Various optimizations

EvenPerfectBitScanner.Tests/ProgramTests.cs

@@ -55,7 +55,7 @@ public class ProgramTests
     [InlineData(81UL, false)]
     public void IsPrime_identifies_primes_correctly(ulong n, bool expected)
     {
-        new PrimeTester().IsPrime(n, CancellationToken.None).Should().Be(expected);
+        PrimeTester.IsPrime(n, CancellationToken.None).Should().Be(expected);
     }
 
     [Fact]

EvenPerfectBitScanner/Program.cs

@@ -622,7 +622,7 @@ internal static class Program
 		if (!_cliArguments.UseByDivisor)
 		{
 			if (!(GpuContextPool.ForceCpu
-											? PrimeTesters.Value!.IsPrime(p, CancellationToken.None)
+											? PrimeTester.IsPrime(p, CancellationToken.None)
 											: PrimeTesters.Value!.IsPrimeGpu(p, CancellationToken.None)))
 			{
 				_lastCompositeP = true;

PerfectNumbers.Core.Tests/PrimeTester/PrimeTesterIsPrimeTests.cs

@@ -26,7 +26,7 @@ public class PrimeTesterIsPrimeTests
     {
         var tester = new PrimeTester();
 
-        tester.IsPrime(n, CancellationToken.None).Should().Be(expected);
+		PrimeTester.IsPrime(n, CancellationToken.None).Should().Be(expected);
     }
 
     [Theory]

PerfectNumbers.Core/Gpu/GpuPrimeWorkLimiter.cs

@@ -3,7 +3,6 @@ namespace PerfectNumbers.Core.Gpu;
 public static class GpuPrimeWorkLimiter
 {
     private static SemaphoreSlim _semaphore = new(1, int.MaxValue);
-    private static readonly object _lock = new();
     private static int _currentLimit = 1;
 
     public static Lease Acquire()
@@ -11,29 +10,21 @@ public static class GpuPrimeWorkLimiter
         // TODO: Inline the pooled limiter guard from the GpuLimiterThroughputBenchmarks so prime-sieve
         // GPU jobs stop allocating Lease instances on every acquisition.
         var sem = _semaphore;
-        sem.Wait();
+        // sem.Wait();
         return new Lease(sem);
     }
 
     public static void SetLimit(int value)
     {
-        if (value < 1)
-        {
-            value = 1;
-        }
+		if (value == _currentLimit)
+		{
+			return;
+		}
 
-        lock (_lock)
-        {
-            if (value == _currentLimit)
-            {
-                return;
-            }
-
-            // TODO: Switch to a shared limiter implementation with GpuWorkLimiter so we can coordinate CPU/GPU prime work using
-            // the same semaphore pool without reallocating per adjustment.
-            _semaphore = new SemaphoreSlim(value, value);
-            _currentLimit = value;
-        }
+		// TODO: Switch to a shared limiter implementation with GpuWorkLimiter so we can coordinate CPU/GPU prime work using
+		// the same semaphore pool without reallocating per adjustment.
+		_semaphore = new SemaphoreSlim(value, value);
+		_currentLimit = value;
     }
 
     public sealed class Lease
@@ -47,7 +38,7 @@ public static class GpuPrimeWorkLimiter
 
         public void Dispose()
         {
-            _sem.Release();
+            // _sem.Release();
         }
     }
 }

PerfectNumbers.Core/Gpu/MersenneNumberDivisorByDivisorGpuTester.cs

@@ -94,7 +94,7 @@ public sealed class MersenneNumberDivisorByDivisorGpuTester : IMersenneNumberDiv
         var gpuLease = RentAccelerator();
         var accelerator = gpuLease.Accelerator;
 
-        Monitor.Enter(gpuLease.ExecutionLock);
+        // Monitor.Enter(gpuLease.ExecutionLock);
 
         var kernel = GetKernel(accelerator);
         var resources = RentBatchResources(accelerator, batchCapacity);
@@ -121,7 +121,7 @@ public sealed class MersenneNumberDivisorByDivisorGpuTester : IMersenneNumberDiv
             out processedCount);
 
