英特尔x86_64处理器不仅是流水线架构，还是超标量？ [英] Are Intel x86_64 processors not only pipelined architecture, but also superscalar?
问题描述
Intel x86_64处理器不仅是流水线架构，还是超标量吗？
Are Intel x86_64 processors not only pipelined architecture, but also superscalar?
-
流水线-这两个序列并行执行（同一时钟中同一管道单元的不同阶段，例如具有4个阶段的ADD）：
Pipelining - these two sequences execute in parallel (different stages of the same pipeline-unit in the same clock, for example ADD with 4 stages):
- 舞台1->舞台2->舞台3->舞台4->什么都没有
- 什么都没有->舞台1->舞台2->舞台3->舞台4
超标量-这两个序列并行执行（可以在同一时钟中向不同的流水线单元启动两条指令，例如ADD和MUL） ：
Superscalar - these two sequences execute in parallel (two instructions can be launched to different pipeline-units in the same clock, for example ADD and MUL):
- ADD（stage1）-> ADD（stage2）-> ADD（stage3）
- MUL（stage1）-> MUL（stage2）-> MUL（stage3）
推荐答案
是的，当代的英特尔处理器都是流水线和超标量。
Yes, contemporary Intel processors are both pipelined and superscalar.
执行一条指令需要花费许多纳秒的时间。这包括从指令存储器中获取指令，对指令进行解码，获取操作数，执行任何计算，从存储器中获取数据以及写入结果。为了获得甚至基本的性能，将这些步骤分解为几个小步骤并允许多个指令同时处于管线中是很重要的。
It takes many nanoseconds to execute a single instruction. That includes fetching the instruction from the instruction memory, decoding the instruction, fetching the operands, performing any computations, fetching data from memory, and writing the results. For even basic performance it is important to break these into small steps and allow multiple instructions to be "in the pipeline" simultaneously.
同样，处理器管道会消耗大量资源（面积，功耗，设计复杂性等）。将1宽处理器变成2宽超标量处理器相对非常便宜。性能的提高远大于成本（尽管一旦您拥有4宽处理器，您将看到收益递减范围越来越大，成本成倍增长）。
Likewise, a processor pipeline consumes a lot of resources (area, power, design complexity, etc.). It is relatively very cheap to turn a 1-wide processor into a 2-wide, superscalar processor. The performance improvement is greater than the costs (although once you have a 4-wide processor, you will see diminishing returns going wider, and the costs grow exponentially).
因此，您可以期望任何现代处理器（绝对最小的处理器除外）都是超标量的。经验法则-如果连接到内存层次结构，它将是超标量。那是因为内存层次结构消耗了大量的能量和能量，因此您需要超标量处理器来使其繁忙。
Thus, you can expect any modern processor (except the absolutely smallest of the small) to be superscalar. Rule of thumb - if it's connected to a memory hierarchy, it will be superscalar. That's because the memory hierarchy consumes so much power and energy that you need a superscalar processor to keep it busy.
这篇关于英特尔x86_64处理器不仅是流水线架构，还是超标量？的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！
