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

No. of divisor cycle iterations made configurable

Browse Source
This commit is contained in:
Marek Lemańczyk committed 2025-09-11 01:20:15 +02:00
1 parent c8c80f2f91
commit f17f83e326
3 files changed
+298 -290
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

EvenPerfectBitScanner/Program.cs
+253 -247
View File
@@ -26,13 +26,12 @@ internal static class Program
private static int _primeCount; private static int _primeCount;
private static bool _primeFoundAfterInit; private static bool _primeFoundAfterInit;
private static bool _useGcdFilter; private static bool _useGcdFilter;
private static bool _useResidue; private static bool _useResidue;
private static bool _useDivisor; private static bool _useDivisor;
private static ulong _divisor; private static ulong _divisor;
private static MersenneNumberDivisorGpuTester? _divisorTester; private static MersenneNumberDivisorGpuTester? _divisorTester;
private static (ulong divisor, uint cycle)[]? _divisorCandidates; private static (ulong divisor, uint cycle)[]? _divisorCandidates;
private const int ExtraDivisorCycleSearchLimit = 32; private static ulong? _orderWarmupLimitOverride;
private static ulong? _orderWarmupLimitOverride;
private static unsafe delegate*<ulong, ref ulong, ulong> _transformP; private static unsafe delegate*<ulong, ref ulong, ulong> _transformP;
private static long _state; private static long _state;
private static bool _limitReached; private static bool _limitReached;
@@ -62,11 +61,11 @@ private static ulong? _orderWarmupLimitOverride;
bool useOrder = false; bool useOrder = false;
bool showHelp = false; bool showHelp = false;
bool useBitTransform = false; bool useBitTransform = false;
bool useLucas = false; bool useLucas = false;
bool useResidue = false; // M_p test via residue divisors (replaces LL/incremental) bool useResidue = false; // M_p test via residue divisors (replaces LL/incremental)
bool useDivisor = false; // M_p divisibility by specific divisor bool useDivisor = false; // M_p divisibility by specific divisor
ulong divisor = 0UL; ulong divisor = 0UL;
// Device routing // Device routing
bool useGpuCycles = true; bool useGpuCycles = true;
bool mersenneOnGpu = true; // controls Lucas/incremental/pow2mod device bool mersenneOnGpu = true; // controls Lucas/incremental/pow2mod device
bool orderOnGpu = true; // controls order computations device bool orderOnGpu = true; // controls order computations device
@@ -76,6 +75,7 @@ private static ulong? _orderWarmupLimitOverride;
string cyclesPath = DefaultCyclesPath; string cyclesPath = DefaultCyclesPath;
int cyclesBatchSize = 512; int cyclesBatchSize = 512;
bool continueCyclesGeneration = false; bool continueCyclesGeneration = false;
int divisorCyclesSearchLimit = PerfectNumberConstants.ExtraDivisorCycleSearchLimit;
// NTT backend selection (GPU): reference vs staged // NTT backend selection (GPU): reference vs staged
for (int i = 0; i < args.Length; i++) for (int i = 0; i < args.Length; i++)
@@ -108,35 +108,40 @@ private static ulong? _orderWarmupLimitOverride;
{ {
int eq = arg.IndexOf('='); int eq = arg.IndexOf('=');
ReadOnlySpan<char> value = arg.AsSpan(eq + 1); ReadOnlySpan<char> value = arg.AsSpan(eq + 1);
if (value.Equals("pow2mod", StringComparison.OrdinalIgnoreCase)) if (value.Equals("pow2mod", StringComparison.OrdinalIgnoreCase))
{ {
