Class CudaCompiledKernelExtensions

Namespace

DotCompute.Backends.CUDA.Compilation

Assembly

DotCompute.Backends.CUDA.dll

Extensions and utilities for CudaCompiledKernel with modern architecture support.

public static class CudaCompiledKernelExtensions

Inheritance

object

CudaCompiledKernelExtensions

Inherited Members

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.MemberwiseClone()

object.ReferenceEquals(object, object)

object.ToString()

Methods

CreateEnhanced(CudaContext, string, string, byte[], CompilationOptions?, ILogger, ICompilationMetadata?)

Creates an enhanced compiled kernel with optimizations for modern architectures.

public static CudaCompiledKernel CreateEnhanced(CudaContext context, string name, string entryPoint, byte[] compiledCode, CompilationOptions? options, ILogger logger, ICompilationMetadata? metadata = null)

Parameters

context CudaContext

name string

entryPoint string

compiledCode byte[]

options CompilationOptions

logger ILogger

metadata ICompilationMetadata

Returns

CudaCompiledKernel

GetOptimalLaunchConfigForModernGPU(CudaCompiledKernel, int, int)

Gets optimal launch configuration for Ada generation and newer architectures.

public static CudaLaunchConfig GetOptimalLaunchConfigForModernGPU(this CudaCompiledKernel kernel, int totalElements, int deviceId)

Parameters

kernel CudaCompiledKernel

totalElements int

deviceId int

Returns

CudaLaunchConfig

GetPerformanceHints(int)

Gets architecture-specific performance hints for kernel optimization.

public static string[] GetPerformanceHints(int deviceId)

Parameters

deviceId int

Returns

string[]

GetRecommendedSharedMemory(string, int, int)

Gets recommended shared memory configuration for the kernel and architecture.

public static uint GetRecommendedSharedMemory(string kernelName, int deviceId, int blockSize)

Parameters

kernelName string

deviceId int

blockSize int

Returns

uint

ValidateEnhancedLaunchConfig(CudaLaunchConfig, int)

Validates launch configuration against modern GPU limits.

public static bool ValidateEnhancedLaunchConfig(CudaLaunchConfig config, int deviceId)

Parameters

config CudaLaunchConfig

deviceId int

Returns

bool