Workload Partitioning via System Progression Rate for PDES
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current workload partitioning methods for null message-based parallel and distributed discrete event simulations (PDES) lack an accurate and quantitative performance model, making it difficult to identify critical simulation tasks and balance computation and prioritization, leading to suboptimal partitioning results.
Innovation Solution
A generic partitioning approach using a performance evaluation method that computes system progression rates based on lookahead values and transit times to efficiently identify and prioritize critical tasks, providing a system-level view of simulation performance and optimizing partitioning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing partitioning approaches use rule of thumb to identify critical LPs and force them to the same computing device, then critical task prioritization is improved, but computation balancing deteriorates
Solution Approach 1:
The patent changes the parameter used for partitioning from communication traffic impact to system progression rate. By using progression rate as the metric, the system can simultaneously identify critical tasks and balance computation across devices, resolving the contradiction between prioritizing critical tasks and balancing workload.
Solution Approach 2:
The patent implements a feedback mechanism by computing system progression rate for different partitioning schemes and using this information to evaluate and improve partitioning quality. This allows the system to iteratively optimize both critical task prioritization and computation balancing.
2Speed
If partitioning minimizes inter-LP communication delays, then communication efficiency is improved, but simulation performance evaluation accuracy deteriorates
Solution Approach 1:
The patent introduces system progression rate as an intermediary metric that connects communication efficiency to overall simulation performance. This intermediary allows accurate evaluation of partitioning quality by considering both communication delays and computation balance, rather than focusing solely on communication efficiency.
3Productivity
If partitioning balances computing resource requirements across sets, then computation balancing is improved, but critical task prioritization deteriorates
Solution Approach 1:
The patent changes the partitioning metric from communication traffic-based measures to system progression rate. This parameter change enables the system to naturally balance computation across devices while simultaneously identifying and prioritizing critical tasks, as progression rate inherently reflects both factors.
Data Source
AI summary
An embodiment of the invention includes applying a first partition to a plurality of LPs, wherein a particular LP is assigned to a first set of LPs. A second partition is applied to the LPs, wherein the particular LP is assigned to an LP set different from the first set. For both the first and second partitions, lookahead values and transit times are determined for each of the LPs and related links. For the first partition, a first system progression rate is computed using a specified function with the lookahead values and transit times determined for the first partition. For the second partition, a second system progression rate is computed using the specified function with the lookahead values and transit times determined for the second partition. The first and second system progression rates are compared to determine which is the lowest.