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

Naming improvements and small cleanup

Browse Source
This commit is contained in:
mlemanczyk committed 2025-11-18 03:43:52 +01:00
1 parent bf7c4bab39
commit 66ca969929
10 files changed
+35 -30
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

PerfectNumbers.Core/Gpu/Accelerators/HeuristicGpuCombinedTables.cs → PerfectNumbers.Core/Gpu/Accelerators/HeuristicCombinedGpuTables.cs
+5 -5
View File
@@ -3,17 +3,17 @@ using ILGPU.Runtime;
namespace PerfectNumbers.Core.Gpu.Accelerators; namespace PerfectNumbers.Core.Gpu.Accelerators;
internal sealed class HeuristicGpuCombinedTables internal readonly struct HeuristicCombinedGpuTables
{ {
private static readonly Accelerator[] _accelerators = AcceleratorPool.Shared.Accelerators; private static readonly Accelerator[] _accelerators = AcceleratorPool.Shared.Accelerators;
private static readonly HeuristicGpuCombinedTables[] _sharedTables = new HeuristicGpuCombinedTables[PerfectNumberConstants.RollingAccelerators]; private static readonly HeuristicCombinedGpuTables[] _sharedTables = new HeuristicCombinedGpuTables[PerfectNumberConstants.RollingAccelerators];
internal static void WarmUp(int acceleratorIndex, AcceleratorStream stream) => _sharedTables[acceleratorIndex] = new(_accelerators[acceleratorIndex], stream); internal static void WarmUp(int acceleratorIndex, AcceleratorStream stream) => _sharedTables[acceleratorIndex] = new(_accelerators[acceleratorIndex], stream);
internal static HeuristicGpuCombinedTables GetStaticTables(int acceleratorIndex) => _sharedTables[acceleratorIndex]; internal static HeuristicCombinedGpuTables GetStaticTables(int acceleratorIndex) => _sharedTables[acceleratorIndex];
internal HeuristicGpuCombinedTables(Accelerator accelerator, AcceleratorStream stream) internal HeuristicCombinedGpuTables(Accelerator accelerator, AcceleratorStream stream)
{ {
var combinedEnding1 = HeuristicCombinedPrimeTester.CombinedDivisorsEnding1; var combinedEnding1 = HeuristicCombinedPrimeTester.CombinedDivisorsEnding1;
HeuristicCombinedDivisorsEnding1 = accelerator.CopySpanToDevice(stream, combinedEnding1); HeuristicCombinedDivisorsEnding1 = accelerator.CopySpanToDevice(stream, combinedEnding1);
@@ -45,7 +45,7 @@ internal sealed class HeuristicGpuCombinedTables
internal readonly MemoryBuffer1D<ulong, Stride1D.Dense> HeuristicCombinedDivisorsEnding9; internal readonly MemoryBuffer1D<ulong, Stride1D.Dense> HeuristicCombinedDivisorsEnding9;
internal readonly MemoryBuffer1D<ulong, Stride1D.Dense> HeuristicCombinedDivisorSquaresEnding9; internal readonly MemoryBuffer1D<ulong, Stride1D.Dense> HeuristicCombinedDivisorSquaresEnding9;
internal HeuristicGpuCombinedDivisorTables CreateHeuristicDivisorTables() => new( internal HeuristicCombinedGpuViews CreateViews() => new(
HeuristicCombinedDivisorsEnding1.View, HeuristicCombinedDivisorsEnding1.View,
HeuristicCombinedDivisorsEnding3.View, HeuristicCombinedDivisorsEnding3.View,
HeuristicCombinedDivisorsEnding7.View, HeuristicCombinedDivisorsEnding7.View,
PerfectNumbers.Core/Gpu/Accelerators/HeuristicGpuGroupABTables.cs → PerfectNumbers.Core/Gpu/Accelerators/HeuristicGroupABGpuTables.cs
+5 -5
View File
@@ -3,15 +3,15 @@ using ILGPU.Runtime;
namespace PerfectNumbers.Core.Gpu.Accelerators; namespace PerfectNumbers.Core.Gpu.Accelerators;
internal sealed class HeuristicGpuGroupABTables internal sealed class HeuristicGroupABGpuTables
{ {
