From 407848ada7f1ac35451cbbdd9492e965a462d6db Mon Sep 17 00:00:00 2001
From: Derek Christ <christ.derek@gmail.com>
Date: Mon, 25 Mar 2024 16:37:22 +0100
Subject: [PATCH] Add wallclock-time plots

---
 plots/matrix_infinite.tex |  2 +-
 plots/vector_infinite.tex |  2 +-
 plots/wallclock_time.csv  |  3 +++
 plots/wallclock_time.tex  | 40 +++++++++++++++++++++++++++++++++++++++
 samplepaper.tex           | 11 +++++++++++
 5 files changed, 56 insertions(+), 2 deletions(-)
 create mode 100644 plots/wallclock_time.csv
 create mode 100644 plots/wallclock_time.tex

diff --git a/plots/matrix_infinite.tex b/plots/matrix_infinite.tex
index 915acd6..80eef81 100644
--- a/plots/matrix_infinite.tex
+++ b/plots/matrix_infinite.tex
@@ -8,7 +8,7 @@
             ymin=0,
             ymax=20,
             ymajorgrids,
-            ylabel={Relative Performance},
+            ylabel={Speedup},
             tick pos=left,
             xtick=data,
             xticklabels from table={\csv}{level},
diff --git a/plots/vector_infinite.tex b/plots/vector_infinite.tex
index c961ce2..12888c5 100644
--- a/plots/vector_infinite.tex
+++ b/plots/vector_infinite.tex
@@ -8,7 +8,7 @@
             ymin=0,
             ymax=20,
             ymajorgrids,
-            ylabel={Relative Performance},
+            ylabel={Speedup},
             tick pos=left,
             xtick=data,
             xticklabels from table={\csv}{level},
diff --git a/plots/wallclock_time.csv b/plots/wallclock_time.csv
new file mode 100644
index 0000000..bcbd391
--- /dev/null
+++ b/plots/wallclock_time.csv
@@ -0,0 +1,3 @@
+system,haxpy,dnn,gemv,vmul,vadd
+PIM-HBM,135.58,49.17,256.96,128.43,136.81
+HBM,3084.85,941.57,6788.0,1701.99,2136.37
diff --git a/plots/wallclock_time.tex b/plots/wallclock_time.tex
new file mode 100644
index 0000000..efceaf8
--- /dev/null
+++ b/plots/wallclock_time.tex
@@ -0,0 +1,40 @@
+\begin{tikzpicture}
+    \pgfplotstableread[col sep=comma]{plots/wallclock_time.csv}\csv
+    \begin{axis}[
+            width=10cm,
+            height=4cm,
+            ybar=1pt,
+            bar width = 5pt,
+            ymin=0,
+            ymax=1e4,
+			ymode=log,
+            ymajorgrids,
+            ylabel={Runtime [s]},
+            tick pos=left,
+            xtick=data,
+            xticklabels from table={\csv}{system},
+            enlarge x limits=0.5,
+            legend style={
+                at={(current bounding box.south-|current axis.south)},
+                anchor=north,
+                legend columns=-1,
+                draw=none,
+                /tikz/every even column/.append style={column sep=0.5cm}
+            },
+        ]
+        \addplot[fill=_darkblue!90] table [x expr=\coordindex, y={vadd}]{\csv};
+        \addlegendentry{VADD}
+
+        \addplot[fill=_blue!90] table [x expr=\coordindex, y={vmul}]{\csv};
+        \addlegendentry{VMUL}
+
+        \addplot[fill=_green!90] table [x expr=\coordindex, y={haxpy}]{\csv};
+        \addlegendentry{HAXPY}
+
+        \addplot[fill=_orange!90] table [x expr=\coordindex, y={gemv}]{\csv};
+        \addlegendentry{GEMV}
+
+        \addplot[fill=yellow!90] table [x expr=\coordindex, y={dnn}]{\csv};
+        \addlegendentry{DNN}
+    \end{axis}
+\end{tikzpicture}
diff --git a/samplepaper.tex b/samplepaper.tex
index 0e3a0df..71089e5 100644
--- a/samplepaper.tex
+++ b/samplepaper.tex
@@ -351,6 +351,17 @@ However, this memory barrier has also been implemented in our VADD kernel, which
 
 The \ac{gemv} microbenchmark on the other hand shows a more matching result with an average speedup value of $\qty{8.3}{\times}$ for Samsung's real system and \qty{2.6}{\times} for their virtual prototype, while this paper achieved an average speedup of $\qty{9.0}{\times}$, which is well within the reach of the real hardware implementation.
 
+\begin{figure}
+    \centering
+    \input{plots/wallclock_time}
+    \caption{Runtimes of the simulation workloads on the host system.}
+    \label{fig:wallclock_time}
+\end{figure}
+
+\Cref{fig:wallclock_time} shows the simulation runtimes of the various workloads on the host system.
+With \ac{pim} enabled, the runtime drops by about an order of magnitude for some workloads, indicating the reduced simulation effort on gem5's complex processor model, as only new memory requests are issued by the model during operation of \ac{pim}.
+Therefore, exploring the effectiveness of different \ac{pim}-enabled workloads may be less time-consuming due to the reduced simulation complexity.
+
 \section{Conclusion}
 % TODO Lukas/Matthias
 %