7.46. cudaKernelNodeParamsV2

struct cudaKernelNodeParamsV2

CUDA GPU kernel node parameters.

Public Members

union cudaKernelNodeParamsV2::[anonymous] [anonymous]

cudaLaunchAttribute *attrs

List of attributes; nullable if cudaKernelNodeParamsV2::numAttrs == 0.

If the only attribute(s) in the array is(are) cudaLaunchAttributeIgnore, the node’s current attributes will be cleared. If cudaKernelNodeParamsV2::numAttrs == 0, the node’s current attributes are unchanged.

uint3 blockDim

Block dimensions.

cudaExecutionContext_t ctx

Context in which to run the kernel.

If NULL will try to use the current context.

cudaFunction_t cuFunc

functionType = cudaKernelFucntionTypeFunction

void **extra

Pointer to kernel arguments in the “extra” format.

void *func

functionType = cudaKernelFucntionTypeDevice

enum cudaKernelFunctionType functionType

Type of handle passed in the func/kern/cuFunc union above.

uint3 gridDim

Grid dimensions.

cudaKernel_t kern

functionType = cudaKernelFucntionTypeKernel

void **kernelParams

Array of pointers to individual kernel arguments.

unsigned int numAttrs

Number of attributes populated in cudaKernelNodeParamsV2::attrs.

unsigned int sharedMemBytes

Dynamic shared-memory size per thread block in bytes.