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

Revert "Merge pull request #24 from mlemanczyk/codex/modify-evenperfectbitscanner-for-residue-mode"

Browse Source 
This reverts commit 2d024cc708, reversing
changes made to b1c29ad16e.
This commit is contained in:
Marek Lemańczyk committed 2025-09-21 02:25:18 +02:00
1 parent deb25796a8
commit 8a6fc0c6a5
7 files changed
+265 -551
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

PerfectNumbers.Core/Gpu/MersenneNumberResidueGpuTester.cs
+50 -86
View File
@@ -12,36 +12,22 @@ public class MersenneNumberResidueGpuTester(bool useGpuOrder)
// via GpuKernelPool to avoid host round-trips.
// GPU residue variant: check 2^p % q == 1 for q = 2*p*k + 1.
public void Scan(
ulong exponent,
UInt128 twoP,
bool lastIsSeven,
UInt128 perSetLimit,
UInt128 setCount,
UInt128 overallLimit,
ref bool isPrime,
ref bool divisorsExhausted)
{
if (setCount == UInt128.Zero || perSetLimit == UInt128.Zero || overallLimit == UInt128.Zero)
{
return;
}
var gpuLease = GpuKernelPool.GetKernel(_useGpuOrder);
var execution = gpuLease.EnterExecutionScope();
public void Scan(ulong exponent, UInt128 twoP, bool lastIsSeven, UInt128 maxK, ref bool isPrime)
{
var gpuLease = GpuKernelPool.GetKernel(_useGpuOrder);
var accelerator = gpuLease.Accelerator;
var stream = gpuLease.Stream;
// Ensure device has small cycles and primes tables for in-kernel lookup
var smallCyclesView = GpuKernelPool.EnsureSmallCyclesOnDevice(accelerator);
ResiduePrimeViews primeViews = GpuKernelPool.EnsureSmallPrimesOnDevice(accelerator);
int batchSize = GpuConstants.ScanBatchSize;
byte last = lastIsSeven ? (byte)1 : (byte)0;
var kernel = gpuLease.Pow2ModKernel;
ulong step10 = ((exponent % 10UL) << 1) % 10UL;
ulong step8 = ((exponent & 7UL) << 1) & 7UL;
ulong step3 = ((exponent % 3UL) << 1) % 3UL;
ulong step5 = ((exponent % 5UL) << 1) % 5UL;
GpuUInt128 twoPGpu = (GpuUInt128)twoP;
// Ensure device has small cycles and primes tables for in-kernel lookup
var smallCyclesView = GpuKernelPool.EnsureSmallCyclesOnDevice(accelerator);
ResiduePrimeViews primeViews = GpuKernelPool.EnsureSmallPrimesOnDevice(accelerator);
int batchSize = GpuConstants.ScanBatchSize;
UInt128 kStart = 1UL;
byte last = lastIsSeven ? (byte)1 : (byte)0;
var kernel = gpuLease.Pow2ModKernel;
ulong step10 = ((exponent % 10UL) << 1) % 10UL;
ulong step8 = ((exponent & 7UL) << 1) & 7UL;
ulong step3 = ((exponent % 3UL) << 1) % 3UL;
ulong step5 = ((exponent % 5UL) << 1) % 5UL;
GpuUInt128 twoPGpu = (GpuUInt128)twoP;
var orderBuffer = accelerator.Allocate1D<ulong>(batchSize);
// Device memory returned by Allocate1D is not guaranteed to be zeroed.
@@ -52,79 +38,57 @@ public class MersenneNumberResidueGpuTester(bool useGpuOrder)
// testers such as the order kernels.
orderBuffer.MemSetToZero();
ulong[] orderArray = ArrayPool<ulong>.Shared.Rent(batchSize);
UInt128 batchSize128 = (UInt128)batchSize;
UInt128 limitInclusive = overallLimit == UInt128.MaxValue
? overallLimit
: checked(overallLimit + UInt128.One);
UInt128 remaining;
int currentSize;
int i;
UInt128 batchSize128 = (UInt128)batchSize, q;
Span<ulong> orders = orderArray.AsSpan(0, batchSize);
try
{
for (UInt128 setIndex = UInt128.Zero; setIndex < setCount && Volatile.Read(ref isPrime); setIndex++)
while (kStart < maxK && Volatile.Read(ref isPrime))
{
UInt128 setOffset = checked(perSetLimit * setIndex);
UInt128 setStart = checked(setOffset + UInt128.One);
if (setStart >= limitInclusive)
remaining = maxK - kStart;
if (remaining > batchSize128)
{
currentSize = batchSize;
}
else
{
currentSize = (int)remaining;
orders = orderArray.AsSpan(0, currentSize);
}
q = twoP * kStart + UInt128.One;
q.Mod10_8_5_3(out ulong q0m10, out ulong q0m8, out ulong q0m5, out ulong q0m3);
var ra = new ResidueAutomatonArgs(q0m10, step10, q0m8, step8, q0m3, step3, q0m5, step5);
kernel(currentSize, exponent, twoPGpu, (GpuUInt128)kStart, last, 0UL,
ra, orderBuffer.View, smallCyclesView, primeViews.LastOne, primeViews.LastSeven, primeViews.LastOnePow2, primeViews.LastSevenPow2);
accelerator.Synchronize();
orderBuffer.View.CopyToCPU(ref MemoryMarshal.GetReference(orders), currentSize);
if (!Volatile.Read(ref isPrime))
{
break;
}
UInt128 setLimitExclusive = checked(setOffset + perSetLimit);
setLimitExclusive = setLimitExclusive >= limitInclusive
? limitInclusive
: checked(setLimitExclusive + UInt128.One);
if (setLimitExclusive > limitInclusive)
for (i = 0; i < currentSize; i++)
{
setLimitExclusive = limitInclusive;
}
UInt128 kStart = setStart;
while (kStart < setLimitExclusive && Volatile.Read(ref isPrime))
{
UInt128 remaining = checked(setLimitExclusive - kStart);
int currentSize = remaining > batchSize128 ? batchSize : (int)remaining;
if (currentSize <= 0)
if (orders[i] != 0UL)
{
Volatile.Write(ref isPrime, false);
break;
}
orders = orderArray.AsSpan(0, currentSize);
UInt128 q = checked(twoP * kStart);
q = checked(q + UInt128.One);
q.Mod10_8_5_3(out ulong q0m10, out ulong q0m8, out ulong q0m5, out ulong q0m3);
var ra = new ResidueAutomatonArgs(q0m10, step10, q0m8, step8, q0m3, step3, q0m5, step5);
kernel(stream, currentSize, exponent, twoPGpu, (GpuUInt128)kStart, last, 0UL,
ra, orderBuffer.View, smallCyclesView, primeViews.LastOne, primeViews.LastSeven, primeViews.LastOnePow2, primeViews.LastSevenPow2);
stream.Synchronize();
orderBuffer.View.CopyToCPU(ref MemoryMarshal.GetReference(orders), currentSize);
stream.Synchronize();
if (!Volatile.Read(ref isPrime))
{
break;
}
for (int i = 0; i < currentSize; i++)
{
if (orders[i] != 0UL)
{
Volatile.Write(ref isPrime, false);
break;
}
}
kStart = checked(kStart + (UInt128)currentSize);
}
kStart += batchSize128;
}
}
finally
{
ArrayPool<ulong>.Shared.Return(orderArray);
orderBuffer.Dispose();
execution.Dispose();
gpuLease.Dispose();
}
divisorsExhausted = true;
}
orderBuffer.Dispose();
gpuLease.Dispose();
}
}
}