☰

重现 CPU 图

tag-icon • tag-icon tikz-pgf
重现 CPU 图

我正在尝试重现一个简化的英特尔处理器图，没有任何好运气。

我不明白如何使 L3 部分变得通用，并且我的代码产生了以下内容：

徒劳的努力

我想要实现的是 L3 的大小等于图表的宽度（不包括 RAM）：如果我再添加两个核心，它应该会相应地拉伸。

这是我的最小可编译代码，这里我遗漏了什么？

谢谢！

\documentclass{article}

\usepackage{tikz}
\usepackage{subfigure}
\usetikzlibrary{calc,trees,positioning,arrows,chains,shapes.geometric,%
    decorations.pathreplacing,decorations.pathmorphing,shapes,%
    matrix,shapes.symbols}

\tikzset
{
    % nodes
    darkstyle/.style =
    {
        circle,draw,fill = gray!20
    },
    % trajectory
    serpent/.style =
    {
        line join  = round,
        line width = 2pt,
        line cap   = round,
        opacity    = .7,
        red
    },
    >=stealth',
    core/.style = 
    {
        rectangle, 
        rounded corners, 
        draw           = black, thick,
        text width     = 5em, 
        minimum height = 3em, 
        text centered, 
        on chain
    },
    cache/.style = 
    {
        rectangle, 
        rounded corners, 
        draw           = gray!40, thin,
        text width     = 5em, 
        minimum height = 3em, 
        text centered, 
        on chain
    },
    line/.style =
    {
        draw, thick, <-
    },
    element/.style =
    {
        tape,
        top color      = white,
        bottom color   = blue!50!black!60!,
        minimum width  = 8em,
        draw           = blue!40!black!90, very thick,
        text width     = 10em, 
        minimum height = 3.5em, 
        text centered, 
        on chain
    },
    every join/.style =
    {
        ->, thick,shorten >=1pt
    },
    decoration = 
    {
        brace
    },
    tuborg/.style = 
    {
        decorate
    },
    tubnode/.style = 
    {
        midway, right = 2pt
    },
}

\begin{document}

\begin{figure}
    \centering
    \begin{tikzpicture}[node distance = .8cm, start chain = going below]
        % cores
        \node [core]  (c1) {Core 1};
        \node [core]  (c2) [right = of c1] {Core 2};
        % caches
        \node [cache] (l11) [below = of c1]  {L1 \\  64 KB} edge [-] (c1);
        \node [cache] (l12) [below = of c2]  {L1 \\  64 KB} edge [-] (c2);
        \node [cache] (l21) [below = of l11] {L2 \\ 256 KB} edge [-] (l11);
        \node [cache] (l22) [below = of l12] {L2 \\ 256 KB} edge [-] (l12);
        % common cache
        \node [core,text width = 10em,xshift=-20.5pt]  (l3)  {L3 \\   8 MB} edge [-] (l21) edge [-] (l22);
        % ram
        \node [core]  (ram)  [right = of l3] {RAM} edge [-] (l3);
    \end{tikzpicture}
    \label{fig:cache}
    \caption{Cache hierarchy diagram for the 2016 Intel Kaby Lake Microarchitechture.}
\end{figure}

\end{document}

答案1

![在此处输入图片描述

\documentclass[tikz, margin=3mm]{standalone}
\usetikzlibrary{arrows, calc, chains, positioning}

\tikzset
{
 core/.style = {
        rectangle,
        rounded corners,
        draw           = black, thick,
        text width     = 5em, 
        minimum height = 3em,
        align=center,
        on chain
                },
cache/.style = {
        core,
        draw           = gray
                }
}% end of tikzset

\begin{document}
    \begin{tikzpicture}[
node distance = 8mm, 
start chain = going below]
% left branch
\node [core]                (c1)  {Core 1};
\node [cache,join]          (l11) {L1 \\  64 KB}; 
\node [cache,join]          (l12) {L2 \\ 256 KB};
% right branch
\node [core,right=of c1]    (c2)  {Core 1};
\node [cache,join]          (l21) {L1 \\  64 KB};
\node [cache,join]          (l22) {L2 \\ 256 KB};
% common cache
\path   let \p1 = ($(c2.east)-(c1.west)$),
            \n1 = {veclen(\x1,\y1)} in
        node [core,minimum width = \n1,
              below=of $(l12.south)!0.5!(l22.south)$]  (l3)  {L3 \\   8 MB};
        \draw   (l12) -- (l12 |- l3.north)   (l22) -- (l22 |- l3.north);
% ram
\node [core,join]  (ram)  [right = of l3] {RAM};
    \end{tikzpicture}
\end{document}

无关： 包subfigure已过时，它已被替换subfig，但你可以通过 实现对子图形的更好控制subcaption。在上面的例子中没有使用图形环境，但将图像代码放在其中（并使用article文档类）应该不是大问题 :-)

对于 L3 块的宽度，使用calc带宏的库veclen。还利用chain库来定位垂直分支。

相关内容