kernelType = GpuKernelType.Pow2Mod; kernelType = GpuKernelType.Pow2Mod;
} }
else if (value.Equals("lucas", StringComparison.OrdinalIgnoreCase)) else if (value.Equals("lucas", StringComparison.OrdinalIgnoreCase))
{ {
useLucas = true; useLucas = true;
} }
else if (value.Equals("residue", StringComparison.OrdinalIgnoreCase)) else if (value.Equals("residue", StringComparison.OrdinalIgnoreCase))
{ {
useResidue = true; useResidue = true;
useLucas = false; useLucas = false;
} }
else if (value.Equals("divisor", StringComparison.OrdinalIgnoreCase)) else if (value.Equals("divisor", StringComparison.OrdinalIgnoreCase))
{ {
useDivisor = true; useDivisor = true;
useLucas = false; useLucas = false;
useResidue = false; useResidue = false;
} }
else else
{ {
kernelType = GpuKernelType.Incremental; kernelType = GpuKernelType.Incremental;
} }
} }
else if (arg.StartsWith("--divisor=", StringComparison.OrdinalIgnoreCase)) else if (arg.StartsWith("--divisor=", StringComparison.OrdinalIgnoreCase))
{ {
int eq = arg.IndexOf('='); int eq = arg.IndexOf('=');
divisor = ulong.Parse(arg.AsSpan(eq + 1)); divisor = ulong.Parse(arg.AsSpan(eq + 1));
} }
else if (arg.StartsWith("--divisor-cycles-limit=", StringComparison.OrdinalIgnoreCase))
{
int eq = arg.IndexOf('=');
divisorCyclesSearchLimit = int.Parse(arg.AsSpan(eq + 1));
}
else if (arg.StartsWith("--residue-max-k=", StringComparison.OrdinalIgnoreCase)) else if (arg.StartsWith("--residue-max-k=", StringComparison.OrdinalIgnoreCase))
{ {
int eq = arg.IndexOf('='); int eq = arg.IndexOf('=');
@@ -354,14 +359,14 @@ private static ulong? _orderWarmupLimitOverride;
} }
} }
if (showHelp) if (showHelp)
{ {
PrintHelp(); PrintHelp();
return; return;
} }
// Apply GPU prime sieve runtime configuration // Apply GPU prime sieve runtime configuration
GpuPrimeWorkLimiter.SetLimit(gpuPrimeThreads); GpuPrimeWorkLimiter.SetLimit(gpuPrimeThreads);
PerfectNumbers.Core.PrimeTester.GpuBatchSize = Math.Max(1, gpuPrimeBatch); PerfectNumbers.Core.PrimeTester.GpuBatchSize = Math.Max(1, gpuPrimeBatch);
MersenneDivisorCycles mersenneDivisorCycles = new(); MersenneDivisorCycles mersenneDivisorCycles = new();
@@ -405,47 +410,47 @@ private static ulong? _orderWarmupLimitOverride;
MersenneDivisorCycles.Shared.LoadFrom(cyclesPath); MersenneDivisorCycles.Shared.LoadFrom(cyclesPath);
} }
Console.WriteLine("Divisor cycles are ready"); Console.WriteLine("Divisor cycles are ready");
_useResidue = useResidue; _useResidue = useResidue;
_useDivisor = useDivisor; _useDivisor = useDivisor;
_divisor = divisor; _divisor = divisor;
_transformP = useBitTransform ? &TransformPBit : &TransformPAdd; _transformP = useBitTransform ? &TransformPBit : &TransformPAdd;
PResidue = new ThreadLocal<ModResidueTracker>(() => new ModResidueTracker(ResidueModel.Identity, initialNumber: currentP, initialized: true), trackAllValues: true); PResidue = new ThreadLocal<ModResidueTracker>(() => new ModResidueTracker(ResidueModel.Identity, initialNumber: currentP, initialized: true), trackAllValues: true);
PrimeTesters = new ThreadLocal<PrimeTester>(() => new PrimeTester(), trackAllValues: true); PrimeTesters = new ThreadLocal<PrimeTester>(() => new PrimeTester(), trackAllValues: true);
// Note: --primes-device controls default device for library kernels; p primality remains CPU here. // Note: --primes-device controls default device for library kernels; p primality remains CPU here.
