derek/master-thesis
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Minor fixes and compiler-based approach in colcusion

Browse Source
This commit is contained in:
Derek Christ
2024-02-19 17:12:44 +01:00
parent e597305a58
commit 2e6187c25a
3 changed files with 24 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
src/chapters/conclusion.tex
View File

@@ -19,6 +19,10 @@ For a better evaluation of the performance gains of \aca{fimdram}, it should be
Effects such as the initialization overhead of \aca{fimdram} can only be evaluated in such an environment. Effects such as the initialization overhead of \aca{fimdram} can only be evaluated in such an environment.
Furthermore, the integration of \aca{fimdram} should be extended to \acp{gpu} or \acp{tpu}, so that the comparison can be extended to the deployment of the real \ac{dnn} applications. Furthermore, the integration of \aca{fimdram} should be extended to \acp{gpu} or \acp{tpu}, so that the comparison can be extended to the deployment of the real \ac{dnn} applications.
Further research could also investigate whether the library-based approach of leveraging \ac{pim} could be replaced by a compiler-based approach.
A special compiler extension would be able to generate the necessary \ac{ld} and \ac{st} instructions by analyzing the data types of the operands.
This extension might also make use of so-called non-temporal instructions that bypass the cache hierarchy on a per-instruction basis.
In conclusion, \ac{pim} is a promising approach to address the future processing needs of \ac{ai} and possibly other applications. In conclusion, \ac{pim} is a promising approach to address the future processing needs of \ac{ai} and possibly other applications.
Not only the architecture itself has to be considered, but also the integration of \ac{pim} into the applications at the software level. Not only the architecture itself has to be considered, but also the integration of \ac{pim} into the applications at the software level.
By overcoming these challenges, \ac{pim} could be part of the solution to increase the performance and energy efficiency of future computing platforms. By overcoming these challenges, \ac{pim} could be part of the solution to increase the performance and energy efficiency of future computing platforms.

4
src/chapters/pim.tex
View File

@@ -352,7 +352,7 @@ An example with a weight matrix of dimensions (128$\times$8), an input vector of
With the processing unit \textit{i}, the number of iterations \textit{j}, the input vector \textit{a} and the weight matrix \textit{w}, the partial sum $psum[i,0:15]$ is calculated as described in \cref{eq:partial_sum}: With the processing unit \textit{i}, the number of iterations \textit{j}, the input vector \textit{a} and the weight matrix \textit{w}, the partial sum $psum[i,0:15]$ is calculated as described in \cref{eq:partial_sum}:
\begin{equation} \begin{equation}
psum[i,0:15]=\sum_{j=0}^{8}(a[j \cdot 16:j \cdot 16+15] \cdot w[i,j \cdot 16:j \cdot 16+15]) psum[i,0:15]=\sum_{j=0}^{7}(a[j \cdot 16:j \cdot 16+15] \cdot w[i,j \cdot 16:j \cdot 16+15])
\label{eq:partial_sum} \label{eq:partial_sum}
\end{equation} \end{equation}
@@ -365,7 +365,7 @@ The operation of this concrete \ac{gemv} microkernel is illustrated in \cref{img
\begin{figure} \begin{figure}
\centering \centering
\includegraphics[width=0.8\linewidth]{images/memory_layout} \includegraphics[width=0.8\linewidth]{images/memory_layout}
\caption[Procedure to perform a (128)$\times$(128$\times$8) \ac{gemv} operation]{Procedure to perform a (128)$\times$(128$\times$8) \ac{gemv} operation. One cell represents 16 \ac{fp16} elements forming a $\qty{32}{\byte}$ block \cite{kang2022}.} \caption[Procedure to perform a (128$\times$8)$\times$(128) \ac{gemv} operation]{Procedure to perform a (128$\times$8)$\times$(128) \ac{gemv} operation. One cell represents 16 \ac{fp16} elements, forming a $\qty{32}{\byte}$ block \cite{kang2022}.}
\label{img:memory_layout} \label{img:memory_layout}
\end{figure} \end{figure}

30
src/images/input_vector.tex
View File

@@ -1,19 +1,25 @@
\begin{tikzpicture} \begin{tikzpicture}
\tiny \tiny
\node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=TealBlue!30] (inputchunk0) {a[0:15]}; \definecolor{_darkblue}{RGB}{68, 114, 196}
\node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=TealBlue!30,right=0 of inputchunk0] (inputchunk1) {a[0:15]}; \definecolor{_blue}{RGB}{91, 155, 213}
\node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=RoyalBlue!30,right=0 of inputchunk1] (inputchunk2) {a[16:31]}; \definecolor{_green}{RGB}{112, 173, 71}
\node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=RoyalBlue!30,right=0 of inputchunk2] (inputchunk3) {a[16:31]}; \definecolor{_orange}{RGB}{237, 125, 49}
\node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=Blue!30,right=0 of inputchunk3] (inputchunk4) {a[32:47]}; \definecolor{_yellow}{RGB}{255, 192, 0}
\node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=Blue!30,right=0 of inputchunk4] (inputchunk5) {a[32:47]};
\node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=Green!30,below=0 of inputchunk0] {Bank 0}; \node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=_orange] (inputchunk0) {a[0:15]};
\node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=SpringGreen!30,below=0 of inputchunk1] {Bank 1}; \node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=_orange,right=0 of inputchunk0] (inputchunk1) {a[0:15]};
\node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=Green!30,below=0 of inputchunk2] {Bank 0}; \node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=_yellow,right=0 of inputchunk1] (inputchunk2) {a[16:31]};
\node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=SpringGreen!30,below=0 of inputchunk3] {Bank 1}; \node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=_yellow,right=0 of inputchunk2] (inputchunk3) {a[16:31]};
\node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=Green!30,below=0 of inputchunk4] {Bank 0}; \node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=_green, right=0 of inputchunk3] (inputchunk4) {a[32:47]};
\node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=SpringGreen!30,below=0 of inputchunk5] {Bank 1}; \node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=_green, right=0 of inputchunk4] (inputchunk5) {a[32:47]};
\node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=_darkblue!80,below=0 of inputchunk0] {Bank 0};
\node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=_blue!80, below=0 of inputchunk1] {Bank 1};
\node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=_darkblue!80,below=0 of inputchunk2] {Bank 0};
\node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=_blue!80, below=0 of inputchunk3] {Bank 1};
\node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=_darkblue!80,below=0 of inputchunk4] {Bank 0};
\node[draw,outer sep=0,minimum width=1.5cm,minimum height=4mm,fill=_blue!80, below=0 of inputchunk5] {Bank 1};
\node[right=of inputchunk5.south east,anchor=east] (inputchunk6) {\normalsize\dots}; \node[right=of inputchunk5.south east,anchor=east] (inputchunk6) {\normalsize\dots};