Design and Analysis of Multi-core Software

The main challenges outlined in the acquisition program objectives are: 1. API abstraction is inadequate for complex architectures 2. Multi-core architectures are non-portable across hardware/OS 3. Testing and validation of concurrent code are complex and hard to model We already have a product called zNet, which is on the market and solves most of the above challenges. 1. zNet uses a Programming Model of a higher abstraction level to overcome multi-threading and parallel coding challenges and to support platform independence. This solves issues #1 & #2, above. 2. zNet partially addresses the challenge #3 listed above. zNet is capable of introspective processes, which look inside concurrent execution logic of applications components, allowing detection of data flow congestion points and provides hints on scalability improvements. zNet still lacks a formal analysis methodology that would specify efficiency constraints for concurrent systems and allow formal modeling of such systems prior to coding. To address this need, we propose to build and validate a simulation tool that will allow the user to test and validate our programming methodology in numerous environments. This tool will imitate large sets of multi- and distributed-cores on a desktop and apply various load constraints to mimic a real-time load.