private static readonly Accelerator[] _accelerators = AcceleratorPool.Shared.Accelerators; private static readonly Accelerator[] _accelerators = AcceleratorPool.Shared.Accelerators;
private static readonly HeuristicGpuGroupABTables[] _sharedTables = new HeuristicGpuGroupABTables[PerfectNumberConstants.RollingAccelerators]; private static readonly HeuristicGroupABGpuTables[] _sharedTables = new HeuristicGroupABGpuTables[PerfectNumberConstants.RollingAccelerators];
private static readonly SemaphoreSlim[] _locks = [..Enumerable.Range(1, PerfectNumberConstants.RollingAccelerators).Select(_ => new SemaphoreSlim(1))]; private static readonly SemaphoreSlim[] _locks = [..Enumerable.Range(1, PerfectNumberConstants.RollingAccelerators).Select(_ => new SemaphoreSlim(1))];
internal static HeuristicGpuGroupABTables EnsureStaticTables(int acceleratorIndex, AcceleratorStream stream) internal static HeuristicGroupABGpuTables EnsureStaticTables(int acceleratorIndex, AcceleratorStream stream)
{ {
var @lock = _locks[acceleratorIndex]; var @lock = _locks[acceleratorIndex];
@lock.Wait(); @lock.Wait();
@@ -28,7 +28,7 @@ internal sealed class HeuristicGpuGroupABTables
return existing; return existing;
} }
internal HeuristicGpuGroupABTables(Accelerator accelerator, AcceleratorStream stream) internal HeuristicGroupABGpuTables(Accelerator accelerator, AcceleratorStream stream)
{ {
var heuristicGroupA = HeuristicPrimeSieves.GroupADivisorsStorage; var heuristicGroupA = HeuristicPrimeSieves.GroupADivisorsStorage;
HeuristicGroupADivisors = accelerator.CopySpanToDevice(stream, heuristicGroupA); HeuristicGroupADivisors = accelerator.CopySpanToDevice(stream, heuristicGroupA);
@@ -55,7 +55,7 @@ internal sealed class HeuristicGpuGroupABTables
internal readonly MemoryBuffer1D<ulong, Stride1D.Dense> HeuristicGroupBDivisorsEnding9; internal readonly MemoryBuffer1D<ulong, Stride1D.Dense> HeuristicGroupBDivisorsEnding9;
internal readonly MemoryBuffer1D<ulong, Stride1D.Dense> HeuristicGroupBDivisorSquaresEnding9; internal readonly MemoryBuffer1D<ulong, Stride1D.Dense> HeuristicGroupBDivisorSquaresEnding9;
internal HeuristicGpuGroupABDivisorTables CreateHeuristicDivisorTables() => new( internal HeuristicGroupABGpuViews CreateViews() => new(
HeuristicGroupADivisors.View, HeuristicGroupADivisors.View,
HeuristicGroupADivisorSquares.View, HeuristicGroupADivisorSquares.View,
HeuristicGroupBDivisorsEnding1.View, HeuristicGroupBDivisorsEnding1.View,
PerfectNumbers.Core/Gpu/Accelerators/HeuristicGroupABPrimeTesterAccelerator.cs
+6 -6
View File
@@ -19,7 +19,7 @@ internal sealed class HeuristicGroupABPrimeTesterAccelerator
Console.WriteLine($"Preparing heuristic accelerator {i}..."); Console.WriteLine($"Preparing heuristic accelerator {i}...");
var accelerator = accelerators[i]; var accelerator = accelerators[i];
AcceleratorStream stream = accelerator.CreateStream(); AcceleratorStream stream = accelerator.CreateStream();
_ = HeuristicGpuGroupABTables.EnsureStaticTables(i, stream); _ = HeuristicGroupABGpuTables.EnsureStaticTables(i, stream);
// _ = GpuKernelPool.GetOrAddKernels(accelerator, stream, KernelType.None); // _ = GpuKernelPool.GetOrAddKernels(accelerator, stream, KernelType.None);
// KernelContainer kernels = GpuKernelPool.GetOrAddKernels(accelerator, stream); // KernelContainer kernels = GpuKernelPool.GetOrAddKernels(accelerator, stream);
