76 lines
2.7 KiB
TeX
76 lines
2.7 KiB
TeX
\documentclass[a4paper, 12pt, english]{article}
|
|
|
|
\usepackage[
|
|
headsep = 25pt
|
|
]{geometry}
|
|
|
|
\usepackage[utf8]{inputenc}
|
|
\usepackage[T1]{fontenc}
|
|
\usepackage{listings}
|
|
\usepackage{xcolor}
|
|
\usepackage{hyperref}
|
|
\usepackage{multicol}
|
|
\usepackage{setspace}
|
|
\usepackage{graphicx}
|
|
%\usepackage[acronym]{glossaries-extra}
|
|
|
|
\usepackage{enumitem}
|
|
\setlist{nosep}
|
|
|
|
|
|
\title{Advanced Operating Systems\\Assignment Week 11}
|
|
\author{Paul Lödige \\ Student ID: 37-229753}
|
|
|
|
\makeatletter
|
|
\newcommand*{\toccontents}{\@starttoc{toc}}
|
|
\makeatother
|
|
|
|
% acronyms
|
|
%\makeglossaries
|
|
|
|
\begin{document}
|
|
\maketitle
|
|
%\toccontents
|
|
|
|
%\printglossaries
|
|
|
|
\section{Assignment}%
|
|
\label{sec:Assignment}
|
|
Explain within 400 words the problem of using GPUs in Linux-based real-time systems.
|
|
%\clearpage
|
|
|
|
\section{Explanation}%
|
|
\label{sec:Explanation}
|
|
One of the main challenges in using GPUs in real-time systems
|
|
is the lack of predictability in the scheduling of GPU tasks.
|
|
In a typical Linux-based operating system,
|
|
the scheduler is designed to prioritize tasks based on a number of factors,
|
|
such as priority levels and resource requirements.
|
|
However, this can lead to unpredictable delays and jitter in the execution of GPU tasks,
|
|
which can be problematic for real-time systems that require deterministic behaviour.
|
|
|
|
Another issue with using GPUs in real-time systems
|
|
is the difficulty in ensuring determinism and reproducibility of results.
|
|
Due to the parallel nature of GPU processing,
|
|
it can be difficult to predict the exact order in which tasks will be executed,
|
|
which can lead to variations in the results of computations.
|
|
Additionally, the use of GPU-specific libraries and APIs can introduce further complexity
|
|
and make it difficult to guarantee that results will be consistent across different systems.
|
|
This problem is amplified by the fact that many GPU drivers for Linux are closed-source and proprietary,
|
|
which can make it difficult to customize or optimize the GPU for a specific real-time application.
|
|
While the availability of Open Source drivers has improved,
|
|
a lot of the GPU drivers for Linux are not designed for real-time use
|
|
and can introduce additional jitter and variability in the response times of the system.
|
|
|
|
There are also additional technical challenges associated with using GPUs in real-time systems.
|
|
For example,
|
|
the use of GPU memory can be difficult to manage in real-time systems,
|
|
and it can be challenging to ensure that data is transferred between the CPU and GPU in a timely and efficient manner.
|
|
|
|
In summary, the use of GPUs in Linux-based real-time systems can present several challenges,
|
|
including issues with scheduling and latency,
|
|
as well as difficulties in ensuring determinism and reproducibility of results.
|
|
|
|
|
|
\end{document}
|