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

Use interface power again

Browse Source
This commit is contained in:
Derek Christ
2024-11-14 16:26:45 +01:00
parent 62331510a7
commit d47c8d8f7c
6 changed files with 3861 additions and 3863 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2570
data/hynix.csv
View File

File diff suppressed because it is too large Load Diff

2570
data/micron.csv
View File

File diff suppressed because it is too large Load Diff

2570
data/samsung.csv
View File

File diff suppressed because it is too large Load Diff

4
drampower-main.tex
View File

@@ -1233,14 +1233,14 @@ As it was not possible to reproduce the usual $I_{DD}0$ pattern of ACT-PRE for t
The initial simulations are based on the current values specified in the datasheet of the specific vendor. The initial simulations are based on the current values specified in the datasheet of the specific vendor.
Then, based on the actual measurements, the current values are reapplied to a second simulation. Then, based on the actual measurements, the current values are reapplied to a second simulation.
The results are shown in Figure~\ref{fig:power_plot}. The results are shown in Figure~\ref{fig:power_plot}.
%\begin{figure} % \begin{figure}
% \centering % \centering
% \resizebox{\linewidth}{!}{% % \resizebox{\linewidth}{!}{%
% \input{img/power_plot} % \input{img/power_plot}
% } % }
% \caption{Average Power Consumption of Simulations and Measurements for Different Vendors} % \caption{Average Power Consumption of Simulations and Measurements for Different Vendors}
% \label{fig:power_plot} % \label{fig:power_plot}
%\end{figure} % \end{figure}
As it can be seen, the $I_{DD}$ currents in the datasheet are overly pessimistic for all vendors: As it can be seen, the $I_{DD}$ currents in the datasheet are overly pessimistic for all vendors:
The simulations based on the datasheets show on average a $4.8\times$ higher power consumption than the actual power measurements. The simulations based on the datasheets show on average a $4.8\times$ higher power consumption than the actual power measurements.
However, when the measured currents are applied to the simulation, there is still a small discrepancy: However, when the measured currents are applied to the simulation, there is still a small discrepancy:

4
drampower.bib
View File

@@ -193,9 +193,7 @@ series = {ASPLOS '18}
publisher = {{ACM}}, publisher = {{ACM}},
author = {Feldmann, Johannes and Steiner, Lukas and Christ, Derek and Psota, Thomas and Jung, Matthias and Wehn, Norbert}, author = {Feldmann, Johannes and Steiner, Lukas and Christ, Derek and Psota, Thomas and Jung, Matthias and Wehn, Norbert},
urldate = {2024-11-14}, urldate = {2024-11-14},
date = {2023-10-02}, date = {2023-10-02}
langid = {english},
file = {Feldmann et al. - 2023 - A Precise Measurement Platform for LPDDR4 Memories.pdf:/home/derek/.local/share/zotero/storage/C7RSPK9K/Feldmann et al. - 2023 - A Precise Measurement Platform for LPDDR4 Memories.pdf:application/pdf},
} }
@inproceedings{vog_10, @inproceedings{vog_10,

6
img/power_plot.tex
View File

@@ -45,7 +45,7 @@
\addplot[color=green!50] table [x={time}, y={datasheet}]{\hynix}; \addplot[color=green!50] table [x={time}, y={datasheet}]{\hynix};
\addplot[color=blue!50] table [x={time}, y={measured}]{\hynix}; \addplot[color=blue!50] table [x={time}, y={measured}]{\hynix};
\addplot[color=red!50] table [x={time}, y={optimized}]{\hynix}; \addplot[color=red!50] table [x={time}, y={optimized}]{\hynix};
\addplot[color=brown!50] table [x={time}, y={optimized_no_if}]{\hynix}; % \addplot[color=red!50] table [x={time}, y={optimized_no_if}]{\hynix};
\nextgroupplot[ \nextgroupplot[
title=Vendor 2, title=Vendor 2,
@@ -67,7 +67,7 @@
\addplot[color=green!50] table [x={time}, y={datasheet}]{\micron}; \addplot[color=green!50] table [x={time}, y={datasheet}]{\micron};
\addplot[color=blue!50] table [x={time}, y={measured}]{\micron}; \addplot[color=blue!50] table [x={time}, y={measured}]{\micron};
\addplot[color=red!50] table [x={time}, y={optimized}]{\micron}; \addplot[color=red!50] table [x={time}, y={optimized}]{\micron};
\addplot[color=brown!50] table [x={time}, y={optimized_no_if}]{\micron}; % \addplot[color=red!50] table [x={time}, y={optimized_no_if}]{\micron};
\nextgroupplot[ \nextgroupplot[
title=Vendor 3, title=Vendor 3,
@@ -89,7 +89,7 @@
\addplot[color=green!50] table [x={time}, y={datasheet}]{\samsung}; \addplot[color=green!50] table [x={time}, y={datasheet}]{\samsung};
\addplot[color=blue!50] table [x={time}, y={measured}]{\samsung}; \addplot[color=blue!50] table [x={time}, y={measured}]{\samsung};
\addplot[color=red!50] table [x={time}, y={optimized}]{\samsung}; \addplot[color=red!50] table [x={time}, y={optimized}]{\samsung};
\addplot[color=brown!50] table [x={time}, y={optimized_no_if}]{\samsung}; % \addplot[color=red!50] table [x={time}, y={optimized_no_if}]{\samsung};
\end{groupplot} \end{groupplot}
% Legend % Legend