Compiling method, chip, electronic device and computer readable storage medium for executing program
CN114691142BActive Publication Date: 2026-07-07STREAM COMPUTING INC
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- STREAM COMPUTING INC
- Filing Date
- 2020-12-31
- Publication Date
- 2026-07-07
AI Technical Summary
Technical Problem
In existing technologies, multi-core CPUs or GPUs have low cache hit rates when processing neural network tasks, leading to frequent access to DDR memory, which reduces computing speed and increases power consumption.
Method used
By obtaining the total amount of data for each subroutine in the original program, allocating processing cores according to the data storage area size of the processing cores, determining the synchronization point location, and compiling into an executable program, the dependence on external memory is reduced.
Benefits of technology
It improves the effective computing power of multi-core systems, reduces latency and power consumption, and reduces bandwidth pressure on external memory.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN114691142B_ABST
Abstract
Embodiments of the present disclosure disclose a compiling method of an execution program, a chip, an electronic device and a computer readable storage medium. The compiling method of the execution program comprises: obtaining a total amount of data of each subprogram in an original program; determining a number of processing cores required for executing each subprogram according to a size of a data storage area of a plurality of processing cores and the total amount of data of each subprogram; grouping the subprograms according to the number of processing cores required for executing each subprogram; determining a position of a synchronization point according to a size of a parameter of each subprogram; and compiling the original program into an execution program according to the grouping and the position of the synchronization point. The above method compiles the execution program according to the total amount of input and output data of the original program to group the subprograms and generate the execution program, thereby solving the technical problem that the execution program in the prior art needs to frequently access an external memory when being executed.
Need to check novelty before this filing date? Find Prior Art