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

GpuScratchBuffer / -Pool naively integrated with PriemOrderCalculatorAccelerator (formerly Pow2MontgomeryAccelerator). FactorEntry replaced with 2 arrays - unused field removed. Small cleanup.

Browse Source
This commit is contained in:
mlemanczyk committed 2025-11-13 14:41:27 +01:00
1 parent 43863d03fb
commit c53aa517f7
14 files changed
+295 -420
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

PerfectNumbers.Core/Gpu/Accelerators/LastDigitGpuTables.cs
+15 -16
View File
@@ -1,4 +1,4 @@
using System.Collections.Concurrent;
using System.Runtime.CompilerServices;
using ILGPU;
using ILGPU.Runtime;
@@ -6,10 +6,13 @@ namespace PerfectNumbers.Core.Gpu.Accelerators;
internal sealed class LastDigitGpuTables
{
private static readonly ConcurrentDictionary<Accelerator, LastDigitGpuTables> _sharedTables = new(20_480, PerfectNumberConstants.RollingAccelerators * 2);
private static readonly Dictionary<Accelerator, LastDigitGpuTables> _sharedTables = new(PerfectNumberConstants.RollingAccelerators);
internal static LastDigitGpuTables EnsureStaticTables(Accelerator accelerator, AcceleratorStream stream)
=> _sharedTables.GetOrAdd(accelerator, _ => new LastDigitGpuTables(accelerator, stream));
[MethodImpl(MethodImplOptions.AggressiveInlining)]
internal static LastDigitGpuTables EnsureStaticTables(Accelerator accelerator) => _sharedTables[accelerator];
[MethodImpl(MethodImplOptions.AggressiveInlining)]
internal static void WarmUp(Accelerator accelerator, AcceleratorStream stream) => _sharedTables[accelerator] = new LastDigitGpuTables(accelerator, stream);
internal LastDigitGpuTables(Accelerator accelerator, AcceleratorStream stream)
{
@@ -21,18 +24,14 @@ internal sealed class LastDigitGpuTables
var primesPow2LastSeven = DivisorGenerator.SmallPrimesPow2LastSeven;
var primesPow2LastThree = DivisorGenerator.SmallPrimesPow2LastThree;
var primesPow2LastNine = DivisorGenerator.SmallPrimesPow2LastNine;
// lock (accelerator)
{
DevicePrimesLastOne = accelerator.Allocate1D(stream, primesLastOne);
DevicePrimesLastSeven = accelerator.Allocate1D(stream, primesLastSeven);
DevicePrimesLastThree = accelerator.Allocate1D(stream, primesLastThree);
DevicePrimesLastNine = accelerator.Allocate1D(stream, primesLastNine);
DevicePrimesPow2LastOne = accelerator.Allocate1D(stream, primesPow2LastOne);
DevicePrimesPow2LastSeven = accelerator.Allocate1D(stream, primesPow2LastSeven);
DevicePrimesPow2LastThree = accelerator.Allocate1D(stream, primesPow2LastThree);
DevicePrimesPow2LastNine = accelerator.Allocate1D(stream, primesPow2LastNine);
}
DevicePrimesLastOne = accelerator.Allocate1D(stream, primesLastOne);
DevicePrimesLastSeven = accelerator.Allocate1D(stream, primesLastSeven);
DevicePrimesLastThree = accelerator.Allocate1D(stream, primesLastThree);
DevicePrimesLastNine = accelerator.Allocate1D(stream, primesLastNine);
DevicePrimesPow2LastOne = accelerator.Allocate1D(stream, primesPow2LastOne);
DevicePrimesPow2LastSeven = accelerator.Allocate1D(stream, primesPow2LastSeven);
DevicePrimesPow2LastThree = accelerator.Allocate1D(stream, primesPow2LastThree);
DevicePrimesPow2LastNine = accelerator.Allocate1D(stream, primesPow2LastNine);
}
internal readonly MemoryBuffer1D<uint, Stride1D.Dense> DevicePrimesLastOne;