aclrtSetOpExecuteTimeOut

Applicability

Description

Sets the timeout (in seconds) for operator execution.

For the following product models, you are advised to use the aclrtSetOpExecuteTimeOutV2 API, which returns the actual timeout:

Atlas A3 training products/Atlas A3 inference products

Atlas A2 training products/Atlas A2 inference products

Prototype

aclError aclrtSetOpExecuteTimeOut(uint32_t timeout)

Parameters

Returns

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

Restrictions

• If this API is called multiple times in a process, the last configured timeout is used.
• If this API is not called, the default timeout for operator execution is shown in the following table.

  Model	Default Timeout of AI Core Operators	Default Timeout of AI CPU Operators
  Atlas A3 training products/Atlas A3 inference products	1091 seconds	60 seconds
  Atlas A2 training products/Atlas A2 inference products	1091 seconds	28 seconds
  Atlas 200I/500 A2 inference products	1091 seconds	28 seconds
  Atlas inference products	547 seconds	28 seconds
  Atlas training products	68 seconds	28 seconds

• The maximum timeout of AI Core and AI CPU operators varies according to the architecture of the product model.
  • For the following product models, the maximum timeout of operators is interval × 254 (unit: μs). interval can be obtained through the aclrtGetOpTimeoutInterval API. The interval of AI Core operators is the same as that of AI CPU operators. Therefore, the maximum timeout of operators is "interval × 254".

    Atlas A3 training products/Atlas A3 inference products

    Atlas A2 training products/Atlas A2 inference products

  • The following table lists the maximum timeouts of AI Core and AI CPU operators for different product models.

    Model	Maximum Timeout Interval of AI Core Operators	Maximum Timeout Interval of AI CPU Operators
    Atlas 200I/500 A2 inference products	1091 seconds	1091 seconds
    Atlas inference products	547 seconds	Not supported
    Atlas training products	2176 seconds	2176 seconds