Simulation Model of IBM-Watson Intelligent System for Early Software Development

IBM-Watson is among the leading intelligent systems available today. It is extremely complex, employing multiple processors, peripherals, interconnects, along with specialized software and applications. Developing software for this architecture, in absence of target platform, is an extremely error-prone affair. Bringup of software once the hardware is available detects large amount of bugs, throwing the project cost and schedule out of control. This paper introduces a methodology based on IBM-Watson system-level simulation, developed using high-level simulation models of all the components of the target architecture. This methodology helps to debug, verify and fine-tune the IBM-Watson software much before the availability of the target hardware. Use of this methodology enables detecting the bugs much earlier in the development cycle. Majority of the defects are removed much earlier and software bringup-time on actual hardware has reduced from months to days!