// Initialize per-thread p residue tracker (Identity model) at currentP // Initialize per-thread p residue tracker (Identity model) at currentP
if (!useDivisor) if (!useDivisor)
{ {
MersenneTesters = new ThreadLocal<MersenneNumberTester>(() => MersenneTesters = new ThreadLocal<MersenneNumberTester>(() =>
{ {
var tester = new MersenneNumberTester( var tester = new MersenneNumberTester(
useIncremental: !useLucas, useIncremental: !useLucas,
useOrderCache: false, useOrderCache: false,
kernelType: kernelType, kernelType: kernelType,
useModuloWorkaround: useModuloWorkaround, useModuloWorkaround: useModuloWorkaround,
useOrder: useOrder, useOrder: useOrder,
useGpuLucas: mersenneOnGpu, useGpuLucas: mersenneOnGpu,
useGpuScan: mersenneOnGpu, useGpuScan: mersenneOnGpu,
useGpuOrder: orderOnGpu, useGpuOrder: orderOnGpu,
useResidue: useResidue, useResidue: useResidue,
maxK: residueKMax); maxK: residueKMax);
if (!useLucas) if (!useLucas)
{ {
tester.WarmUpOrders(currentP, _orderWarmupLimitOverride ?? 5_000_000UL); tester.WarmUpOrders(currentP, _orderWarmupLimitOverride ?? 5_000_000UL);
} }
return tester; return tester;
}, trackAllValues: true); }, trackAllValues: true);
} }
else else
{ {
_divisorTester = new MersenneNumberDivisorGpuTester(); _divisorTester = new MersenneNumberDivisorGpuTester();
if (divisor == 0UL) if (divisor == 0UL)
{ {
_divisorCandidates = BuildDivisorCandidates(); _divisorCandidates = BuildDivisorCandidates();
} }
} }
// Load RLE blacklist (optional) // Load RLE blacklist (optional)
if (!string.IsNullOrEmpty(_rleBlacklistPath)) if (!string.IsNullOrEmpty(_rleBlacklistPath))
@@ -471,26 +476,26 @@ private static ulong? _orderWarmupLimitOverride;
Console.WriteLine("Initialization..."); Console.WriteLine("Initialization...");
// Compose a results file name that encodes configuration (before opening file) // Compose a results file name that encodes configuration (before opening file)
var builtName = BuildResultsFileName( var builtName = BuildResultsFileName(
useBitTransform, useBitTransform,
threadCount, threadCount,
blockSize, blockSize,
kernelType, kernelType,
useLucas, useLucas,
useDivisor, useDivisor,
mersenneOnGpu, mersenneOnGpu,
useOrder, useOrder,
useModuloWorkaround, useModuloWorkaround,
_useGcdFilter, _useGcdFilter,
NttGpuMath.GpuTransformBackend, NttGpuMath.GpuTransformBackend,
gpuPrimeThreads, gpuPrimeThreads,
sliceSize, sliceSize,
scanBatchSize, scanBatchSize,
_orderWarmupLimitOverride ?? 5_000_000UL, _orderWarmupLimitOverride ?? 5_000_000UL,
NttGpuMath.ReductionMode, NttGpuMath.ReductionMode,
mersenneOnGpu ? "gpu" : "cpu", mersenneOnGpu ? "gpu" : "cpu",
(GpuContextPool.ForceCpu ? "cpu" : "gpu"), (GpuContextPool.ForceCpu ? "cpu" : "gpu"),
orderOnGpu ? "gpu" : "cpu"); orderOnGpu ? "gpu" : "cpu");
if (!string.IsNullOrEmpty(_resultsPrefix)) if (!string.IsNullOrEmpty(_resultsPrefix))
{ {
@@ -550,38 +555,38 @@ private static ulong? _orderWarmupLimitOverride;
PerfectNumbers.Core.PrimeTester.GpuBatchSize = gpuPrimeBatch; PerfectNumbers.Core.PrimeTester.GpuBatchSize = gpuPrimeBatch;
// Compose a results file name that encodes configuration // Compose a results file name that encodes configuration
ResultsFileName = BuildResultsFileName( ResultsFileName = BuildResultsFileName(
useBitTransform, useBitTransform,
