Performance and energy optimization of ternary optical computers based on tandem queuing system

Area of Science:

• Computer Science
• Optical Computing
• Parallel Processing

Background:

• Ternary optical computers (TOCs) are emerging technologies crucial for cloud computing and big data.
• Prior research on TOCs primarily focused on performance, neglecting energy consumption impacts.
• Optimizing the performance-energy trade-off is essential for practical TOC applications.

Purpose of the Study:

• To investigate the optimization trade-off between performance and energy consumption in TOC systems.
• To develop a bi-objective optimization model for TOCs considering response time and energy usage.
• To propose a strategy for achieving balanced performance and energy efficiency under varying load conditions.

Main Methods:

• Constructed a TOC service model using M/M/1 and M/M/c queuing theories within a tandem queueing system framework.
• Analyzed the impact of processor partitioning strategy and the number of small TOCs (STOCs) on system performance and energy consumption.
• Developed a bi-objective optimization model incorporating response time and energy consumption.

Main Results:

• Increasing active STOCs significantly improves performance when response time is critical.
• Diminishing returns are observed with more STOCs, leading to increased energy costs.
• The study identified optimal strategies for partitioning and STOC allocation based on load.

Conclusions:

• A bi-objective optimization strategy effectively balances performance and energy consumption in TOCs.
• The proposed approach enables efficient resource management for TOC systems in cloud and big data environments.
• Tailoring the number of active STOCs and partitioning strategy to specific load conditions is key for optimal TOC operation.