// GpuStaticTableInitializer.EnsureStaticTables(accelerator, kernels, stream); // GpuStaticTableInitializer.EnsureStaticTables(accelerator, kernels, stream);
@@ -75,7 +75,7 @@ internal sealed class HeuristicGroupABPrimeTesterAccelerator
#endregion #endregion
public readonly HeuristicGpuGroupABDivisorTables DivisorTables; public readonly HeuristicGroupABGpuViews DivisorTables;
internal readonly Context Context; internal readonly Context Context;
public readonly Accelerator Accelerator; public readonly Accelerator Accelerator;
@@ -102,7 +102,7 @@ internal sealed class HeuristicGroupABPrimeTesterAccelerator
public MemoryBuffer1D<int, Stride1D.Dense>? HeuristicFlag = null!; public MemoryBuffer1D<int, Stride1D.Dense>? HeuristicFlag = null!;
public int BufferCapacity; public int BufferCapacity;
private HeuristicGpuGroupABTables _sharedTables; private HeuristicGroupABGpuTables _sharedTables;
public MemoryBuffer1D<ulong, Stride1D.Dense> HeuristicGroupADivisors => _sharedTables.HeuristicGroupADivisors; public MemoryBuffer1D<ulong, Stride1D.Dense> HeuristicGroupADivisors => _sharedTables.HeuristicGroupADivisors;
public MemoryBuffer1D<ulong, Stride1D.Dense> HeuristicGroupADivisorSquares => _sharedTables.HeuristicGroupADivisorSquares; public MemoryBuffer1D<ulong, Stride1D.Dense> HeuristicGroupADivisorSquares => _sharedTables.HeuristicGroupADivisorSquares;
public MemoryBuffer1D<ulong, Stride1D.Dense> HeuristicGroupBDivisorsEnding1 => _sharedTables.HeuristicGroupBDivisorsEnding1; public MemoryBuffer1D<ulong, Stride1D.Dense> HeuristicGroupBDivisorsEnding1 => _sharedTables.HeuristicGroupBDivisorsEnding1;
@@ -125,14 +125,14 @@ internal sealed class HeuristicGroupABPrimeTesterAccelerator
Accelerator = accelerator; Accelerator = accelerator;
EnsureCapacity(minBufferCapacity); EnsureCapacity(minBufferCapacity);
HeuristicTrialDivisionKernel = KernelUtil.GetKernel(accelerator.LoadAutoGroupedStreamKernel<Index1D, ArrayView<int>, ulong, ulong, HeuristicGpuDivisorTableKind, HeuristicGpuGroupABDivisorTables>(PrimeTesterKernels.HeuristicTrialGroupABDivisionKernel)); HeuristicTrialDivisionKernel = KernelUtil.GetKernel(accelerator.LoadAutoGroupedStreamKernel<Index1D, ArrayView<int>, ulong, ulong, HeuristicGpuDivisorTableKind, HeuristicGroupABGpuViews>(PrimeTesterKernels.HeuristicTrialGroupABDivisionKernel));
AcceleratorStream stream = AcceleratorStreamPool.Rent(acceleratorIndex); AcceleratorStream stream = AcceleratorStreamPool.Rent(acceleratorIndex);
// GpuStaticTableInitializer.EnsureStaticTables(accelerator, kernels, stream); // GpuStaticTableInitializer.EnsureStaticTables(accelerator, kernels, stream);
_sharedTables = HeuristicGpuGroupABTables.EnsureStaticTables(acceleratorIndex, stream); _sharedTables = HeuristicGroupABGpuTables.EnsureStaticTables(acceleratorIndex, stream);
DivisorTables = _sharedTables.CreateHeuristicDivisorTables(); DivisorTables = _sharedTables.CreateViews();
stream.Synchronize(); stream.Synchronize();
AcceleratorStreamPool.Return(acceleratorIndex, stream); AcceleratorStreamPool.Return(acceleratorIndex, stream);
PerfectNumbers.Core/Gpu/Accelerators/PrimeOrderCalculatorAccelerator.cs
+12 -7
View File
@@ -69,7 +69,7 @@ public sealed class PrimeOrderCalculatorAccelerator
AcceleratorStream stream = accelerator.CreateStream(); AcceleratorStream stream = accelerator.CreateStream();
// TODO: Review which tables are needed when final execution path is defined. // TODO: Review which tables are needed when final execution path is defined.
