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

PrimeOrderCalculatorAccelerator.Rent is now created outside of the loops and re-used for all calculations.

Browse Source
This commit is contained in:
mlemanczyk committed 2025-11-24 22:51:53 +01:00
1 parent 57a50e2a37
commit b9ca95bf8c
53 files changed
+4639 -4229
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

PerfectNumbers.Core/Gpu/Pow2MontgomeryGpuCalculator.cs
+1 -33
View File
@@ -8,12 +8,11 @@ namespace PerfectNumbers.Core.Gpu;
public static class Pow2MontgomeryGpuCalculator
{
[MethodImpl(MethodImplOptions.AggressiveInlining)]
public static ulong CalculateKeep(ulong exponent, in MontgomeryDivisorData divisor)
public static ulong CalculateKeep(PrimeOrderCalculatorAccelerator gpu, ulong exponent, in MontgomeryDivisorData divisor)
{
// GpuPrimeWorkLimiter.Acquire();
ulong result = 0UL;
var gpu = PrimeOrderCalculatorAccelerator.Rent(1);
int acceleratorIndex = gpu.AcceleratorIndex;
Accelerator? accelerator = gpu.Accelerator;
MemoryBuffer1D<ulong, Stride1D.Dense> exponentBuffer = gpu.Input;
@@ -32,37 +31,6 @@ public static class Pow2MontgomeryGpuCalculator
AcceleratorStreamPool.Return(acceleratorIndex, stream);
// GpuPrimeWorkLimiter.Release();
PrimeOrderCalculatorAccelerator.Return(gpu);
return result;
}
[MethodImpl(MethodImplOptions.AggressiveInlining)]
public static ulong CalculateConvert(ulong exponent, in MontgomeryDivisorData divisor)
{
// GpuPrimeWorkLimiter.Acquire();
ulong result = 0UL;
var gpu = PrimeOrderCalculatorAccelerator.Rent(1);
MemoryBuffer1D<ulong, Stride1D.Dense> exponentBuffer = gpu.Input;
MemoryBuffer1D<ulong, Stride1D.Dense> resultBuffer = gpu.OutputUlong;
int acceleratorIndex = gpu.AcceleratorIndex;
AcceleratorStream stream = AcceleratorStreamPool.Rent(acceleratorIndex);
exponentBuffer.View.CopyFromCPU(stream, ref exponent, 1);
// We don't need to worry about any left-overs here.
// resultBuffer.MemSetToZero(stream);
var kernelLauncher = gpu.ConvertToStandardKernelLauncher!;
kernelLauncher(stream, 1, exponentBuffer.View, divisor.Modulus, divisor.NPrime, divisor.MontgomeryOne, divisor.MontgomeryTwo, divisor.MontgomeryTwoSquared, resultBuffer.View);
resultBuffer.View.CopyToCPU(stream, ref result, 1);
stream.Synchronize();
AcceleratorStreamPool.Return(acceleratorIndex, stream);
// GpuPrimeWorkLimiter.Release();
PrimeOrderCalculatorAccelerator.Return(gpu);
return result;
}