mlemanczyk/even-perfect-numbers-scanner
1
0 Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Initial BitTree implementation added.

Browse Source
This commit is contained in:
mlemanczyk committed 2026-01-09 16:13:16 +01:00
1 parent 45e3455ff3
commit d88ad25476
11 files changed
+3817 -367
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

PerfectNumbers.Core/Gpu/Accelerators/PrimeOrderCalculatorAccelerator.cs
+6 -6
View File
@@ -149,7 +149,7 @@ public sealed class PrimeOrderCalculatorAccelerator
private sealed class CalculatorKernels(Accelerator accelerator)
{
public readonly Action<AcceleratorStream, Index1D, ulong, CalculateOrderKernelConfig, ulong, ulong, ulong, ulong, ulong, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, int, OrderKernelBuffers> CalculateOrderKernelLauncher = KernelUtil.GetKernel(accelerator.LoadAutoGroupedStreamKernel<Index1D, ulong, CalculateOrderKernelConfig, ulong, ulong, ulong, ulong, ulong, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, int, OrderKernelBuffers>(PrimeOrderKernels.CalculateOrderKernel)).CreateLauncherDelegate<Action<AcceleratorStream, Index1D, ulong, CalculateOrderKernelConfig, ulong, ulong, ulong, ulong, ulong, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, int, OrderKernelBuffers>>();
// public readonly Action<AcceleratorStream, Index1D, ulong, CalculateOrderKernelConfig, ulong, ulong, ulong, ulong, ulong, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, int, OrderKernelBuffers> CalculateOrderKernelLauncher = KernelUtil.GetKernel(accelerator.LoadAutoGroupedStreamKernel<Index1D, ulong, CalculateOrderKernelConfig, ulong, ulong, ulong, ulong, ulong, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, int, OrderKernelBuffers>(PrimeOrderKernels.CalculateOrderKernel)).CreateLauncherDelegate<Action<AcceleratorStream, Index1D, ulong, CalculateOrderKernelConfig, ulong, ulong, ulong, ulong, ulong, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, int, OrderKernelBuffers>>();
public readonly Kernel CheckFactorsKernel = KernelUtil.GetKernel(accelerator.LoadStreamKernel<int, ulong, ArrayView1D<KeyValuePair<ulong, int>, Stride1D.Dense>, ulong, ulong, ulong, ulong, ulong, ArrayView1D<ulong, Stride1D.Dense>>(PrimeOrderKernels.CheckFactorsKernel));
@@ -177,7 +177,7 @@ public sealed class PrimeOrderCalculatorAccelerator
public readonly Action<AcceleratorStream, Index1D, ulong, ArrayView<ulong>, int, ulong, ArrayView<ulong>> SpecialMaxKernelLauncher = KernelUtil.GetKernel(accelerator.LoadAutoGroupedStreamKernel<Index1D, ulong, ArrayView<ulong>, int, ulong, ArrayView<ulong>>(PrimeOrderGpuHeuristics.EvaluateSpecialMaxCandidatesKernel)).CreateLauncherDelegate<Action<AcceleratorStream, Index1D, ulong, ArrayView<ulong>, int, ulong, ArrayView<ulong>>>();
}
public readonly Action<AcceleratorStream, Index1D, ulong, CalculateOrderKernelConfig, ulong, ulong, ulong, ulong, ulong, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, int, OrderKernelBuffers> CalculateOrderKernelLauncher;
// public readonly Action<AcceleratorStream, Index1D, ulong, CalculateOrderKernelConfig, ulong, ulong, ulong, ulong, ulong, ArrayView1D<uint, Stride1D.Dense>, ArrayView1D<ulong, Stride1D.Dense>, int, OrderKernelBuffers> CalculateOrderKernelLauncher;
public readonly Kernel CheckFactorsKernel;
public readonly Action<AcceleratorStream, Index1D, ArrayView1D<ulong, Stride1D.Dense>, ulong, ulong, ulong, ulong, ulong, ArrayView1D<ulong, Stride1D.Dense>> ConvertToStandardKernelLauncher;
public readonly Action<AcceleratorStream, Index1D, ArrayView<int>, ulong, ulong, ArrayView<ulong>, ArrayView<ulong>> HeuristicCombinedTrialDivisionKernelLauncher;
@@ -421,9 +421,9 @@ public sealed class PrimeOrderCalculatorAccelerator
DevicePrimesPow2LastThree = lastDigitSharedTables.DevicePrimesPow2LastThree.View;
DevicePrimesPow2LastNine = lastDigitSharedTables.DevicePrimesPow2LastNine.View;
DivisorTables = HeuristicCombinedGpuTables
.GetStaticTables(acceleratorIndex)
.CreateViews();
// DivisorTables = HeuristicCombinedGpuTables
// .GetStaticTables(acceleratorIndex)
// .CreateViews();
var smallPrimeFactorTables = SmallPrimeFactorGpuTables.GetStaticTables(acceleratorIndex);
@@ -436,7 +436,7 @@ public sealed class PrimeOrderCalculatorAccelerator
var kernels = _kernels[acceleratorIndex];
CalculateOrderKernelLauncher = kernels.CalculateOrderKernelLauncher;
// CalculateOrderKernelLauncher = kernels.CalculateOrderKernelLauncher;
CheckFactorsKernel = kernels.CheckFactorsKernel;
ConvertToStandardKernelLauncher = kernels.ConvertToStandardKernelLauncher;
HeuristicCombinedTrialDivisionKernelLauncher = kernels.HeuristicCombinedTrialDivisionKernelLauncher;
PerfectNumbers.Core/Gpu/PrimeOrderGpuHeuristics.cs
+15 -15
View File
@@ -272,27 +272,27 @@ internal static partial class PrimeOrderGpuHeuristics
var kernelConfig = new CalculateOrderKernelConfig(previousValue, hasPrevious, limit, config.MaxPowChecks, config.StrictMode);
ref var buffers = ref gpu.CalculateOrderKernelBuffers;
var kernelLauncher = gpu.CalculateOrderKernelLauncher;
// var kernelLauncher = gpu.CalculateOrderKernelLauncher;
AcceleratorStream stream = AcceleratorStreamPool.Rent(acceleratorIndex);
// TODO: Remove the cleaning after the order kernel is modified to always set the result.
resultBufferView.MemSetToZero(stream);
statusBufferView.MemSetToZero(stream);
kernelLauncher(
stream,
1,
prime,
kernelConfig,
divisorData.Modulus,
divisorData.NPrime,
divisorData.MontgomeryOne,
divisorData.MontgomeryTwo,
divisorData.MontgomeryTwoSquared,
smallPrimesView,
smallSquaresView,
(int)smallPrimesView.Length,
buffers);
// kernelLauncher(
// stream,
// 1,
// prime,
// kernelConfig,
// divisorData.Modulus,
// divisorData.NPrime,
// divisorData.MontgomeryOne,
// divisorData.MontgomeryTwo,
// divisorData.MontgomeryTwoSquared,
// smallPrimesView,
// smallSquaresView,
// (int)smallPrimesView.Length,
// buffers);
byte status = 0;