LastDigitGpuTables.WarmUp(i, stream); LastDigitGpuTables.WarmUp(i, stream);
HeuristicGpuCombinedTables.WarmUp(i, stream); HeuristicCombinedGpuTables.WarmUp(i, stream);
// SmallPrimeFactorGpuTables.WarmUp(i, stream); // SmallPrimeFactorGpuTables.WarmUp(i, stream);
// SharedHeuristicGpuTables.EnsureStaticTables(accelerator, stream); // SharedHeuristicGpuTables.EnsureStaticTables(accelerator, stream);
// _ = GpuKernelPool.GetOrAddKernels(accelerator, stream, KernelType.None); // _ = GpuKernelPool.GetOrAddKernels(accelerator, stream, KernelType.None);
@@ -111,8 +111,9 @@ public sealed class PrimeOrderCalculatorAccelerator
public readonly MemoryBuffer1D<int, Stride1D.Dense> SmallPrimeFactorCountSlot; public readonly MemoryBuffer1D<int, Stride1D.Dense> SmallPrimeFactorCountSlot;
public readonly MemoryBuffer1D<ulong, Stride1D.Dense> SpecialMaxResult; public readonly MemoryBuffer1D<ulong, Stride1D.Dense> SpecialMaxResult;
public readonly HeuristicGpuCombinedDivisorTables DivisorTables; public readonly HeuristicCombinedGpuViews DivisorTables;
public readonly SmallPrimeFactorGpuTables SmallPrimeFactorTables; public readonly MemoryBuffer1D<uint, Stride1D.Dense> SmallPrimeFactorPrimes;
public readonly MemoryBuffer1D<ulong, Stride1D.Dense> SmallPrimeFactorSquares;
public readonly Kernel CheckFactorsKernel; public readonly Kernel CheckFactorsKernel;
public readonly Kernel ConvertToStandardKernel; public readonly Kernel ConvertToStandardKernel;
@@ -217,9 +218,13 @@ public sealed class PrimeOrderCalculatorAccelerator
DevicePrimesPow2LastThree = lastDigitSharedTables.DevicePrimesPow2LastThree; DevicePrimesPow2LastThree = lastDigitSharedTables.DevicePrimesPow2LastThree;
DevicePrimesPow2LastNine = lastDigitSharedTables.DevicePrimesPow2LastNine; DevicePrimesPow2LastNine = lastDigitSharedTables.DevicePrimesPow2LastNine;
var heuristicSharedTables = HeuristicGpuCombinedTables.GetStaticTables(acceleratorIndex); var heuristicSharedTables = HeuristicCombinedGpuTables.GetStaticTables(acceleratorIndex);
DivisorTables = heuristicSharedTables.CreateHeuristicDivisorTables();
SmallPrimeFactorTables = SmallPrimeFactorGpuTables.GetStaticTables(acceleratorIndex); DivisorTables = heuristicSharedTables.CreateViews();
var smallPrimeFactorTables = SmallPrimeFactorGpuTables.GetStaticTables(acceleratorIndex);
SmallPrimeFactorPrimes = smallPrimeFactorTables.Primes;
SmallPrimeFactorSquares = smallPrimeFactorTables.Squares;
var checkFactorsKernel = accelerator.LoadStreamKernel<int, ulong, ArrayView1D<KeyValuePair<ulong, int>, Stride1D.Dense>, MontgomeryDivisorData, ArrayView1D<ulong, Stride1D.Dense>>(PrimeOrderGpuHeuristics.CheckFactorsKernel); var checkFactorsKernel = accelerator.LoadStreamKernel<int, ulong, ArrayView1D<KeyValuePair<ulong, int>, Stride1D.Dense>, MontgomeryDivisorData, ArrayView1D<ulong, Stride1D.Dense>>(PrimeOrderGpuHeuristics.CheckFactorsKernel);
CheckFactorsKernel = KernelUtil.GetKernel(checkFactorsKernel); CheckFactorsKernel = KernelUtil.GetKernel(checkFactorsKernel);
@@ -227,7 +232,7 @@ public sealed class PrimeOrderCalculatorAccelerator
