在单指令流单数据流结构中，有单一的指令周期，在执行前，单个处理机按序取操作数。
In an SISD architecture there is a single instruction cycle; operands are fetched serially into a single processing unit before execution.
设计一个能够在一个时钟周期执行一条指令的简单指令系统才是更有效的。
It is more efficient to design a simple instruction set that enable the execution of one instruction per clock cycle.
CUDA的单指令、多线程（SIMT）的执行模型，很适合大型数据上并行执行相同的操作；
The execute model of single instruction, multiple threads (SIMT) of CUDA is very suitable for parallel to execute the same operations for large-scale data;
CUDA的单指令、多线程（SIMT）的执行模型，很适合大型数据上并行执行相同的操作；
The execute model of single instruction, multiple threads (SIMT) of CUDA is very suitable for parallel to execute the same operations for large-scale data;
应用推荐