PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
March 29, 2026Программирование / Programming and Computer Software0 citations

Application of Parallelization Capabilities of Modern Computing Architectures to Optimizing Software Performance

View Full Paper
OSO.A. Slavin

Key Points

  • The aim is to explore optimization methods for enhancing software performance on modern computing architectures.
  • Analyzed various optimization methods including algorithmic and hardware-oriented techniques.
  • Examined the capabilities of CPUs with multi-core architecture, RAM, and cache memory systems.
  • Considered instruction-level parallelism, data parallelism (SIMD), and multithreading for performance improvement.
  • Utilized examples on x86–64, ARM, and VLIW platforms to illustrate the optimization approaches.
  • Demonstrated significant improvements in software performance utilizing parallel processing methods.
  • Illustrated the effects of processor architecture and memory organization on program efficiency.
  • Showed applicability of proposed methods to image processing and cyclic computational algorithms.

Abstract

The paper considers software performance optimization technologies. The requirements for program performance, especially on mobile platforms, necessitate the use of various optimization methods, including high-level (algorithmic) and low-level (hardware-oriented) optimization. The paper considers methods for increasing program performance, taking into account the capabilities of modern computing platforms, including central processing units (CPUs) with multi-core architecture, random access memory (RAM), and cache memory systems. Attention is paid to parallelism at the instruction level (ILP), data (SIMD), and threads (multithreading). Tools for applying parallelism using compilers are considered. Several examples are given to illustrate the features of accelerating the implemented algorithms on various platforms. The x86–64 (Intel Core i9), ARM (Apple M1 Pro), and VLIW (Elbrus 2C3 and 8B) platforms are considered. The influence of the processor architecture and memory organization, including cache memory, on program performance is shown. The proposed approaches to optimizing program performance are applicable to the implementation of image processing and recognition algorithms, as well as to computational algorithms that perform cyclic information processing.

Ask AI
Helpful
Bookmark
Share
View Full Paper

Cite This Study

O.A. Slavin (2025) studied this question.

synapsesocial.com/papers/69c8c28cde0f0f753b39ce29https://doi.org/10.7868/s3034584725060033
Ask AI
Helpful
Bookmark
Share
View Full Paper