aclrtLaunchKernel

Product	Supported
Ascend 950PR/Ascend 950DT	√
Atlas A3 training products/Atlas A3 inference products	√
Atlas A2 training products/Atlas A2 inference products	√
Atlas 200I/500 A2 inference products	√
Atlas inference products	√
Atlas training products	√

Starts the compute task of an operator. This is an asynchronous API. The operator here is a custom operator developed by using Ascend C.

aclError aclrtLaunchKernel(aclrtFuncHandle funcHandle, uint32_t numBlocks, const void *argsData, size_t argsSize, aclrtStream stream)

Parameter	Input/Output	Description
funcHandle	Input	Kernel function handle. For details about the type definition, see aclrtFuncHandle. Call aclrtBinaryGetFunction to obtain funcHandle based on kernelName.
numBlocks	Input	Number of cores on which the kernel function will be executed.
argsData	Input	Pointer to the device memory address that stores all input parameters of the kernel function. For details about memory allocation APIs, see Memory Management. The device that executes the task delivered by this API must be the same as the device that contains the memory used by argsData.
argsSize	Input	Size of parameter values in argsData, in bytes.
stream	Input	Stream for executing the task. For details about the type definition, see aclrtStream.

0 on success; otherwise, failure. For details, see aclError.

All the APIs listed in the table below can start the compute task of an operator. However, their functions and usage methods are different.

API	How Kernel Function Parameter Values Are Passed	Where Kernel Function Parameter Values Are Stored	Task Configurable
aclrtLaunchKernel	Specify the pointer to the device memory address that stores all input parameters of the kernel function.	Device memory	No
aclrtLaunchKernelV2	Specify the pointer to the device memory address that stores all input parameters of the kernel function.	Device memory	Yes
aclrtLaunchKernelWithConfig	Specify the parameter list handle aclrtArgsHandle.	Host memory	Yes
aclrtLaunchKernelWithHostArgs	Specify the pointer to the host memory address that stores all input parameters of the kernel function.	Host memory	Yes
aclrtLaunchKernelWithArgsArray	Specify the parameter array that stores all input parameters of the kernel function. Each element in the array points to the data of a kernel function parameter.	Host memory	Yes

Parent topic: Kernel Loading and Execution