The evolution of DSP architectures: towards parallelism exploitation
R. Sernec, Matej Zajc, J.F. Tasič
Abstract
R. Sernec, Matej Zajc, J.F. Tasič
Abstract
This paper presents a path of parallelism exploitation in commercial programmable DSP processors. DSP processors have gained in their complexity and have adopted some very sophisticated parallelism extraction techniques, namely very long instruction word (VLIW) and SIMD designs. The intention is to show a development path of digital signal processors (DSP) and focuses on their features that allow parallel processing of algorithms.
OpenAlex reports 1 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.
This paper presents a path of parallelism exploitation in commercial programmable DSP processors. DSP processors have gained in their complexity and have adopted some very sophisticated parallelism extraction techniques, namely very long instruction word (VLIW) and SIMD designs. The intention is to show a development path of digital signal processors (DSP) and focuses on their features that allow parallel processing of algorithms.
Key concepts: Very long instruction word, Digital signal processing, Computer science, Parallelism (grammar), SIMD, Parallel computing, Instruction-level parallelism, Path (computing)