Tiny cleanup
This commit is contained in:
Marek Lemańczyk committed
1 parent
2c2bc62cb2
commit
860d64baba
1 file changed
+121
-121
@@ -9,11 +9,11 @@ namespace PerfectNumbers.Core.Gpu;
|
|||||||
|
|
||||||
public sealed class MersenneNumberDivisorGpuTester
|
public sealed class MersenneNumberDivisorGpuTester
|
||||||
{
|
{
|
||||||
private readonly ConcurrentDictionary<Accelerator, Action<Index1D, ulong, GpuUInt128, ArrayView<byte>>> _kernelCache = new();
|
private readonly ConcurrentDictionary<Accelerator, Action<Index1D, ulong, GpuUInt128, ArrayView<byte>>> _kernelCache = new();
|
||||||
private readonly ConcurrentDictionary<Accelerator, MemoryBuffer1D<byte, Stride1D.Dense>> _resultBuffers = new();
|
private readonly ConcurrentDictionary<Accelerator, MemoryBuffer1D<byte, Stride1D.Dense>> _resultBuffers = new();
|
||||||
|
|
||||||
private Action<Index1D, ulong, GpuUInt128, ArrayView<byte>> GetKernel(Accelerator accelerator) =>
|
private Action<Index1D, ulong, GpuUInt128, ArrayView<byte>> GetKernel(Accelerator accelerator) =>
|
||||||
_kernelCache.GetOrAdd(accelerator, acc => acc.LoadAutoGroupedStreamKernel<Index1D, ulong, GpuUInt128, ArrayView<byte>>(Kernel));
|
_kernelCache.GetOrAdd(accelerator, acc => acc.LoadAutoGroupedStreamKernel<Index1D, ulong, GpuUInt128, ArrayView<byte>>(Kernel));
|
||||||
|
|
||||||
public static void BuildDivisorCandidates()
|
public static void BuildDivisorCandidates()
|
||||||
{
|
{
|
||||||
@@ -32,137 +32,137 @@ public sealed class MersenneNumberDivisorGpuTester
|
|||||||
list[count++] = ((ulong)i, cycle);
|
list[count++] = ((ulong)i, cycle);
|
||||||
}
|
}
|
||||||
|
|
||||||
_divisorCandidates = count == 0 ? Array.Empty<(ulong divisor, uint cycle)>() : list[..count].ToArray();
|
_divisorCandidates = count == 0 ? [] : list[..count];
|
||||||
}
|
}
|
||||||
|
|
||||||
public bool IsDivisible(ulong exponent, UInt128 divisor)
|
public bool IsDivisible(ulong exponent, UInt128 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);
|
||||||
var resultBuffer = _resultBuffers.GetOrAdd(accelerator, acc => acc.Allocate1D<byte>(1));
|
var resultBuffer = _resultBuffers.GetOrAdd(accelerator, acc => acc.Allocate1D<byte>(1));
|
||||||
resultBuffer.MemSetToZero();
|
resultBuffer.MemSetToZero();
|
||||||
kernel(1, exponent, (GpuUInt128)divisor, resultBuffer.View);
|
kernel(1, exponent, (GpuUInt128)divisor, resultBuffer.View);
|
||||||
accelerator.Synchronize();
|
accelerator.Synchronize();
|
||||||
Span<byte> result = stackalloc byte[1];
|
Span<byte> result = stackalloc byte[1];
|
||||||
resultBuffer.View.CopyToCPU(ref result[0], 1);
|
resultBuffer.View.CopyToCPU(ref result[0], 1);
|
||||||
bool divisible = result[0] != 0;
|
bool divisible = result[0] != 0;
|
||||||
result[0] = 0;
|
result[0] = 0;
|
||||||
resultBuffer.View.CopyFromCPU(ref result[0], 1);
|
resultBuffer.View.CopyFromCPU(ref result[0], 1);