         ReturnBatchResources(resources);
-        Monitor.Exit(gpuLease.ExecutionLock);
+        // Monitor.Exit(gpuLease.ExecutionLock);
         ReturnAccelerator(gpuLease);
 
         if (composite)
@@ -560,7 +560,7 @@ public sealed class MersenneNumberDivisorByDivisorGpuTester : IMersenneNumberDiv
                 }
             }
 
-            Monitor.Enter(_lease.ExecutionLock);
+            // Monitor.Enter(_lease.ExecutionLock);
 
             EnsureExecutionResourcesLocked(length);
 
@@ -601,7 +601,7 @@ public sealed class MersenneNumberDivisorByDivisorGpuTester : IMersenneNumberDiv
             ref byte hitRef = ref MemoryMarshal.GetReference(hitSlice);
             hitView.CopyToCPU(ref hitRef, length);
 
-            Monitor.Exit(_lease.ExecutionLock);
+            // Monitor.Exit(_lease.ExecutionLock);
         }
 
         public void Dispose()

PerfectNumbers.Core/Gpu/MersenneNumberResidueGpuTester.cs

@@ -30,7 +30,7 @@ public class MersenneNumberResidueGpuTester(bool useGpuOrder)
         var gpuLease = RentAccelerator();
         var accelerator = gpuLease.Accelerator;
 
-        Monitor.Enter(gpuLease.ExecutionLock);
+        // Monitor.Enter(gpuLease.ExecutionLock);
 
         var stream = accelerator.CreateStream();
         var kernel = GetPow2ModKernel(accelerator);
@@ -108,7 +108,7 @@ public class MersenneNumberResidueGpuTester(bool useGpuOrder)
 
         ReturnResources(accelerator, resources);
         stream.Dispose();
-        Monitor.Exit(gpuLease.ExecutionLock);
+        // Monitor.Exit(gpuLease.ExecutionLock);
         ReturnAccelerator(gpuLease);
         limiter.Dispose();
     }

PerfectNumbers.Core/HeuristicPrimeTester.cs

@@ -122,10 +122,10 @@ public sealed class HeuristicPrimeTester
             return EvaluateWithOpenNumericFallback(n);
         }
 
-        if (!UseHeuristicGroupBTrialDivision)
-        {
-            return HeuristicTrialDivisionCpu(n, sqrtLimit, nMod10, includeGroupB: false);
-        }
+        // if (!UseHeuristicGroupBTrialDivision)
+        // {
+        //     return HeuristicTrialDivisionCpu(n, sqrtLimit, nMod10, includeGroupB: false);
+        // }
 
         bool compositeDetected = HeuristicTrialDivisionGpuDetectsDivisor(n, sqrtLimit, nMod10);
         return !compositeDetected;
@@ -260,8 +260,8 @@ public sealed class HeuristicPrimeTester
 
     private bool HeuristicTrialDivisionGpuDetectsDivisor(ulong n, ulong sqrtLimit, byte nMod10)
     {
-        int batchCapacity = Math.Max(1, HeuristicGpuDivisorBatchSize);
-        int interleaveBatchSize = Math.Max(1, HeuristicDivisorInterleaveBatchSize);
+        int batchCapacity = HeuristicGpuDivisorBatchSize;
+        int interleaveBatchSize = HeuristicDivisorInterleaveBatchSize;
         var divisorPool = ThreadStaticPools.UlongPool;
         var hitPool = ThreadStaticPools.BytePool;
 
@@ -277,7 +277,7 @@ public sealed class HeuristicPrimeTester
         bool compositeDetected = false;
         int count = 0;
 
-        lock (gpu.ExecutionLock)
+        // lock (gpu.ExecutionLock)
         {
             var input = scratch.Input;
             var output = scratch.Output;