threadCount, threadCount,
blockSize, blockSize,
kernelType, kernelType,
useLucas, useLucas,
useDivisor, useDivisor,
mersenneOnGpu, mersenneOnGpu,
useOrder, useOrder,
useModuloWorkaround, useModuloWorkaround,
_useGcdFilter, _useGcdFilter,
NttGpuMath.GpuTransformBackend, NttGpuMath.GpuTransformBackend,
gpuPrimeThreads, gpuPrimeThreads,
sliceSize, sliceSize,
scanBatchSize, scanBatchSize,
_orderWarmupLimitOverride ?? 5_000_000UL, _orderWarmupLimitOverride ?? 5_000_000UL,
NttGpuMath.ReductionMode, NttGpuMath.ReductionMode,
mersenneOnGpu ? "gpu" : "cpu", mersenneOnGpu ? "gpu" : "cpu",
GpuContextPool.ForceCpu ? "cpu" : "gpu", GpuContextPool.ForceCpu ? "cpu" : "gpu",
orderOnGpu ? "gpu" : "cpu"); orderOnGpu ? "gpu" : "cpu");
if (!useDivisor) if (!useDivisor)
{ {
Console.WriteLine("Warming up orders..."); Console.WriteLine("Warming up orders...");
threadCount = Math.Max(1, threadCount); threadCount = Math.Max(1, threadCount);
_ = MersenneTesters.Value; _ = MersenneTesters.Value;
if (!useLucas) if (!useLucas)
{ {
MersenneTesters.Value!.WarmUpOrders(currentP, _orderWarmupLimitOverride ?? 5_000_000UL); MersenneTesters.Value!.WarmUpOrders(currentP, _orderWarmupLimitOverride ?? 5_000_000UL);
} }
} }
Console.WriteLine("Starting scan..."); Console.WriteLine("Starting scan...");
_state = ((long)currentP << 3) | (long)remainder; _state = ((long)currentP << 3) | (long)remainder;
@@ -690,12 +695,12 @@ private static ulong? _orderWarmupLimitOverride;
Console.WriteLine(" --increment=bit|add exponent increment method"); Console.WriteLine(" --increment=bit|add exponent increment method");
Console.WriteLine(" --threads=<value> number of worker threads"); Console.WriteLine(" --threads=<value> number of worker threads");
Console.WriteLine(" --block-size=<value> values processed per thread batch"); Console.WriteLine(" --block-size=<value> values processed per thread batch");
Console.WriteLine(" --mersenne=pow2mod|incremental|lucas|residue|divisor Mersenne test method"); Console.WriteLine(" --mersenne=pow2mod|incremental|lucas|residue|divisor Mersenne test method");
Console.WriteLine(" --divisor=<value> optional divisor for --mersenne=divisor mode"); Console.WriteLine(" --divisor=<value> optional divisor for --mersenne=divisor mode");
Console.WriteLine(" --residue-max-k=<value> max k for residue Mersenne test (q = 2*p*k + 1)"); Console.WriteLine(" --residue-max-k=<value> max k for residue Mersenne test (q = 2*p*k + 1)");
Console.WriteLine(" --mersenne-device=cpu|gpu Device for Mersenne method (default gpu)"); Console.WriteLine(" --mersenne-device=cpu|gpu Device for Mersenne method (default gpu)");
Console.WriteLine(" --primes-device=cpu|gpu Device for prime-scan kernels (default gpu)"); Console.WriteLine(" --primes-device=cpu|gpu Device for prime-scan kernels (default gpu)");
Console.WriteLine(" --gpu-prime-batch=<n> Batch size for GPU primality sieve (default 262144)"); Console.WriteLine(" --gpu-prime-batch=<n> Batch size for GPU primality sieve (default 262144)");
Console.WriteLine(" --order-device=cpu|gpu Device for order computations (default gpu)"); Console.WriteLine(" --order-device=cpu|gpu Device for order computations (default gpu)");
Console.WriteLine(" --ntt=reference|staged GPU NTT backend (default staged)"); Console.WriteLine(" --ntt=reference|staged GPU NTT backend (default staged)");