var convertKernel = accelerator.LoadAutoGroupedStreamKernel<Index1D, ArrayView1D<ulong, Stride1D.Dense>, MontgomeryDivisorData, ArrayView1D<ulong, Stride1D.Dense>>(Pow2MontgomeryKernels.Pow2MontgomeryKernelConvertToStandard); var convertKernel = accelerator.LoadAutoGroupedStreamKernel<Index1D, ArrayView1D<ulong, Stride1D.Dense>, MontgomeryDivisorData, ArrayView1D<ulong, Stride1D.Dense>>(Pow2MontgomeryKernels.Pow2MontgomeryKernelConvertToStandard);
ConvertToStandardKernel = KernelUtil.GetKernel(convertKernel); ConvertToStandardKernel = KernelUtil.GetKernel(convertKernel);
HeuristicCombinedTrialDivisionKernel = KernelUtil.GetKernel(accelerator.LoadAutoGroupedStreamKernel<Index1D, byte, ArrayView<int>, ulong, ulong, HeuristicGpuCombinedDivisorTables>(PrimeTesterKernels.HeuristicTrialCombinedDivisionKernel)); HeuristicCombinedTrialDivisionKernel = KernelUtil.GetKernel(accelerator.LoadAutoGroupedStreamKernel<Index1D, byte, ArrayView<int>, ulong, ulong, HeuristicCombinedGpuViews>(PrimeTesterKernels.HeuristicTrialCombinedDivisionKernel));
var keepKernel = accelerator.LoadAutoGroupedStreamKernel<Index1D, ArrayView1D<ulong, Stride1D.Dense>, MontgomeryDivisorData, ArrayView1D<ulong, Stride1D.Dense>>(Pow2MontgomeryKernels.Pow2MontgomeryKernelKeepMontgomery); var keepKernel = accelerator.LoadAutoGroupedStreamKernel<Index1D, ArrayView1D<ulong, Stride1D.Dense>, MontgomeryDivisorData, ArrayView1D<ulong, Stride1D.Dense>>(Pow2MontgomeryKernels.Pow2MontgomeryKernelKeepMontgomery);
KeepMontgomeryKernel = KernelUtil.GetKernel(keepKernel); KeepMontgomeryKernel = KernelUtil.GetKernel(keepKernel);
PerfectNumbers.Core/Gpu/HeuristicGpuCombinedDivisorTables.cs → PerfectNumbers.Core/Gpu/HeuristicCombinedGpuViews.cs
+1 -1
View File
@@ -4,7 +4,7 @@ using ILGPU.Runtime;
namespace PerfectNumbers.Core.Gpu; namespace PerfectNumbers.Core.Gpu;
public readonly struct HeuristicGpuCombinedDivisorTables( public readonly struct HeuristicCombinedGpuViews(
ArrayView1D<ulong, Stride1D.Dense> combinedDivisorsEnding1, ArrayView1D<ulong, Stride1D.Dense> combinedDivisorsEnding1,
ArrayView1D<ulong, Stride1D.Dense> combinedDivisorsEnding3, ArrayView1D<ulong, Stride1D.Dense> combinedDivisorsEnding3,
ArrayView1D<ulong, Stride1D.Dense> combinedDivisorsEnding7, ArrayView1D<ulong, Stride1D.Dense> combinedDivisorsEnding7,
PerfectNumbers.Core/Gpu/HeuristicGpuGroupABDivisorTables.cs → PerfectNumbers.Core/Gpu/HeuristicGroupABGpuViews.cs
+1 -1
View File
@@ -4,7 +4,7 @@ using ILGPU.Runtime;
namespace PerfectNumbers.Core.Gpu; namespace PerfectNumbers.Core.Gpu;
public readonly struct HeuristicGpuGroupABDivisorTables( public readonly struct HeuristicGroupABGpuViews(
ArrayView1D<ulong, Stride1D.Dense> groupADivisors, ArrayView1D<ulong, Stride1D.Dense> groupADivisors,
ArrayView1D<ulong, Stride1D.Dense> groupADivisorSquares, ArrayView1D<ulong, Stride1D.Dense> groupADivisorSquares,
ArrayView1D<ulong, Stride1D.Dense> groupBDivisorsEnding1, ArrayView1D<ulong, Stride1D.Dense> groupBDivisorsEnding1,
PerfectNumbers.Core/Gpu/Kernels/PrimeTesterKernels.cs
+2 -2
View File
@@ -83,7 +83,7 @@ internal static class PrimeTesterKernels
ArrayView<int> resultFlag, ArrayView<int> resultFlag,
ulong n, ulong n,
