Method for reducing resource consumption of instruction memory on stream processor chip

Abstract

The invention discloses a method for reducing resource consumption of an instruction memory on a stream processor chip and aims at effectively reducing the resource consumption of the instruction memory without the addition of a complex compiling algorithm based on the prior mature hardware memory structure. The method adopts the following technical scheme: correcting the instruction memory for pure soft management in a stream processor into an instruction memory with software and hardware mixing; adding a kernel hot-code searching module in a compiler and searching kernel hot codes in the stream application according to a theorem for judging the hot codes; adding an instruction stream loading instruction before all kernel hot codes in the stream compiling; adopting a software management static memory to store the kernel hot codes so as to ensure high hit ratio, storing other instructions by adopting a cache for hardware management, and reducing the requirements of the instruction memory to memory capacity by shortening the instruction occupying time of the instruction memory, thereby the capacity of the instruction memory can be reduced. The invention can reduce the resource consumption of the instruction memory in a chip.

Description

technical field [0001] The invention relates to a method for reducing resource consumption of instruction memory in a chip, in particular to a method for reducing resource consumption of instruction memory on a stream processor chip. Background technique [0002] The stream processor based on the stream architecture is a typical representative of a new generation of intensive computing-oriented high-performance microprocessors, which are specially oriented to stream applications. Streaming applications are mainly divided into two categories: one is media applications; the other is scientific computing. Streaming applications have the following main characteristics: Computational intensity, compared with traditional desktop applications, streaming applications perform a large number of arithmetic operations on each data fetched from memory; parallelism, mainly data-level parallelism, There are both instruction-level and task-level parallelism; locality refers to the locality...

AI Technical Summary

Problems solved by technology

This method relies too much on the capabilities of the software compiler. If the compiler predicts wrongly, the program performance will be extremely low.

Images