aclSysParamOpt

Data Format	Description
ACL_OPT_DETERMINISTIC = 0	Whether to enable deterministic computing. 0 (default): disabled 1: enabled When deterministic computing is enabled, the same output is generated if an operator is executed for multiple times with the same hardware and input. This often slows down operator execution. By default, deterministic computing is disabled. The results of multiple executions of an operator with the same hardware and input may be different. This is generally caused by asynchronous multi-thread executions during operator implementation, which changes the accumulation sequence of floating point numbers. You are advised not to enable deterministic computing because it slows down operator execution and affects performance. If the execution results of a model are different for multiple times or the precision needs to be optimized, you can enable deterministic computing to assist model debugging and optimization.
ACL_OPT_STRONG_CONSISTENCY = 2	Whether to enable strong consistency for computation. Reserved parameter. The current configuration is invalid. 0: disabled (default) 1: enabled If enabled, the computation result is deterministic, meaning that multiple executions will generate the same result. In addition, the computation result is irrelevant to the data location. For example, when performing matrix multiplication, the order of accumulation across different rows may vary, which can lead to slight differences in results for the same data in different rows. However, when strong consistency is enabled, the results will remain consistent across rows as long as the inputs are the same. By default, strong consistency is not enabled, which may cause inconsistencies in results when the same data appears in different rows. It is a convention not to enable strong consistency, as doing so slows down operator execution and downgrades performance. Enable this feature when you need strict consistency of results for identical data in different positions, or when you want to apply precision tuning to model debugging and optimization.

Data Format

Description

ACL_OPT_DETERMINISTIC = 0

Whether to enable deterministic computing.

0 (default): disabled
1: enabled

When deterministic computing is enabled, the same output is generated if an operator is executed for multiple times with the same hardware and input. This often slows down operator execution.

By default, deterministic computing is disabled. The results of multiple executions of an operator with the same hardware and input may be different. This is generally caused by asynchronous multi-thread executions during operator implementation, which changes the accumulation sequence of floating point numbers.

You are advised not to enable deterministic computing because it slows down operator execution and affects performance. If the execution results of a model are different for multiple times or the precision needs to be optimized, you can enable deterministic computing to assist model debugging and optimization.

ACL_OPT_STRONG_CONSISTENCY = 2

Whether to enable strong consistency for computation. Reserved parameter. The current configuration is invalid.

0: disabled (default)

1: enabled

If enabled, the computation result is deterministic, meaning that multiple executions will generate the same result. In addition, the computation result is irrelevant to the data location. For example, when performing matrix multiplication, the order of accumulation across different rows may vary, which can lead to slight differences in results for the same data in different rows. However, when strong consistency is enabled, the results will remain consistent across rows as long as the inputs are the same.

By default, strong consistency is not enabled, which may cause inconsistencies in results when the same data appears in different rows.

It is a convention not to enable strong consistency, as doing so slows down operator execution and downgrades performance. Enable this feature when you need strict consistency of results for identical data in different positions, or when you want to apply precision tuning to model debugging and optimization.

Parent topic: Data Types and Operation APIs