SBIR Phase I: Parallel-SHIFT: A Simulation Programming Language

Meta-X的这个小型企业创新研究第一阶段项目提出设计和实现一个并行版本的Java语言，该语言将在多处理器共享内存计算机上运行。 多个并行化方法将被设计，实施和测试。ESTA，一个特殊用途的面向对象的编程语言，在加州大学伯克利分校开发的模拟语义，提供了适当的抽象层次的描述有限状态机和微分方程描述的复杂系统。 规范被编译成C程序。 当运行这些程序时，将模拟源程序中指定的设计。 由于对象模型中的细节水平，当前的并行计算仿真仅限于少量对象以获得及时的仿真结果。 因此，目前的实施方案的商业上是不可行的。 它被用于各种学术应用，包括高速公路自动化，自主潜艇，化学蒸馏塔控制和空中交通管理。可以利用并行语言的语法和语义来开发并行实现，其性能可以随着给定工作站上可用的处理器的数量而扩展。 该项目所带来的性能改进将使AQUEL-EQUAL成为汽车、航空航天、过程控制和制造业的商业上可行的建模、设计、仿真和评估工具。

This Small Business Innovation Research Phase I project from Meta-X proposes to design and implement a parallel version of the SHIFT languageto be run on multi-processor shared-memory computers. Multiple parallelization approaches will be designed, implemented, and tested.SHIFT, a special purpose object-oriented programming language with simulation semantics developed at UC Berkeley, offers the proper level of abstraction for characterizing complex systems described by finite state machines and differential equations. SHIFT specifications are compiled into C programs. When run these programs simulate the design specified in the SHIFT source programs. Due to the level of detail in object models, current SHIFT simulations are restricted to a small number of objects for timely simulation results. Hence, current implementation of SHIFT is commercially not viable. SHIFT is in use in a variety of academic applications including highway automation, autonomous submarines, chemical distillation column control, and air traffic management. SHIFT language syntax and semantics can be exploited to develop a parallel implementation whose performance can scale with the number of processors available on a given workstation. The resulting performance improvements that this project offers will make Parallel-SHIFT a commercially viable modeling, design, simulation, and evaluation tool for automotive, aerospace, process control, and manufacturing industries.