Skip to content
Snippets Groups Projects
Commit 52ae6106 authored by Peter Gerwinski's avatar Peter Gerwinski
Browse files

Vortragsfolien, Tafelbild und Beispiele 15.6.2021

parent fbb61174
Branches
No related tags found
No related merge requests found
Show whitespace changes
Inline Side-by-side
Showing
with 625 additions and 0 deletions
20210615/Buegeln.pdf 0 → 120000
+ 1
− 0
View file @ 52ae6106
../common/Buegeln.pdf
\ No newline at end of file
20210615/Trommeltrocknen.pdf 0 → 120000
+ 1
− 0
View file @ 52ae6106
../common/Trommeltrocknen.pdf
\ No newline at end of file
20210615/Waschen_30.pdf 0 → 120000
+ 1
− 0
View file @ 52ae6106
../common/Waschen_30.pdf
\ No newline at end of file
20210615/exploit-2.c 0 → 100644
+ 11
− 0
View file @ 52ae6106
#include <stdio.h>
#include <stdint.h>
int main (int argc, char **argv)
{
char loop[8];
uint64_t *loop_int = (uint64_t *) loop;
*loop_int = 0x555555555185;
printf ("Prof. Dr. rer. nat. Dipl.-Phys. Peter Ge%s\n", loop);
return 0;
}
20210615/exploit-2.txt 0 → 100644
+ 1
− 0
View file @ 52ae6106
Prof. Dr. rer. nat. Dipl.-Phys. Peter GeQUUUU
20210615/logo-hochschule-bochum-cvh-text-v2.pdf 0 → 120000
+ 1
− 0
View file @ 52ae6106
../common/logo-hochschule-bochum-cvh-text-v2.pdf
\ No newline at end of file
20210615/logo-hochschule-bochum.pdf 0 → 120000
+ 1
− 0
View file @ 52ae6106
../common/logo-hochschule-bochum.pdf
\ No newline at end of file
20210615/pgslides.sty 0 → 120000
+ 1
− 0
View file @ 52ae6106
../common/pgslides.sty
\ No newline at end of file
20210615/rtech-20210615.pdf 0 → 100644
+ 0
− 0
View file @ 52ae6106
File added
20210615/rtech-20210615.tex 0 → 100644
+ 593
− 0
View file @ 52ae6106
% rtech-20210615.pdf - Lecture Slides on Computer Technology
% Copyright (C) 2012, 2013, 2014, 2021 Peter Gerwinski
%
% This document is free software: you can redistribute it and/or
% modify it either under the terms of the Creative Commons
% Attribution-ShareAlike 3.0 License, or under the terms of the
% GNU General Public License as published by the Free Software
% Foundation, either version 3 of the License, or (at your option)
% any later version.
%
% This document is distributed in the hope that it will be useful,
% but WITHOUT ANY WARRANTY; without even the implied warranty of
% MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
% GNU General Public License for more details.
%
% You should have received a copy of the GNU General Public License
% along with this document. If not, see <http://www.gnu.org/licenses/>.
%
% You should have received a copy of the Creative Commons
% Attribution-ShareAlike 3.0 Unported License along with this
% document. If not, see <http://creativecommons.org/licenses/>.
