Class KernelLaunchParameters

Namespace

Orleans.GpuBridge.Abstractions.Models

Assembly

Orleans.GpuBridge.Abstractions.dll

Defines the parameters required for launching a kernel on a compute device. These parameters specify the execution configuration, work distribution, memory requirements, and kernel arguments needed for proper kernel execution.

public sealed record KernelLaunchParameters : IEquatable<KernelLaunchParameters>

Inheritance

object

KernelLaunchParameters

Implements

IEquatable<KernelLaunchParameters>

Inherited Members

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.ReferenceEquals(object, object)

object.ToString()

Constructors

KernelLaunchParameters(int[], int[]?, int, IReadOnlyDictionary<string, object>?)

Defines the parameters required for launching a kernel on a compute device. These parameters specify the execution configuration, work distribution, memory requirements, and kernel arguments needed for proper kernel execution.

public KernelLaunchParameters(int[] GlobalWorkSize, int[]? LocalWorkSize = null, int DynamicSharedMemoryBytes = 0, IReadOnlyDictionary<string, object>? Arguments = null)

Parameters

GlobalWorkSize int[]

The global work size for kernel execution, specified as an array of dimensions. This defines the total amount of work to be performed across all compute units. For example, [1024, 512] might specify a 2D grid with 1024 items in the X dimension and 512 items in the Y dimension. The number of dimensions must match the kernel's expected dimensionality (typically 1D, 2D, or 3D).

LocalWorkSize int[]

The local work size (workgroup size) for kernel execution, specified as an array of dimensions. This defines how the global work is divided into local workgroups that execute together on the same compute unit. If null, the backend will automatically select an appropriate local work size based on device characteristics and kernel requirements. When specified, the dimensions must match those of GlobalWorkSize.

DynamicSharedMemoryBytes int

The amount of dynamic shared memory to allocate per workgroup, in bytes. Shared memory is fast on-chip memory that can be accessed by all threads within a workgroup. This is in addition to any statically declared shared memory in the kernel code. Default is 0, meaning no additional shared memory is allocated.

Arguments IReadOnlyDictionary<string, object>

Kernel arguments as key-value pairs where keys are parameter names and values are argument data. This dictionary contains all the parameters that need to be passed to the kernel function. The keys should match the parameter names in the kernel signature, and values must be compatible with the expected parameter types. If null, the kernel is assumed to have no parameters.

Properties

Arguments

Kernel arguments as key-value pairs where keys are parameter names and values are argument data. This dictionary contains all the parameters that need to be passed to the kernel function. The keys should match the parameter names in the kernel signature, and values must be compatible with the expected parameter types. If null, the kernel is assumed to have no parameters.

public IReadOnlyDictionary<string, object>? Arguments { get; init; }

Property Value

IReadOnlyDictionary<string, object>

DynamicSharedMemoryBytes

The amount of dynamic shared memory to allocate per workgroup, in bytes. Shared memory is fast on-chip memory that can be accessed by all threads within a workgroup. This is in addition to any statically declared shared memory in the kernel code. Default is 0, meaning no additional shared memory is allocated.

public int DynamicSharedMemoryBytes { get; init; }

Property Value

int

GlobalWorkSize

The global work size for kernel execution, specified as an array of dimensions. This defines the total amount of work to be performed across all compute units. For example, [1024, 512] might specify a 2D grid with 1024 items in the X dimension and 512 items in the Y dimension. The number of dimensions must match the kernel's expected dimensionality (typically 1D, 2D, or 3D).

public int[] GlobalWorkSize { get; init; }

Property Value

int[]

LocalWorkSize

The local work size (workgroup size) for kernel execution, specified as an array of dimensions. This defines how the global work is divided into local workgroups that execute together on the same compute unit. If null, the backend will automatically select an appropriate local work size based on device characteristics and kernel requirements. When specified, the dimensions must match those of GlobalWorkSize.

public int[]? LocalWorkSize { get; init; }

Property Value

int[]