From 86990ae0f79bd67545c2e337bece694fc56ec66a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Marek=20Lema=C5=84czyk?= <marek@lemanczyk.eu>
Date: Tue, 28 Oct 2025 13:30:47 +0100
Subject: [PATCH] Remove kernel initialization lock from GPU leases

---
 PerfectNumbers.Core/Gpu/GpuKernelLease.cs | 78 +++++++++--------------
 1 file changed, 30 insertions(+), 48 deletions(-)

diff --git a/PerfectNumbers.Core/Gpu/GpuKernelLease.cs b/PerfectNumbers.Core/Gpu/GpuKernelLease.cs
index 0603ef9..a79af7a 100644
--- a/PerfectNumbers.Core/Gpu/GpuKernelLease.cs
+++ b/PerfectNumbers.Core/Gpu/GpuKernelLease.cs
@@ -56,15 +56,12 @@ public sealed class GpuKernelLease
         get
         {
             var accel = Accelerator; // avoid capturing 'this' in lambda
-            lock (_gpu!.Value.ExecutionLock)
+            return KernelContainer.InitOnce(ref _kernels!.Order, () =>
             {
-                return KernelContainer.InitOnce(ref _kernels!.Order, () =>
-                {
-                    var loaded = accel.LoadAutoGroupedStreamKernel<Index1D, ulong, ulong, ArrayView<GpuUInt128>, ArrayView<ulong>>(OrderKernels.OrderKernelScan);
-                    var kernel = KernelUtil.GetKernel(loaded);
-                    return kernel.CreateLauncherDelegate<Action<AcceleratorStream, Index1D, ulong, ulong, ArrayView<GpuUInt128>, ArrayView<ulong>>>();
-                });
-            }
+                var loaded = accel.LoadAutoGroupedStreamKernel<Index1D, ulong, ulong, ArrayView<GpuUInt128>, ArrayView<ulong>>(OrderKernels.OrderKernelScan);
+                var kernel = KernelUtil.GetKernel(loaded);
+                return kernel.CreateLauncherDelegate<Action<AcceleratorStream, Index1D, ulong, ulong, ArrayView<GpuUInt128>, ArrayView<ulong>>>();
+            });
         }
     }
 
@@ -73,15 +70,12 @@ public sealed class GpuKernelLease
         get
         {
             var accel = Accelerator;
-            lock (_gpu!.Value.ExecutionLock)
+            return KernelContainer.InitOnce(ref _kernels!.Pow2Mod, () =>
             {
-                return KernelContainer.InitOnce(ref _kernels!.Pow2Mod, () =>
-                {
-                    var loaded = accel.LoadAutoGroupedStreamKernel<Index1D, ulong, GpuUInt128, GpuUInt128, byte, ulong, ResidueAutomatonArgs, ArrayView<ulong>, ArrayView1D<ulong, Stride1D.Dense>, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>>(Pow2ModKernels.Pow2ModKernelScan);
-                    var kernel = KernelUtil.GetKernel(loaded);
-                    return kernel.CreateLauncherDelegate<Action<AcceleratorStream, Index1D, ulong, GpuUInt128, GpuUInt128, byte, ulong, ResidueAutomatonArgs, ArrayView<ulong>, ArrayView1D<ulong, Stride1D.Dense>, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>>>();
-                });
-            }
+                var loaded = accel.LoadAutoGroupedStreamKernel<Index1D, ulong, GpuUInt128, GpuUInt128, byte, ulong, ResidueAutomatonArgs, ArrayView<ulong>, ArrayView1D<ulong, Stride1D.Dense>, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>>(Pow2ModKernels.Pow2ModKernelScan);
+                var kernel = KernelUtil.GetKernel(loaded);
+                return kernel.CreateLauncherDelegate<Action<AcceleratorStream, Index1D, ulong, GpuUInt128, GpuUInt128, byte, ulong, ResidueAutomatonArgs, ArrayView<ulong>, ArrayView1D<ulong, Stride1D.Dense>, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>>>();
+            });
         }
     }
 
@@ -90,15 +84,12 @@ public sealed class GpuKernelLease
         get
         {
             var accel = Accelerator;
-            lock (_gpu!.Value.ExecutionLock)
+            return KernelContainer.InitOnce(ref _kernels!.Incremental, () =>
             {
-                return KernelContainer.InitOnce(ref _kernels!.Incremental, () =>
-                {
-                    var loaded = accel.LoadAutoGroupedStreamKernel<Index1D, ulong, GpuUInt128, GpuUInt128, byte, ulong, ulong, ulong, ulong, ulong, ArrayView<ulong>, ArrayView1D<ulong, Stride1D.Dense>>(IncrementalKernels.IncrementalKernelScan);
-                    var kernel = KernelUtil.GetKernel(loaded);
-                    return kernel.CreateLauncherDelegate<Action<AcceleratorStream, Index1D, ulong, GpuUInt128, GpuUInt128, byte, ulong, ulong, ulong, ulong, ulong, ArrayView<ulong>, ArrayView1D<ulong, Stride1D.Dense>>>();
-                });
-            }
+                var loaded = accel.LoadAutoGroupedStreamKernel<Index1D, ulong, GpuUInt128, GpuUInt128, byte, ulong, ulong, ulong, ulong, ulong, ArrayView<ulong>, ArrayView1D<ulong, Stride1D.Dense>>(IncrementalKernels.IncrementalKernelScan);
+                var kernel = KernelUtil.GetKernel(loaded);
+                return kernel.CreateLauncherDelegate<Action<AcceleratorStream, Index1D, ulong, GpuUInt128, GpuUInt128, byte, ulong, ulong, ulong, ulong, ulong, ArrayView<ulong>, ArrayView1D<ulong, Stride1D.Dense>>>();
+            });
         }
     }
 