|
||||||
accelerator.Synchronize();
|
accelerator.Synchronize();
|
||||||
gpu.Dispose();
|
gpu.Dispose();
|
||||||
return divisible;
|
return divisible;
|
||||||
}
|
}
|
||||||
|
|
||||||
private static (ulong divisor, uint cycle)[]? _divisorCandidates = Array.Empty<(ulong divisor, uint cycle)>();
|
private static (ulong divisor, uint cycle)[]? _divisorCandidates = Array.Empty<(ulong divisor, uint cycle)>();
|
||||||
|
|
||||||
public bool IsPrime(ulong p, UInt128 d, ulong divisorCyclesSearchLimit, out bool divisorsExhausted)
|
public bool IsPrime(ulong p, UInt128 d, ulong divisorCyclesSearchLimit, out bool divisorsExhausted)
|
||||||
{
|
{
|
||||||
if (d != UInt128.Zero)
|
if (d != UInt128.Zero)
|
||||||
{
|
{
|
||||||
if (IsDivisible(p, d))
|
if (IsDivisible(p, d))
|
||||||
{
|
{
|
||||||
divisorsExhausted = true;
|
divisorsExhausted = true;
|
||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
|
|
||||||
divisorsExhausted = false;
|
divisorsExhausted = false;
|
||||||
return true;
|
return true;
|
||||||
}
|
}
|
||||||
|
|
||||||
if (_divisorCandidates is { Length: > 0 } candidates)
|
if (_divisorCandidates is { Length: > 0 } candidates)
|
||||||
{
|
{
|
||||||
int len = candidates.Length;
|
int len = candidates.Length;
|
||||||
uint cycle;
|
uint cycle;
|
||||||
for (int k = 0; k < len; k++)
|
for (int k = 0; k < len; k++)
|
||||||
{
|
{
|
||||||
(ulong dSmall, cycle) = candidates[k];
|
(ulong dSmall, cycle) = candidates[k];
|
||||||
if (p % cycle != 0UL)
|
if (p % cycle != 0UL)
|
||||||
{
|
{
|
||||||
continue;
|
continue;
|
||||||
}
|
}
|
||||||
|
|
||||||
if (IsDivisible(p, dSmall))
|
if (IsDivisible(p, dSmall))
|
||||||
{
|
{
|
||||||
divisorsExhausted = true;
|
divisorsExhausted = true;
|
||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
UInt128 kMul2;
|
UInt128 kMul2;
|
||||||
var divisorCycles = MersenneDivisorCycles.Shared;
|
var divisorCycles = MersenneDivisorCycles.Shared;
|
||||||
UInt128 maxK2 = UInt128.MaxValue / ((UInt128)p << 1);
|
UInt128 maxK2 = UInt128.MaxValue / ((UInt128)p << 1);
|
||||||
ulong limit = divisorCyclesSearchLimit;
|
ulong limit = divisorCyclesSearchLimit;
|
||||||
if ((UInt128)limit > maxK2)
|
if ((UInt128)limit > maxK2)
|
||||||
{
|
{
|
||||||
limit = (ulong)maxK2;
|
limit = (ulong)maxK2;
|
||||||
}
|
}
|
||||||
|
|
||||||
ulong k2;
|
ulong k2;
|
||||||
for (k2 = 1UL; k2 <= limit; k2++)
|
for (k2 = 1UL; k2 <= limit; k2++)
|
||||||
{
|
{
|
||||||
kMul2 = (UInt128)k2 << 1;
|
kMul2 = (UInt128)k2 << 1;
|
||||||
UInt128 candidate = checked(kMul2 * p);
|
UInt128 candidate = checked(kMul2 * p);
|
||||||
d = checked(candidate + UInt128.One);
|
d = checked(candidate + UInt128.One);
|
||||||
if (p < 64UL && d == ((UInt128)1 << (int)p) - UInt128.One)
|
if (p < 64UL && d == ((UInt128)1 << (int)p) - UInt128.One)
|
||||||
{
|
{
|
||||||
continue;
|
continue;
|
||||||
}
|
}
|
