index1D, ArrayView1D<int, Stride1D.Dense>>(EIntegerKernel);
-        using var buffer = accelerator.Allocate1D<int>(1);
+        var buffer = accelerator.Allocate1D<int>(1);
         kernel(new Index1D((int)buffer.Length), buffer.View);
         accelerator.Synchronize();
+        buffer.Dispose();
         return ProbeResult.Success(
             "PeterO.Numbers.EInteger",
             "Addition, subtraction, multiplication, division, and remainder compile on the CPU accelerator.");
@@ -93,7 +95,7 @@ static ProbeResult ProbeERational(Accelerator accelerator)
     try
     {
         var kernel = accelerator.LoadAutoGroupedStreamKernel<Index1D, ArrayView1D<int, Stride1D.Dense>>(ERationalKernel);
-        using var buffer = accelerator.Allocate1D<int>(1);
+        var buffer = accelerator.Allocate1D<int>(1);
         kernel(new Index1D((int)buffer.Length), buffer.View);
         accelerator.Synchronize();
 
@@ -101,6 +103,7 @@ static ProbeResult ProbeERational(Accelerator accelerator)
         details += hasRemainder
             ? " ERational exposes a remainder API that requires further investigation."
             : " ERational does not expose a remainder/modulo API.";
+        buffer.Dispose();
         return ProbeResult.Success("PeterO.Numbers.ERational", details);
     }
     catch (Exception ex)
@@ -120,9 +123,10 @@ static ProbeResult ProbeOpenNumericPrimesIsPrime(Accelerator accelerator)
     try
     {
         var kernel = accelerator.LoadAutoGroupedStreamKernel<Index1D, ArrayView1D<int, Stride1D.Dense>>(OpenNumericPrimesIsPrimeKernel);
-        using var buffer = accelerator.Allocate1D<int>(4);
+        var buffer = accelerator.Allocate1D<int>(4);
         kernel(new Index1D((int)buffer.Length), buffer.View);
         accelerator.Synchronize();
+        buffer.Dispose();
         return ProbeResult.Success(
             "Open.Numeric.Primes.Prime.Numbers.IsPrime",
             "IsPrime compiles and runs on the CPU accelerator kernel.");