ulong maxDivisorSquare, ulong maxDivisorSquare,
HeuristicGpuCombinedDivisorTables tables) HeuristicCombinedGpuViews tables)
{ {
// byte nMod10 = (byte)(n % 10UL); // byte nMod10 = (byte)(n % 10UL);
ArrayView1D<ulong, Stride1D.Dense> divisors = tables.SelectDivisors(nMod10); ArrayView1D<ulong, Stride1D.Dense> divisors = tables.SelectDivisors(nMod10);
@@ -116,7 +116,7 @@ internal static class PrimeTesterKernels
ulong n, ulong n,
ulong maxDivisorSquare, ulong maxDivisorSquare,
HeuristicGpuDivisorTableKind tableKind, HeuristicGpuDivisorTableKind tableKind,
HeuristicGpuGroupABDivisorTables tables) HeuristicGroupABGpuViews tables)
{ {
ArrayView1D<ulong, Stride1D.Dense> divisors = tables.SelectDivisors(tableKind); ArrayView1D<ulong, Stride1D.Dense> divisors = tables.SelectDivisors(tableKind);
ArrayView1D<ulong, Stride1D.Dense> divisorSquares = tables.SelectDivisorSquares(tableKind); ArrayView1D<ulong, Stride1D.Dense> divisorSquares = tables.SelectDivisorSquares(tableKind);
PerfectNumbers.Core/Gpu/SmallPrimeFactorGpuTables.cs
+1 -1
View File
@@ -3,8 +3,8 @@ using ILGPU;
using ILGPU.Runtime; using ILGPU.Runtime;
namespace PerfectNumbers.Core.Gpu; namespace PerfectNumbers.Core.Gpu;
public readonly struct SmallPrimeFactorGpuTables public readonly struct SmallPrimeFactorGpuTables
{ {
private static readonly Accelerator[] _accelerators = AcceleratorPool.Shared.Accelerators; private static readonly Accelerator[] _accelerators = AcceleratorPool.Shared.Accelerators;
private static readonly SmallPrimeFactorGpuTables[] _sharedTables = new SmallPrimeFactorGpuTables[PerfectNumberConstants.RollingAccelerators]; private static readonly SmallPrimeFactorGpuTables[] _sharedTables = new SmallPrimeFactorGpuTables[PerfectNumberConstants.RollingAccelerators];
PerfectNumbers.Core/HeuristicCombinedPrimeTester.cs
+1 -1
View File
@@ -246,7 +246,7 @@ public sealed class HeuristicCombinedPrimeTester
int compositeFlag = 0; int compositeFlag = 0;
flagView1D.CopyFromCPU(stream, ref compositeFlag, 1); flagView1D.CopyFromCPU(stream, ref compositeFlag, 1);
var kernelLauncher = kernel.CreateLauncherDelegate<Action<AcceleratorStream, Index1D, byte, ArrayView<int>, ulong, ulong, HeuristicGpuCombinedDivisorTables>>(); var kernelLauncher = kernel.CreateLauncherDelegate<Action<AcceleratorStream, Index1D, byte, ArrayView<int>, ulong, ulong, HeuristicCombinedGpuViews>>();
kernelLauncher( kernelLauncher(
stream, stream,
1, 1,
PerfectNumbers.Core/HeuristicPrimeTester.cs
+1 -1
View File
@@ -261,7 +261,7 @@ public sealed class HeuristicPrimeTester
// HeuristicGpuDivisorTableKind tableKind, // HeuristicGpuDivisorTableKind tableKind,
// HeuristicGpuDivisorTables tables) // HeuristicGpuDivisorTables tables)
var kernelLauncher = kernel.CreateLauncherDelegate<Action<AcceleratorStream, Index1D, byte, ArrayView<int>, ulong, ulong, HeuristicGpuDivisorTableKind, HeuristicGpuCombinedDivisorTables>>(); var kernelLauncher = kernel.CreateLauncherDelegate<Action<AcceleratorStream, Index1D, byte, ArrayView<int>, ulong, ulong, HeuristicGpuDivisorTableKind, HeuristicCombinedGpuViews>>();
kernelLauncher( kernelLauncher(
stream, stream,
1, 1,