RunCpuKernelAsync

Description

Asynchronously executes the Compute function of the AI CPU operator kernel.

Prototype

uint32_t RunCpuKernelAsync(CpuKernelContext &ctx, const uint8_t wait_type, const uint32_t wait_id, std::function<uint32_t()> cb)

Parameters

Parameter	Input/Output	Description
ctx	Input	Context of CpuKernel. CpuKernelContext type.
wait_type	Input	Event waiting type. The value can be 0 or 1. 0: Disable the event waiting function. 1: Enable the event waiting function. Note: In the asynchronous scenario, this parameter must be set to 1.
wait_id	Input	Event ID, which is the event waiting ID.
cb	Input	Post-Compute processing callback function.

Returns

uint32_t for the RunCpuKernelAsync execution result.

0: success
Other values: failure

Restrictions

None

Example

CpuKernelContext ctx;
uint8_t wait_type = 1;
uint32_t wait_id = 0;
uint32_t UpdateFWKOutputShape(const CpuKernelContext &ctx) const {
  // Update the specific output shape information.
  // ...
}
uint32_t ret = CpuKernelRegister::Instance().RunCpuKernelAsync(ctx, wait_type, wait_id, [&, ctx]() { return UpdateFWKOutputShape(ctx); });

Parent topic: Class CpuKernelRegister