@@ -107,15 +98,12 @@ public sealed class GpuKernelLease
         get
         {
             var accel = Accelerator;
-            lock (_gpu!.Value.ExecutionLock)
+            return KernelContainer.InitOnce(ref _kernels!.IncrementalOrder, () =>
             {
-                return KernelContainer.InitOnce(ref _kernels!.IncrementalOrder, () =>
-                {
-                    var loaded = accel.LoadAutoGroupedStreamKernel<Index1D, ulong, GpuUInt128, GpuUInt128, byte, ulong, ResidueAutomatonArgs, ArrayView<int>, ArrayView1D<ulong, Stride1D.Dense>>(IncrementalKernels.IncrementalOrderKernelScan);
-                    var kernel = KernelUtil.GetKernel(loaded);
-                    return kernel.CreateLauncherDelegate<Action<AcceleratorStream, Index1D, ulong, GpuUInt128, GpuUInt128, byte, ulong, ResidueAutomatonArgs, ArrayView<int>, ArrayView1D<ulong, Stride1D.Dense>>>();
-                });
-            }
+                var loaded = accel.LoadAutoGroupedStreamKernel<Index1D, ulong, GpuUInt128, GpuUInt128, byte, ulong, ResidueAutomatonArgs, ArrayView<int>, ArrayView1D<ulong, Stride1D.Dense>>(IncrementalKernels.IncrementalOrderKernelScan);
+                var kernel = KernelUtil.GetKernel(loaded);
+                return kernel.CreateLauncherDelegate<Action<AcceleratorStream, Index1D, ulong, GpuUInt128, GpuUInt128, byte, ulong, ResidueAutomatonArgs, ArrayView<int>, ArrayView1D<ulong, Stride1D.Dense>>>();
+            });
         }
     }
 
@@ -124,15 +112,12 @@ public sealed class GpuKernelLease
         get
         {
             var accel = Accelerator;
-            lock (_gpu!.Value.ExecutionLock)
+            return KernelContainer.InitOnce(ref _kernels!.Pow2ModOrder, () =>
             {
-                return KernelContainer.InitOnce(ref _kernels!.Pow2ModOrder, () =>
-                {
-                    var loaded = accel.LoadAutoGroupedStreamKernel<Index1D, ulong, GpuUInt128, GpuUInt128, byte, ulong, ResidueAutomatonArgs, ArrayView<int>, ArrayView1D<ulong, Stride1D.Dense>>(Pow2ModKernels.Pow2ModOrderKernelScan);
-                    var kernel = KernelUtil.GetKernel(loaded);
-                    return kernel.CreateLauncherDelegate<Action<AcceleratorStream, Index1D, ulong, GpuUInt128, GpuUInt128, byte, ulong, ResidueAutomatonArgs, ArrayView<int>, ArrayView1D<ulong, Stride1D.Dense>>>();
-                });
-            }
+                var loaded = accel.LoadAutoGroupedStreamKernel<Index1D, ulong, GpuUInt128, GpuUInt128, byte, ulong, ResidueAutomatonArgs, ArrayView<int>, ArrayView1D<ulong, Stride1D.Dense>>(Pow2ModKernels.Pow2ModOrderKernelScan);
+                var kernel = KernelUtil.GetKernel(loaded);
+                return kernel.CreateLauncherDelegate<Action<AcceleratorStream, Index1D, ulong, GpuUInt128, GpuUInt128, byte, ulong, ResidueAutomatonArgs, ArrayView<int>, ArrayView1D<ulong, Stride1D.Dense>>>();
+            });
         }
     }
 
@@ -141,15 +126,12 @@ public sealed class GpuKernelLease
         get
         {
             var accel = Accelerator;
-            lock (_gpu!.Value.ExecutionLock)
+            return KernelContainer.InitOnce(ref _kernels!.SmallPrimeFactor, () =>
             {
-                return KernelContainer.InitOnce(ref _kernels!.SmallPrimeFactor, () =>
-                {
-                    var loaded = accel.LoadAutoGroupedStreamKernel<Index1D, ulong, uint, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, int, ArrayView1D<ulong, Stride1D.Dense>, ArrayView1D<int, Stride1D.Dense>, ArrayView1D<int, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>>(SmallPrimeFactorKernels.SmallPrimeFactorKernelScan);
-                    var kernel = KernelUtil.GetKernel(loaded);
-                    return kernel.CreateLauncherDelegate<Action<AcceleratorStream, Index1D, ulong, uint, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, int, ArrayView1D<ulong, Stride1D.Dense>, ArrayView1D<int, Stride1D.Dense>, ArrayView1D<int, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>>>();
-                });
-            }
+                var loaded = accel.LoadAutoGroupedStreamKernel<Index1D, ulong, uint, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, int, ArrayView1D<ulong, Stride1D.Dense>, ArrayView1D<int, Stride1D.Dense>, ArrayView1D<int, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>>(SmallPrimeFactorKernels.SmallPrimeFactorKernelScan);
+                var kernel = KernelUtil.GetKernel(loaded);
+                return kernel.CreateLauncherDelegate<Action<AcceleratorStream, Index1D, ulong, uint, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, int, ArrayView1D<ulong, Stride1D.Dense>, ArrayView1D<int, Stride1D.Dense>, ArrayView1D<int, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>>>();
+            });
         }
     }