Console.WriteLine(" --mod-reduction=auto|uint128|mont64|barrett128 staged NTT reduction (default auto)"); Console.WriteLine(" --mod-reduction=auto|uint128|mont64|barrett128 staged NTT reduction (default auto)");
@@ -708,12 +713,13 @@ private static ulong? _orderWarmupLimitOverride;
Console.WriteLine(" --rle-only-last7=true|false apply RLE only when p % 10 == 7 (default true)"); Console.WriteLine(" --rle-only-last7=true|false apply RLE only when p % 10 == 7 (default true)");
Console.WriteLine(" --zero-hard=<f> hard reject if zero_fraction(p) > f (default off)"); Console.WriteLine(" --zero-hard=<f> hard reject if zero_fraction(p) > f (default off)");
Console.WriteLine(" --zero-conj=<f>:<r> hard reject if zero_fraction(p) > f AND max_zero_block >= r (default off)"); Console.WriteLine(" --zero-conj=<f>:<r> hard reject if zero_fraction(p) > f AND max_zero_block >= r (default off)");
Console.WriteLine(" --results-dir=<path> directory for results file"); Console.WriteLine(" --results-dir=<path> directory for results file");
Console.WriteLine(" --results-prefix=<text> prefix to prepend to results filename"); Console.WriteLine(" --results-prefix=<text> prefix to prepend to results filename");
Console.WriteLine(" --divisor-cycles=<path> divisor cycles data file"); Console.WriteLine(" --divisor-cycles=<path> divisor cycles data file");
Console.WriteLine(" --divisor-cycles-device=cpu|gpu device for cycles generation (default gpu)"); Console.WriteLine(" --divisor-cycles-device=cpu|gpu device for cycles generation (default gpu)");
Console.WriteLine(" --divisor-cycles-batch-size=<value> batch size for cycles generation (default 512)"); Console.WriteLine(" --divisor-cycles-batch-size=<value> batch size for cycles generation (default 512)");
Console.WriteLine(" --divisor-cycles-continue continue divisor cycles generation"); Console.WriteLine(" --divisor-cycles-continue continue divisor cycles generation");
Console.WriteLine(" --divisor-cycles-limit=<value> cycle search iterations when --mersenne=divisor");
Console.WriteLine(" --use-order test primality via q order"); Console.WriteLine(" --use-order test primality via q order");
Console.WriteLine(" --workaround-mod avoid '%' operator on the GPU"); Console.WriteLine(" --workaround-mod avoid '%' operator on the GPU");
@@ -724,10 +730,10 @@ private static ulong? _orderWarmupLimitOverride;
Console.WriteLine(" --help, -help, --?, -?, /? show this help message"); Console.WriteLine(" --help, -help, --?, -?, /? show this help message");
} }
private static string BuildResultsFileName(bool bitInc, int threads, int block, GpuKernelType kernelType, bool useLucasFlag, bool useDivisorFlag, bool mersenneOnGpu, bool useOrder, bool useModWorkaround, bool useGcd, NttBackend nttBackend, int gpuPrimeThreads, int llSlice, int gpuScanBatch, ulong warmupLimit, ModReductionMode reduction, string mersenneDevice, string primesDevice, string orderDevice) private static string BuildResultsFileName(bool bitInc, int threads, int block, GpuKernelType kernelType, bool useLucasFlag, bool useDivisorFlag, bool mersenneOnGpu, bool useOrder, bool useModWorkaround, bool useGcd, NttBackend nttBackend, int gpuPrimeThreads, int llSlice, int gpuScanBatch, ulong warmupLimit, ModReductionMode reduction, string mersenneDevice, string primesDevice, string orderDevice)
