The Implementation and Cost-Effectiveness Analysis of 8-Bit and 16-Bit Jump Tables on the RISC-V Computer Architecture in LLVM

In this paper, the implementation and cost-effectiveness analysis of an optimization applied to jump tables for the RISC-V computer architecture within LLVM is presented. The focus of this work is on creating an optimization pass within LLVM that enables the compression of jump tables from 32-bit to 8-bit and 16-bit jump tables. The optimization pass, a phase in the compilation process, has the overall goal of reducing code size of the generated code. The analysis of the optimization's cost-effectiveness was performed using the CSIBE tool, comparing the results with and without the optimization.