% README: Puffer-Überläufe, Pipelining
\documentclass[10pt,t]{beamer}
\usepackage{pgslides}
\usepackage{rotating}
\lstdefinestyle{asm}{basicstyle=\color{structure},
language={},
gobble=4}
\title{Rechnertechnik}
\author{Prof.\ Dr.\ rer.\ nat.\ Peter Gerwinski}
\date{15.\ Juni 2021}
\begin{document}
\maketitleframe
\sectionnonumber{\inserttitle}
\begin{frame}
\showsectionnonumber
\begin{itemize}
\item[\textbf{1}] \textbf{Einführung}
\item[\textbf{2}] \textbf{Vom Schaltkreis zum Computer}
\item[\textbf{3}] \textbf{Architekturmerkmale von Prozessoren}
\item[\textbf{4}] \textbf{Der CPU-Stack}
\begin{itemize}
\item[4.1] Implementation
\item[4.2] Unterprogramme
\item[4.3] Register sichern
\item[4.4] Stack-Überläufe
\color{medgreen}
\item[4.5] Puffer-Überläufe
\end{itemize}
% \color{gray}
% \item[\textbf{5}] \textbf{Hardwarenahe Programmierung}
\item[\textbf{5}] \textbf{Anwender-Software}
\item[\textbf{6}] \textbf{Pipelining}
\begin{itemize}
\color{medgreen}
\item[6.1] Konzept
\item[6.2] Arithmetik-Pipelines
\color{red}
\item[6.3] Instruktions-Pipelines
\end{itemize}
% \item[\textbf{7}] \textbf{Bus-Systeme}
\item[\textbf{\dots\hspace{-0.75em}}]
% \item[\textbf{8}] \textbf{Ausblick}
\end{itemize}
\end{frame}
\setcounter{section}{4}
\section{Anwender-Software}
\setcounter{subsection}{5}
\subsection{Puffer-Überläufe}
\begin{frame}[fragile]
\showsection
\showsubsection
\begin{lstlisting}
#include <stdio.h>
int main (void)
{
int ID;
char buffer[20];
printf ("Your ID, please: ");
gets (buffer);
sscanf (buffer, "%d", &ID);
printf ("Your name, please: ");
gets (buffer);
printf ("Hello, %s!\nYour ID is %d.\n", buffer, ID);
return 0;
}
\end{lstlisting}
\begin{picture}(0,0)
\put(5,6){\makebox(0,0)[tl]{\begin{minipage}{6.7cm}
Die Funktion \lstinline{gets()} prüft nicht, ob \lstinline{buffer[]}
für den eingegebenen Namen ausreicht, und überschreibt ggf.\ die
Variable \lstinline{ID} sowie die Rücksprungadresse des Funktionsaufrufs
von \lstinline{main()}.
\end{minipage}}}
\put(6.0,2.5){\makebox(0,0)[l]{\color{red}gets() nicht verwenden!}}
\end{picture}
\vspace*{-1cm}
\end{frame}
\setcounter{section}{6}
\section{Pipelining}
\subsection{Konzept}
\newcommand{\wm}{\raisebox{-0.3cm}{\includegraphics[width=1.0cm]{Waschen_30.pdf}}}
\newcommand{\wt}{\raisebox{-0.3cm}{\includegraphics[width=1.0cm]{Trommeltrocknen.pdf}}}
\newcommand{\be}{\raisebox{-0.3cm}{\includegraphics[width=1.0cm]{Buegeln.pdf}}}
\begin{frame}
\showsection
% \pause
\showsubsection
\begin{itemize}
\item
Aufgabe in Teilaufgaben zerlegen
\item
Teilaufgaben parallel ausführen
\end{itemize}
% \pause
\bigskip
Beispiel: Wäsche waschen
% \pause
\begin{itemize}
\item
Teilaufgaben: \wm, \wt, \be
% \pause
\item
müssen nacheinander ausgeführt werden: Datenfluß
% \pause
\item
belegen jeweils 1 Ressource
\end{itemize}
\end{frame}
\begin{frame}
3 Ladungen Wäsche
\begin{center}
\begin{onlyenv}<1-2>
\begin{picture}(12,3.5)
\thicklines
\put(1,1){\wm}
\put(2,1){\wt}
\put(3,1){\be}
\put(4,2){\wm}
\put(5,2){\wt}
\put(6,2){\be}