{ {
string inc = bitInc ? "bit" : "add"; string inc = bitInc ? "bit" : "add";
string mers = useDivisorFlag ? "divisor" : (useLucasFlag ? "lucas" : (kernelType == GpuKernelType.Pow2Mod ? "pow2mod" : "incremental")); string mers = useDivisorFlag ? "divisor" : (useLucasFlag ? "lucas" : (kernelType == GpuKernelType.Pow2Mod ? "pow2mod" : "incremental"));
string ntt = nttBackend == NttBackend.Staged ? "staged" : "reference"; string ntt = nttBackend == NttBackend.Staged ? "staged" : "reference";
string red = reduction switch { ModReductionMode.Mont64 => "mont64", ModReductionMode.Barrett128 => "barrett128", ModReductionMode.GpuUInt128 => "uint128", _ => "auto" }; string red = reduction switch { ModReductionMode.Mont64 => "mont64", ModReductionMode.Barrett128 => "barrett128", ModReductionMode.GpuUInt128 => "uint128", _ => "auto" };
string order = useOrder ? "order-on" : "order-off"; string order = useOrder ? "order-on" : "order-off";
@@ -876,9 +882,9 @@ private static ulong? _orderWarmupLimitOverride;
} }
[MethodImpl(MethodImplOptions.AggressiveInlining)] [MethodImpl(MethodImplOptions.AggressiveInlining)]
internal static ulong TransformPAdd(ulong value, ref ulong remainder) internal static ulong TransformPAdd(ulong value, ref ulong remainder)
{ {
ulong next = value; ulong next = value;
value = remainder switch value = remainder switch
{ {
0UL => 1UL, 0UL => 1UL,
@@ -898,49 +904,49 @@ private static ulong? _orderWarmupLimitOverride;
remainder += value; remainder += value;
remainder -= (remainder >= 6UL) ? 6UL : 0UL; remainder -= (remainder >= 6UL) ? 6UL : 0UL;
return next + value; return next + value;
} }
private static (ulong divisor, uint cycle)[] BuildDivisorCandidates() private static (ulong divisor, uint cycle)[] BuildDivisorCandidates()
{ {
uint[] snapshot = MersenneDivisorCycles.Shared.ExportSmallCyclesSnapshot(); uint[] snapshot = MersenneDivisorCycles.Shared.ExportSmallCyclesSnapshot();
List<(ulong divisor, uint cycle)> list = new(snapshot.Length / 2); List<(ulong divisor, uint cycle)> list = new(snapshot.Length / 2);
ulong d; ulong d;
uint cycle; uint cycle;
for (int i = 3; i < snapshot.Length; i += 2) for (int i = 3; i < snapshot.Length; i += 2)
{ {
cycle = snapshot[i]; cycle = snapshot[i];
if (cycle == 0U) if (cycle == 0U)
{ {
continue; continue;
} }
d = (ulong)i; d = (ulong)i;
list.Add((d, cycle)); list.Add((d, cycle));
} }
return list.ToArray(); return list.ToArray();
} }
private static bool IsDivisible(ulong exponent, ulong divisor) private static bool IsDivisible(ulong exponent, ulong divisor)
{ {
if (_divisorTester is not null) if (_divisorTester is not null)
{ {
return _divisorTester.IsDivisible(exponent, divisor); return _divisorTester.IsDivisible(exponent, divisor);
} }
ulong remainder = 1UL; ulong remainder = 1UL;
for (ulong i = 0; i < exponent; i++) for (ulong i = 0; i < exponent; i++)
{ {
remainder <<= 1; remainder <<= 1;
if (remainder >= divisor) if (remainder >= divisor)
{ {
remainder -= divisor; remainder -= divisor;
} }
} }
return remainder == 1UL; return remainder == 1UL;
} }
[MethodImpl(MethodImplOptions.AggressiveInlining)] [MethodImpl(MethodImplOptions.AggressiveInlining)]