||||||
|
|
||||||
UInt128 cycle128 = MersenneDivisorCycles.GetCycle(d);
|
UInt128 cycle128 = MersenneDivisorCycles.GetCycle(d);
|
||||||
if ((UInt128)p % cycle128 != UInt128.Zero)
|
if ((UInt128)p % cycle128 != UInt128.Zero)
|
||||||
{
|
{
|
||||||
continue;
|
continue;
|
||||||
}
|
}
|
||||||
|
|
||||||
if (IsDivisible(p, d))
|
if (IsDivisible(p, d))
|
||||||
{
|
{
|
||||||
divisorsExhausted = true;
|
divisorsExhausted = true;
|
||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
divisorsExhausted = (UInt128)divisorCyclesSearchLimit >= maxK2;
|
divisorsExhausted = (UInt128)divisorCyclesSearchLimit >= maxK2;
|
||||||
return true;
|
return true;
|
||||||
}
|
}
|
||||||
|
|
||||||
private static void Kernel(Index1D _, ulong exponent, GpuUInt128 divisor, ArrayView<byte> result)
|
private static void Kernel(Index1D _, ulong exponent, GpuUInt128 divisor, ArrayView<byte> result)
|
||||||
{
|
{
|
||||||
if (divisor.High == 0UL && divisor.Low <= 1UL)
|
if (divisor.High == 0UL && divisor.Low <= 1UL)
|
||||||
{
|
{
|
||||||
result[0] = 0;
|
result[0] = 0;
|
||||||
return;
|
return;
|
||||||
}
|
}
|
||||||
|
|
||||||
GpuUInt128 baseVal, mod;
|
GpuUInt128 baseVal, mod;
|
||||||
int x = divisor.High == 0UL
|
int x = divisor.High == 0UL
|
||||||
? 63 - XMath.LeadingZeroCount(divisor.Low)
|
? 63 - XMath.LeadingZeroCount(divisor.Low)
|
||||||
: 127 - XMath.LeadingZeroCount(divisor.High);
|
: 127 - XMath.LeadingZeroCount(divisor.High);
|
||||||
mod = divisor;
|
mod = divisor;
|
||||||
if (x == 0)
|
if (x == 0)
|
||||||
{
|
{
|
||||||
baseVal = GpuUInt128.Pow2Mod(exponent, mod);
|
baseVal = GpuUInt128.Pow2Mod(exponent, mod);
|
||||||
result[0] = baseVal.High == 0UL && baseVal.Low == 1UL ? (byte)1 : (byte)0;
|
result[0] = baseVal.High == 0UL && baseVal.Low == 1UL ? (byte)1 : (byte)0;
|
||||||
return;
|
return;
|
||||||
}
|
}
|
||||||
|
|
||||||
ulong ux = (ulong)x;
|
ulong ux = (ulong)x;
|
||||||
if (x >= 64)
|
if (x >= 64)
|
||||||
{
|
{
|
||||||
baseVal = new GpuUInt128(1UL << (x - 64), 0UL);
|
baseVal = new GpuUInt128(1UL << (x - 64), 0UL);
|
||||||
}
|
}
|
||||||
else
|
else
|
||||||
{
|
{
|
||||||
baseVal = new GpuUInt128(0UL, 1UL << x);
|
baseVal = new GpuUInt128(0UL, 1UL << x);
|
||||||
}
|
}
|
||||||
|
|
||||||
GpuUInt128 pow = baseVal;
|
GpuUInt128 pow = baseVal;
|
||||||
pow.ModPow(exponent / ux, mod);
|
pow.ModPow(exponent / ux, mod);
|
||||||
var part2 = GpuUInt128.Pow2Mod(exponent % ux, mod);
|
var part2 = GpuUInt128.Pow2Mod(exponent % ux, mod);
|
||||||
GpuUInt128 product = pow;
|
GpuUInt128 product = pow;
|
||||||
product.MulMod(part2, mod);
|
product.MulMod(part2, mod);
|
||||||
result[0] = product.High == 0UL && product.Low == 1UL ? (byte)1 : (byte)0;
|
result[0] = product.High == 0UL && product.Low == 1UL ? (byte)1 : (byte)0;
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
Reference in new issue
Block a user