\put(7,3){\wm}
\put(8,3){\wt}
\put(9,3){\be}
\put(-0.1,0.5){\vector(1,0){11.1}}
\put(11.5,0.5){\makebox(0,0){$t$}}
\put(10.5,1.0){\makebox(0,0)[r]{ineffizient}}
\end{picture}\\[\medskipamount]
\end{onlyenv}
\begin{onlyenv}<2-4>
\begin{picture}(12,3.5)
\thicklines
\put(1,1){\wm}
\put(2,1){\wt}
\put(3,1){\be}
\put(2,2){\wm}
\put(3,2){\wt}
\put(4,2){\be}
\put(3,3){\wm}
\put(4,3){\wt}
\put(5,3){\be}
\put(-0.1,0.5){\vector(1,0){11.1}}
\put(11.5,0.5){\makebox(0,0){$t$}}
\put(10.5,1.0){\makebox(0,0)[r]{effizient}}
\end{picture}\\[\medskipamount]
\end{onlyenv}
\begin{onlyenv}<3->
\begin{picture}(12,3.5)
\thicklines
\put(1,1){\wm}
\put(2,1){\wt}
\put(3,1){\be}
\put(1,2){\wm}
\put(2,2){\wt}
\put(3,2){\be}
\put(1,3){\wm}
\put(2,3){\wt}
\put(3,3){\be}
\put(-0.1,0.5){\vector(1,0){11.1}}
\put(11.5,0.5){\makebox(0,0){$t$}}
\put(10.5,1.0){\makebox(0,0)[r]{noch effizienter}}
\color{red}
\thicklines
\put(0.5,0.8){\only<4->{\line(4,1){10}}}
\put(0.5,3.3){\only<4->{\line(4,-1){10}}}
\put(7.5,1.7){\only<4->{\mbox{\shortstack{Ressourcen-\\konflikt}}}}
\end{picture}\\[\medskipamount]
\end{onlyenv}
\begin{onlyenv}<5->
\begin{picture}(12,3.5)
\thicklines
\put(1,1){\wm}
\put(2,1){\wt}
\put(3,1){\be}
\put(1,2){\wt}
\put(2,2){\be}
\put(3,2){\wm}
\put(1,3){\be}
\put(2,3){\wm}
\put(3,3){\wt}
\put(-0.1,0.5){\vector(1,0){11.1}}
\put(11.5,0.5){\makebox(0,0){$t$}}
\put(10.5,1.0){\makebox(0,0)[r]{noch effizienter}}
\color{red}
\thicklines
\put(0.5,0.8){\only<6->{\line(4,1){10}}}
\put(0.5,3.3){\only<6->{\line(4,-1){10}}}
\put(7.5,1.7){\only<6->{\mbox{\shortstack{Daten-\\konflikt}}}}
\end{picture}%
\end{onlyenv}
\end{center}
\end{frame}
\subsection{Arithmetik-Pipelines}
\newcommand{\insns}[1]{%
\begin{picture}(0,0)
\put(0,0){\line(1,0){2}}
\put(2,0){\line(0,-1){#1}}
\put(2,-#1){\vector(-1,0){2}}
\end{picture}}
\newcommand{\insnsup}[1]{%
\begin{picture}(0,0)
\put(0,0){\line(1,0){2}}
\put(2,0){\line(0,1){#1}}
\put(2,#1){\vector(-1,0){2}}
\end{picture}}
\begin{frame}
\showsubsection
"`Register-FIFO"'
\pause
\smallskip
Pseudo-Code: Berechnung von
\begin{displaymath}
S = \left(\begin{array}{c}a_1\\a_2\\a_3\end{array}\right)
\cdot\left(\begin{array}{c}b_1\\b_2\\b_3\end{array}\right)
= a_1 \cdot b_1
+ a_2 \cdot b_2
+ a_3 \cdot b_3
\end{displaymath}
mit einer Pipeline der Länge 3
\pause
\medskip
\begingroup
\setlength{\unitlength}{\baselineskip}%
\begin{picture}(0,0)(-5.0,-0.25)
\color{red}
\put(0,0.0){\only<4->{\insns{3}}}
\put(0.5,-1){\only<5->{\insns{2.9}}}
\put(1.0,-2){\only<6->{\insns{3}}}
\put(0.5,-4.1){\only<8->{\insns{2.8}}}
\put(0.5,-7.1){\only<11->{\insns{2.9}}}
\put(-1.5,-9.25){\only<15->{\shortstack{Daten-\\konflikt}}}
\end{picture}%
\color{structure}
push $a_1 \cdot b_1$\\
\pause
\pause
push $a_2 \cdot b_2$\\
\pause
push $a_3 \cdot b_3$\\
\pause
$s_1$ = pop\\
\pause
push $s_1 + \mbox{pop}$\\
\pause
$s_3$ = pop\\
\pause
nop\\
\pause
push $s_3 + \mbox{pop}$\\
\pause
nop\\
\pause
nop\\
\pause
$S$ = pop\\
\endgroup
\pause
\pause
\begin{picture}(0,0)
\color{red}
