Computing-System Performance Simulation via Task Conversion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing simulation methods for computing systems, such as discrete event simulation and analysis model-based simulation, are inefficient for predicting performance due to scalability and parallelism issues, especially when dealing with various applications of different scales and degrees of parallelism.
Innovation Solution
A computing device and method that converts complex tasks into simpler tasks with reduced scale and parallelism, using a simulation to predict performance by adjusting cache memory sizes and obtaining success probabilities for cache hits, allowing for accurate performance estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If discrete event simulation is used to simulate computing system performance, then simulation precision for simple applications is improved, but simulation efficiency deteriorates due to time subdivision and clock frequency constraints
Solution Approach 1:
The patent changes the simulation approach from discrete event simulation to analysis model-based simulation. This involves transforming the simulation parameters and methodology to use mathematical models and statistical analysis instead of time-based discrete events, thereby improving efficiency while maintaining precision for complex applications
Solution Approach 2:
The patent replaces the mechanical/discrete event simulation mechanism with an analysis model-based approach. Instead of simulating each discrete event in time, the system uses mathematical analysis models to predict performance, substituting the simulation mechanism with a more efficient analytical approach
2Productivity
If analysis model-based simulation is used to simulate computing system performance, then simulation efficiency is improved, but adaptability deteriorates due to restrictions on hardware configurations, applications, and program inputs
Solution Approach 1:
The patent enhances the universality of the simulation system by developing a framework that can handle various hardware configurations, applications, and program inputs through a single analysis model-based approach. The system uses configurable parameters and statistical models that adapt to different scenarios without requiring separate simulation methodologies
Solution Approach 2:
The patent employs parameter changes to enable the analysis model to adapt to different hardware configurations, applications, and inputs. By using configurable parameters and statistical distributions, the model can be adjusted to match various real-world scenarios while maintaining the same efficient simulation framework
3Measurement precision
If simulation is performed on large-scale hardware with complex applications, then simulation accuracy is improved, but simulation time increases significantly
Solution Approach 1:
The patent substitutes direct simulation of large-scale complex applications with analysis model-based prediction. Instead of running time-consuming simulations on large hardware configurations, the system uses mathematical models and statistical analysis to predict performance, dramatically reducing simulation time while maintaining accuracy
Solution Approach 2:
The patent applies preliminary action by pre-establishing analysis models and statistical frameworks that can quickly predict performance for large-scale applications. The models are prepared in advance with configurable parameters, allowing rapid assessment without requiring time-consuming real-time simulations
Data Source
AI summary
A method includes: converting a first task assigned to simulated hardware having a first configuration of a simulated computing system into a second task, where the first task requires a first scale and a first degree of parallelism and the second task requires a second scale that is equal to the first scale and a second degree of parallelism; obtaining a probability that the simulated hardware having the first configuration succeeds in a requested event by executing a simulation in which hardware having a second configuration processes the second task; and based on the probability, based on the first scale of the first task, and based on parameter information of the simulated hardware having the first configuration, predicting a statistical performance index of the simulated hardware having the first configuration when the simulated hardware having the first configuration processes the first task.