internal static bool IsEvenPerfectCandidate(ulong p) internal static bool IsEvenPerfectCandidate(ulong p)
@@ -948,7 +954,7 @@ private static ulong? _orderWarmupLimitOverride;
return IsEvenPerfectCandidate(p, out _, out _); return IsEvenPerfectCandidate(p, out _, out _);
} }
internal static bool IsEvenPerfectCandidate(ulong p, out bool searchedMersenne, out bool detailedCheck) internal static bool IsEvenPerfectCandidate(ulong p, int divisorCyclesSearchLimit, out bool searchedMersenne, out bool detailedCheck)
{ {
searchedMersenne = false; searchedMersenne = false;
detailedCheck = false; detailedCheck = false;
@@ -1011,67 +1017,67 @@ private static ulong? _orderWarmupLimitOverride;
return false; return false;
} }
searchedMersenne = true; searchedMersenne = true;
if (_useDivisor) if (_useDivisor)
{ {
if (_divisor != 0UL) if (_divisor != 0UL)
{ {
if (IsDivisible(p, _divisor)) if (IsDivisible(p, _divisor))
{ {
detailedCheck = true; detailedCheck = true;
return false; return false;
} }
detailedCheck = false; detailedCheck = false;
return true; return true;
} }
var candidates = _divisorCandidates!; var candidates = _divisorCandidates!;
int len = candidates.Length; int len = candidates.Length;
for (int i = 0; i < len; i++) for (int i = 0; i < len; i++)
{ {
var (d, cycle) = candidates[i]; var (d, cycle) = candidates[i];
if (p % cycle != 0UL) if (p % cycle != 0UL)
{ {
continue; continue;
} }
if (IsDivisible(p, d)) if (IsDivisible(p, d))
{ {
detailedCheck = true; detailedCheck = true;
return false; return false;
} }
} }
for (int k = 1; k <= ExtraDivisorCycleSearchLimit; k++) for (int k = 1; k <= divisorCyclesSearchLimit; k++)
{ {
ulong kMul2 = 2UL * (ulong)k; ulong kMul2 = 2UL * (ulong)k;
if (p > ulong.MaxValue / kMul2) if (p > ulong.MaxValue / kMul2)
{ {
break; break;
} }
ulong d = kMul2 * p + 1UL; ulong d = kMul2 * p + 1UL;
ulong cycle = MersenneDivisorCycles.Shared.GetCycle(d); ulong cycle = MersenneDivisorCycles.Shared.GetCycle(d);
if (p % cycle != 0UL) if (p % cycle != 0UL)
{ {
continue; continue;
} }
if (IsDivisible(p, d)) if (IsDivisible(p, d))
{ {
detailedCheck = true; detailedCheck = true;
return false; return false;
} }
} }
detailedCheck = false; detailedCheck = false;
return true; return true;
} }
detailedCheck = MersenneTesters.Value!.IsMersennePrime(p); detailedCheck = MersenneTesters.Value!.IsMersennePrime(p);
return detailedCheck; return detailedCheck;
} }
// Use ModResidueTracker with a small set of primes to prefilter composite p. // Use ModResidueTracker with a small set of primes to prefilter composite p.
private static bool IsCompositeByResidues(ulong p) private static bool IsCompositeByResidues(ulong p)
PerfectNumbers.Core/Gpu/MersenneNumberDivisorGpuTester.cs
+42 -42
View File
@@ -7,51 +7,51 @@ namespace PerfectNumbers.Core.Gpu;
public sealed class MersenneNumberDivisorGpuTester public sealed class MersenneNumberDivisorGpuTester
{ {
private readonly ConcurrentDictionary<Accelerator, Action<Index1D, ulong, ulong, ArrayView<byte>>> _kernelCache = new(); private readonly ConcurrentDictionary<Accelerator, Action<Index1D, ulong, ulong, ArrayView<byte>>> _kernelCache = new();
private Action<Index1D, ulong, ulong, ArrayView<byte>> GetKernel(Accelerator accelerator) => private Action<Index1D, ulong, ulong, ArrayView<byte>> GetKernel(Accelerator accelerator) =>
