Function: launch_kernel_with_config

Applicability

Product	Supported
Atlas 350 Accelerator Card	√
Atlas A3 training product/Atlas A3 inference product	√
Atlas A2 training product/Atlas A2 inference product	√
Atlas training product	√
Atlas inference product	√
Atlas 200I/500 A2 inference product	√

Description

Specifies the configuration to be delivered by a task and starts the compute task of the corresponding operator. This API is asynchronous.

If you use this API to deliver a compute task of an AI Core operator, you need to use the acl.rt.binary_load_from_file API to load and parse the operator binary file.

Prototype

C Prototype

aclError aclrtLaunchKernelWithConfig(aclrtFuncHandle funcHandle, uint32_t numBlocks, aclrtStream stream, aclrtLaunchKernelCfg *cfg, aclrtArgsHandle argsHandle, void *reserve)

Python Function

ret = acl.rt.launch_kernel_with_config(func_handle, num_blocks, stream, cfg, args_handle, reserve)

Parameters

Parameter	Description
func_handle	Int, kernel function handle.
num_blocks	Int, number of cores on which the kernel function will be executed.
stream	Int, stream for executing the task.
cfg	List, configuration to be delivered by the task. If no configuration is specified, an empty list is allowed. The data type of each element in the list is dictionary. For details, see aclrtLaunchKernelCfg.
args_handle	Int, handle of the parameter list.
reserve	Int, reserved parameter. Currently, the value is fixed at 0.

Return Value

Return Value	Description
ret	Int, 0 on success; else, failure.

Parent topic: Kernel Loading and Execution