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

Reintroduce MulHigh CPU helper and tighten GPU sieve loop

Browse Source
This commit is contained in:
Marek Lemańczyk committed 2025-10-26 23:41:14 +01:00
1 parent e2f68cedf2
commit 9a0e328995
10 files changed
+222 -344
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

PerfectNumbers.Core/Gpu/GpuKernelPool.cs
+2 -14
View File
@@ -46,7 +46,6 @@ public sealed class KernelContainer
public MemoryBuffer1D<int, Stride1D.Dense>? SmallPrimeFactorExponentSlots;
public MemoryBuffer1D<int, Stride1D.Dense>? SmallPrimeFactorCountSlot;
public MemoryBuffer1D<ulong, Stride1D.Dense>? SmallPrimeFactorRemainingSlot;
public MemoryBuffer1D<byte, Stride1D.Dense>? SmallPrimeFactorFullyFactoredSlot;
public MemoryBuffer1D<uint, Stride1D.Dense>? SmallPrimesLastOne;
public MemoryBuffer1D<ulong, Stride1D.Dense>? SmallPrimesPow2LastOne;
public MemoryBuffer1D<uint, Stride1D.Dense>? SmallPrimesLastSeven;
@@ -108,20 +107,17 @@ public readonly struct SmallPrimeFactorScratch
public readonly MemoryBuffer1D<int, Stride1D.Dense> ExponentSlots;
public readonly MemoryBuffer1D<int, Stride1D.Dense> CountSlot;
public readonly MemoryBuffer1D<ulong, Stride1D.Dense> RemainingSlot;
public readonly MemoryBuffer1D<byte, Stride1D.Dense> FullyFactoredSlot;
public SmallPrimeFactorScratch(
MemoryBuffer1D<ulong, Stride1D.Dense> primeSlots,
MemoryBuffer1D<int, Stride1D.Dense> exponentSlots,
MemoryBuffer1D<int, Stride1D.Dense> countSlot,
MemoryBuffer1D<ulong, Stride1D.Dense> remainingSlot,
MemoryBuffer1D<byte, Stride1D.Dense> fullyFactoredSlot)
MemoryBuffer1D<ulong, Stride1D.Dense> remainingSlot)
{
PrimeSlots = primeSlots;
ExponentSlots = exponentSlots;
CountSlot = countSlot;
RemainingSlot = remainingSlot;
FullyFactoredSlot = fullyFactoredSlot;
}
public void Clear()
@@ -130,7 +126,6 @@ public readonly struct SmallPrimeFactorScratch
ExponentSlots.MemSetToZero();
CountSlot.MemSetToZero();
RemainingSlot.MemSetToZero();
FullyFactoredSlot.MemSetToZero();
}
}
@@ -278,14 +273,7 @@ public class GpuKernelPool
kernels.SmallPrimeFactorRemainingSlot = remainingSlot;
}
MemoryBuffer1D<byte, Stride1D.Dense>? fullyFactoredSlot = kernels.SmallPrimeFactorFullyFactoredSlot;
if (fullyFactoredSlot is null)
{
fullyFactoredSlot = accelerator.Allocate1D<byte>(1);
kernels.SmallPrimeFactorFullyFactoredSlot = fullyFactoredSlot;
}
return new SmallPrimeFactorScratch(primeSlots!, exponentSlots!, countSlot!, remainingSlot!, fullyFactoredSlot!);
return new SmallPrimeFactorScratch(primeSlots!, exponentSlots!, countSlot!, remainingSlot!);
}
}