_kernelCache.GetOrAdd(accelerator, acc => acc.LoadAutoGroupedStreamKernel<Index1D, ulong, ulong, ArrayView<byte>>(Kernel)); _kernelCache.GetOrAdd(accelerator, acc => acc.LoadAutoGroupedStreamKernel<Index1D, ulong, ulong, ArrayView<byte>>(Kernel));
public bool IsDivisible(ulong exponent, ulong divisor) public bool IsDivisible(ulong exponent, ulong divisor)
{ {
var gpu = GpuContextPool.RentPreferred(preferCpu: false); var gpu = GpuContextPool.RentPreferred(preferCpu: false);
var accelerator = gpu.Accelerator; var accelerator = gpu.Accelerator;
var kernel = GetKernel(accelerator); var kernel = GetKernel(accelerator);
using var resultBuffer = accelerator.Allocate1D<byte>(1); using var resultBuffer = accelerator.Allocate1D<byte>(1);
kernel(1, exponent, divisor, resultBuffer.View); kernel(1, exponent, divisor, resultBuffer.View);
accelerator.Synchronize(); accelerator.Synchronize();
bool divisible = resultBuffer.GetAsArray1D()[0] != 0; bool divisible = resultBuffer.GetAsArray1D()[0] != 0;
gpu.Dispose(); gpu.Dispose();
return divisible; return divisible;
} }
private static void Kernel(Index1D _, ulong exponent, ulong divisor, ArrayView<byte> result) private static void Kernel(Index1D _, ulong exponent, ulong divisor, ArrayView<byte> result)
{ {
if (divisor <= 1UL) if (divisor <= 1UL)
{ {
result[0] = 0; result[0] = 0;
return; return;
} }
int x = 63 - XMath.LeadingZeroCount(divisor); int x = 63 - XMath.LeadingZeroCount(divisor);
if (x == 0) if (x == 0)
{ {
var mod = new GpuUInt128(0UL, divisor); var mod = new GpuUInt128(0UL, divisor);
var temp = GpuUInt128.Pow2Mod(exponent, mod); var temp = GpuUInt128.Pow2Mod(exponent, mod);
result[0] = temp.High == 0UL && temp.Low == 1UL ? (byte)1 : (byte)0; result[0] = temp.High == 0UL && temp.Low == 1UL ? (byte)1 : (byte)0;
return; return;
} }
ulong ux = (ulong)x; ulong ux = (ulong)x;
ulong q = exponent / ux; ulong q = exponent / ux;
ulong r = exponent % ux; ulong r = exponent % ux;
ulong pow2x = 1UL << x; ulong pow2x = 1UL << x;
ulong y = divisor - pow2x; ulong y = divisor - pow2x;
var modVal = new GpuUInt128(0UL, divisor); var modVal = new GpuUInt128(0UL, divisor);
var baseVal = new GpuUInt128(0UL, pow2x); // 2^x ≡ -y (mod divisor) var baseVal = new GpuUInt128(0UL, pow2x); // 2^x ≡ -y (mod divisor)
var part1 = baseVal.ModPow(q, modVal); var part1 = baseVal.ModPow(q, modVal);
var part2 = GpuUInt128.Pow2Mod(r, modVal); var part2 = GpuUInt128.Pow2Mod(r, modVal);
var rem = part1.MulMod(part2, modVal); var rem = part1.MulMod(part2, modVal);
result[0] = rem.High == 0UL && rem.Low == 1UL ? (byte)1 : (byte)0; result[0] = rem.High == 0UL && rem.Low == 1UL ? (byte)1 : (byte)0;
} }
} }
PerfectNumbers.Core/PerfectNumberConstants.cs
+3 -1
View File
@@ -4,5 +4,7 @@ public static class PerfectNumberConstants
{ {
public const ulong BiggestKnownEvenPerfectP = 136279841UL; public const ulong BiggestKnownEvenPerfectP = 136279841UL;
public static readonly uint PrimesLimit = 1_000_000; //(ulong)Array.MaxLength;// 1_000_000; public static readonly uint PrimesLimit = 1_000_000; //(ulong)Array.MaxLength;// 1_000_000;
public const int MaxQForDivisorCycles = 4_000_000; public const int MaxQForDivisorCycles = 4_000_000;
public const int ExtraDivisorCycleSearchLimit = 64;
} }