CrossCoreWaitFlag (ISASI)

Applicability

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

Function Usage

Sets the inter-core synchronization control for the separate mode. This API works with CrossCoreSetFlag. For details, see CrossCoreSetFlag.

Prototype

      
           template <uint8_t modeId = 0, pipe_t pipe = PIPE_S>
__aicore__ inline void CrossCoreWaitFlag(uint16_t flagId)

Parameters

**Table 1** Template parameters
Parameter	Description
modeId	Inter-core synchronization mode. The options are as follows: Mode 0: synchronization control between AI Cores. Mode 1: synchronization control between AIV cores in the AI Core. Mode 2: synchronization control between AIC and AIV cores in the AI Core. Mode 4: synchronization control between AIC and AIV cores in the AI Core. AIV0 and AIV1 can trigger AIC wait independently.
pipe	The pipeline type of the instruction. For details about the pipeline types, see Pipelines. Note that the PIPE_S type is supported only by the Atlas 350 Accelerator Card.

**Table 2** Parameters
Parameter	Input/Output	Description
flagId	Input	Flag ID of the inter-core synchronization. For the Atlas A2 training product / Atlas A2 inference product , the value ranges from 0 to 10. For the Atlas A3 training product / Atlas A3 inference product , the value ranges from 0 to 10. For the Atlas 350 Accelerator Card, the value ranges are as follows: The operations with flag IDs 0 to 10 in CrossCoreSetFlag initiated by AIV0 correspond to the operations with flag IDs 0 to 10 in CrossCoreWaitFlag on the AIC. The operations with flag IDs 0 to 10 in CrossCoreSetFlag initiated by AIV1 correspond to the operations with flag IDs 16 to 26 in CrossCoreWaitFlag on the AIC. The operations with flag IDs 0 to 10 in CrossCoreSetFlag initiated by AIC correspond to the operations with flag IDs 0 to 10 in CrossCoreWaitFlag on AIV0. The operations with flag IDs 16 to 26 in CrossCoreSetFlag initiated by AIC correspond to the operations with flag IDs 0 to 10 in CrossCoreWaitFlag on AIV1.

Returns

None

Constraints

When using this synchronization API, you need to set the kernel type according to the following rules:
- In pure vector or cube scenarios, set the kernel type to KERNEL_TYPE_MIX_AIV_1_0 or KERNEL_TYPE_MIX_AIC_1_0.
- In vector and cube fused scenarios, set the kernel type based on the actual situation.

CrossCoreWaitFlag must be used together with CrossCoreSetFlag to prevent the computing core from being always in the blocking phase.
If the value of the counter of flagId is 0 when CrossCoreWaitFlag is executed, all instructions after CrossCoreWaitFlag are blocked until the value of the counter of flagId is not 0. The counter of the same flagId can be set for a maximum of 15 times.
When using mode 0 of this API, you are advised to enable the batch mode so that an operator exclusively occupies all required core resources. Otherwise, a deadlock may occur if the following conditions are met:
- Multi-stream concurrency scenarios (≥ 2 execution streams)
- Concurrent execution of at least two operators.
- The total number of cores of all concurrent operators exceeds the number of physical cores.
- Two or more concurrent operators use inter-core synchronization.
Specifically, in a multi-stream scenario, although n physical cores are allocated to an inter-core synchronization operator of a stream, only (n – m) cores may be scheduled and executed first, and remaining m cores are not started because they are preempted by inter-core synchronization operators of other streams. When the first started (n – m) cores reach inter-core synchronization, they wait for the remaining m cores to complete execution. However, the remaining m cores are occupied by the inter-core synchronization operators of other streams and cannot be released, resulting in a deadlock.

In kernel direct debugging scenarios, use the __schedmode__(mode) qualifier to set the batch mode. In project-based operator development scenarios, use the SetScheduleMode API of TilingContext to set the batch mode. For details, see Basic Data Structures and API List.

Example

For details, see Example.

Parent topic: Inter-Core Synchronization