SIMD extension to VLIW multicluster processors for embedded applications
Domenico Barretta, William Fornaciari, M. Sami, Danilo Pau
Abstract
Domenico Barretta, William Fornaciari, M. Sami, Danilo Pau
Abstract
We propose a retargetable architecture, based on a multicluster VLIW processor that can exploit either instruction level parallelism (ILP) or ILP and data level parallelism (DLP) jointly in a SIMD fashion. Simulation results show that performances may increase significantly when the application is compiled for the proposed architecture.
OpenAlex reports 5 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
We propose a retargetable architecture, based on a multicluster VLIW processor that can exploit either instruction level parallelism (ILP) or ILP and data level parallelism (DLP) jointly in a SIMD fashion. Simulation results show that performances may increase significantly when the application is compiled for the proposed architecture.
Key concepts: Very long instruction word, SIMD, Computer science, Parallel computing, Parallelism (grammar), Instruction-level parallelism, Exploit, Computer architecture