Distributed Hierarchical Digital Simulation Zhang Zhengyu Professor Richard
Distributed Hierarchical Digital Simulation Zhang Zhengyu Professor Richard Newton
Project Goals n n Prototype the next generation “component” based electronic design system using pre-built design objects with embedded simulation capability Demonstrate the feasibility of and evaluate the usability of such a system
Overview n Traditional approach: • coding (at behavior or RTL level) debugging via simulation coding again … • availability of simulator (expensive) an onerous necessity • awkward for collaboration • IP reuse made difficult
Overview (Cont’d) n Approach based on the WELD paradigm • build-once, use-many-times design objects • hierarchical construction of top level design • Plug and Run: flexible design revision and verification • collaboration poised due to its distributed nature • IP reuse
Description n Base component • possess the behavior corresponding to the function modules at the bottom level of the abstraction hierarchy n High level component • wrap components at the lower level and regard their collective behavior as that of a single entity n Distributed Asynchronous Scheduler • have the local knowledge to orchestrate the interactions between various sub-components
Description (Cont’d) n The overall formation • divide and conquer: input vectors along with scheduling messages propagate through the hierarchy, get evaluated at the base level then merge back up to form output • resemble many other distributed asynchronous scheduling system: an extensively studied problem
Current Status singled out Sun’s JDK 1. 1 RMI as the API of the choice n outlined a detailed implementation at pseudo-code level n literature survey of relevant technology. e. g CORBA and DCOM n implemented a primitive version of the system based on hand-coded components n
Future Work A web enabled, user interactive and GUI augmented version n Performance enhancement n
- Slides: 8