\put(5,3.8){\makebox(0,0)[l]{$\left.\rule{0pt}{1.2cm}\right\}$ effizient: Pipeline gefüllt}}
\put(5,1.4){\makebox(0,0)[l]{$\left.\rule{0pt}{1.2cm}\right\}$ ineffizient: "`Blasen"'}}
\end{picture}
\pause
\vspace*{-1.25cm}
\begin{flushright}
\dots\ aber immer noch effizienter als\\
3 Schritte für jede Operation
\end{flushright}
\end{frame}
\subsectionnonumber{Reales Beispiel: \break Vektor-Addition auf i860}
\begin{frame}[fragile]
\begin{minipage}[t]{5cm}
\raggedright
\showsubsectionnonumber
\begin{lstlisting}
.align 8
.globl _vadd
nop
_vadd:
shr 1,r19,r19
bte r19,r0,exitadd
addu 0x000F,r16,r16
andnot 0x000F,r16,r16
adds -16,r16,r16
addu 0x000F,r17,r17
andnot 0x000F,r17,r17
adds -16,r17,r17
addu 0x000F,r18,r18
andnot 0x000F,r18,r18
adds -16,r18,r18
mov -1,r20
\end{lstlisting}
\end{minipage}
\begin{minipage}[t]{5cm}
\begin{lstlisting}
fld.q 16(r16)++,f16
fld.q 16(r17)++,f20
pfadd.dd f16,f20,f0
bla r20,r19,loopadd
pfadd.dd f18,f22,f0
loopadd:
d.pfadd.dd f0,f0,f0
fld.q 16(r16)++,f16
d.pfadd.dd f0,f0,f24
fld.q 16(r17)++,f20
d.pfadd.dd f16,f20,f26
bla r20,r19,loopadd
d.pfadd.dd f18,f22,f0
fst.q f24,16(r18)++
nop
nop
nop
exitadd:
bri r1
nop
\end{lstlisting}
\end{minipage}
\end{frame}
\lstdefinestyle{shybubble}{basicstyle=\color{lightgray},
moredelim=**[is][\color{structure}]{¡}{¿},
moredelim=**[is][\only<7->{\color{red}}]{°}{¿}}
\begin{frame}[fragile]
\begin{minipage}[t]{5cm}
\raggedright
\showsubsectionnonumber
\begin{lstlisting}[style=shybubble]
.align 8
.globl _vadd
nop
_vadd:
shr 1,r19,r19
bte r19,r0,exitadd
addu 0x000F,r16,r16
andnot 0x000F,r16,r16
adds -16,r16,r16
addu 0x000F,r17,r17
andnot 0x000F,r17,r17
adds -16,r17,r17
addu 0x000F,r18,r18
andnot 0x000F,r18,r18
adds -16,r18,r18
mov -1,r20
\end{lstlisting}
\end{minipage}
\begin{minipage}[t]{5cm}
\begin{lstlisting}[style=shybubble]
¡fld.q 16(r16)++,f16
fld.q 16(r17)++,f20
pfadd.dd f16,f20,f0
bla r20,r19,loopadd
pfadd.dd f18,f22,f0
loopadd:
d.pfadd.dd °f0,f0,f0¿
fld.q 16(r16)++,f16
d.pfadd.dd °f0,f0¿,f24
fld.q 16(r17)++,f20
d.pfadd.dd f16,f20,f26
bla r20,r19,loopadd
d.pfadd.dd f18,f22,°f0¿
fst.q f24,16(r18)++¿
nop
nop
nop
exitadd:
bri r1
nop
\end{lstlisting}
\end{minipage}
\pause
\begingroup
\setlength{\unitlength}{\baselineskip}%
\begin{picture}(0,0)(-21.0,-17.6)
\color{red}
\put(0,0){\only<2->{\insns{5.9}}}
\put(1.0,-2){\only<3->{\insns{6.0}}}
\put(-7,-9){\only<4->{\line(-1,0){1}}}
\put(-8,-9){\only<4->{\line(0,-1){2}}}
\put(-8,-11){\only<4->{\line(-1,0){1}}}
\put(-9,-11){\only<4->{\line(0,1){8}}}
\put(-9,-3){\only<4->{\vector(1,0){0.7}}}
\put(0.8,-7.8){\only<5->{\insnsup{1.7}}}
\put(0.8,-10){\only<6->{\insnsup{1.8}}}
\put(-0.5,-12.5){\only<7->{\mbox{6mal f0 = 2 Blasen}}}
\end{picture}%
\endgroup
\vspace{-2cm}
\begin{flushright}
\only<8->{Immerhin: 2 Additionen in 4 Taktzyklen}
\medskip
\only<9->{Dies ist ein \emph{einfaches} Beispiel.}
\end{flushright}
\end{frame}
\subsection{Instruktions-Pipelines}
\begin{frame}[fragile]
\showsubsection
Ein Prozessor benötigt Zeit, um einen Befehl zu verstehen.
