提出了一种新的ＩＬＰ处理器体系结构——前瞻性多线程体系结构，简称ＳＭＡ．它结合了前瞻性执行机制和多线程执行机制，以整个线程为步长进行前瞻性执行，多个线程并行执行并且共享处理器硬件资源．这样，处理器既通过组合每个线程的指令窗口形成一个大的动态指令窗口，开发出程序中更大的ＩＬＰ，又利用多线程执行机制屏蔽各种长延迟操作，达到较高的资源利用率；介绍了ＳＭＡ执行模型，并讨论了ＳＭＡ处理器的实现和其中的关键技术．这些关键技术包括：线程控制、分布式寄存器重命名、指令发射执行、中断处理和处理器对编译优化的要求．性能分析结果表明ＳＭＡ处理器可以适应各种不同类型的程序，以达到高性能，如果采用较好的编译支持和线程预测算法，性能将进一步提高 Proposed a new ILP processor architecture - forward-looking multi-threaded architecture, referred to as SMA. It combines a forward-looking execution mechanism with a multithreaded execution mechanism, proactively executing the entire thread in steps, multiple threads executing in parallel, and sharing processor hardware resources. In this way, the processor not only forms a large dynamic instruction window by combining the instruction windows of each thread, develops a larger ILP in the program, but also shields various long-delay operations by using a multi-thread execution mechanism to achieve high resource utilization ; Introduced the SMA execution model, and discussed the implementation of the SMA processor and the key technologies therein. These key technologies include: Thread Control, Distributed Register Rename, Instruction Launch Execution, Interrupt Handling, and Processor Compilation Optimization Requirements. Performance analysis results show that the SMA processor can adapt to a variety of different types of programs to achieve high performance, performance will be further enhanced if better compiler support and thread prediction algorithms are used