摘要 |
Disclosed is a method and an apparatus for computing precision, determining a braiding factor by the number of units of task encoding a physical thread; determining a value of the braiding factor based on a mixture of precision requirements for each of the task unit; classifying the units of the job with a command for an applied code conversion based on related precision requirements for a processing environment; compressing a command for inputting target registers according to the determined value of the braiding factor with a destination register; and executing compressed commands of other precision, displaying a command set structure comprised to execute the compressed commands in the form of a vector. |