\textarrow\ Während Befehlsausführung nächste Befehle vorauslesen
\medskip
\begin{lstlisting}[style=shy]
².L3:
movw r30,r20¿
³add r30,r18¿
°adc r31,r19¿
¤mov r24,r18
subi r24,lo8(-(1))
st Z,r24
subi r18,lo8(-(1))
sbci r19,hi8(-(1))
cp r22,r18
cpc r23,r19¿
¢brge .L3¿
æret¿
\end{lstlisting}
\begin{picture}(0,0)
\color{red}
\put(0.5,1.00){\only<8->{\line(-1,0){0.4}}}
\put(0.1,1.00){\only<8->{\line(0,1){4.15}}}
\put(0.1,5.15){\only<8->{\vector(1,0){0.4}}}
\put(0.5,0.90){\only<9->{\line(-1,0){0.4}}}
\put(0.1,0.90){\only<9->{\line(0,-1){0.35}}}
\put(0.1,0.55){\only<9->{\vector(1,0){0.4}}}
\put(2.5,0.825){\only<10->{\begin{minipage}[t]{8cm}
bedingter Sprung:
Welche Befehle vorauslesen?\\[\smallskipamount]
\only<11->{Kontrollflußkonflikt}
\end{minipage}}}
\end{picture}
\begin{flushright}
\vspace*{-0.5cm}
\only<12->{Lösungsansatz: Zweigvorhersage}
\vspace*{-1cm}
\end{flushright}
\end{frame}
\subsubsectionnonumber{Zweigvorhersage -- Branch Prediction}
\begin{frame}[fragile]
\showsubsection
\showsubsubsectionnonumber
\begin{itemize}
\setlength{\unitlength}{\baselineskip}%
\pause
\item
Sprünge nach oben sind Schleifen: "`Ja"'\\
Sprünge nach unten sind Auswahl-Verzweigungen: "`Nein"'
\pause
\item
Delayed Branches: Sprungbefehl verspätet ausführen\\
\textarrow\ Optimierung manuell oder durch Compiler
\pause
\medskip
\begin{lstlisting}
loopadd:
d.pfadd.dd f0,f0,f0
fld.q 16(r16)++,f16
d.pfadd.dd f0,f0,f24
fld.q 16(r17)++,f20
d.pfadd.dd f16,f20,f26
bla r20,r19,loopadd
d.pfadd.dd f18,f22,f0
fst.q f24,16(r18)++
\end{lstlisting}
\begin{picture}(0,0)(-9.6,-12.2)
\color{red}
\put(-7,-9){\line(-1,0){1}}
\put(-8,-9){\line(0,-1){2}}
\put(-8,-11){\line(-1,0){1}}
\put(-9,-11){\line(0,1){8}}
\put(-9,-3){\vector(1,0){0.7}}
\end{picture}
\pause
\item
Branch History Table: Sprünge merken
\pause
\item
\dots
\end{itemize}
\end{frame}
\subsectionnonumber{Zusammenfassung}
\begin{frame}
\showsection
\showsubsectionnonumber
\begin{itemize}
\item
Teilaufgaben parallel ausführen
\item
Arithmetik-Pipelines führen Berechnungen parallel aus,\\
Instruktions-Pipelines lesen Befehle voraus
\item
Ressourcen-, Daten- und Kontrollflußkonflikte führen zu "`Blasen"'
\item
Zweigvorhersage reduziert Kontrollflußkonflikte in Instruktions-Pipelines
\begin{itemize}
\item
nach oben / nach unten
\item
Delayed Branches: manuell optimieren
\item
Branch History Table: Sprünge merken
\end{itemize}
\end{itemize}
\end{frame}
\end{document}
20210615/server-2.c 0 → 100644
+ 14
− 0
View file @ 52ae6106
#include <stdio.h>
#include <string.h>
int main (void)
{
char buffer[20];
printf ("Your name, please: ");
gets (buffer);
if (strcmp (buffer, "amerkel") == 0)
printf ("Hello, %s! Here is the secret code:\n13 23 42 137 1117\n", buffer);
else
printf ("Hello, %s! You have restricted guest access.\n", buffer);
return 0;
}
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Please register or to comment

Impressum Datenschutzerklärung