The purpose of this project is to analyze raw CPU temperature data parsed from a log file and calculate the peicewise linear interpolation and global linear least squares formula of the trend.
Table of Contents
[TOC]
The items below are required to compile the project.
Both g++ and Clang++ are tested to work however, the included makefile exclusively targets g++-8 which has been more thoroughly tested.
apt-get install g++-8
apt-get install cmake
The Eigen library for C++ is used for precision matrix and vector mathematics. More information on the library can be found here
The necessary Eigen dependencies are included in this repository in ./libs.
The included makefile assumes:
- g++-8 is available.
Although filesystem is no longer experimental in GCC 8 it still needs
LFLAGS = -lstdc++fsset to build correctly.
The makefile targets the included lib folder containing the necessary Eigen dependancies.
With this in mind the program can be compiled by invoking:
make -f makefile
The program runs from the command line. It accepts arguments pointing to the text files containing the temperature data.
./CS417SP file1.txt
It can also accept multiple files at once:
./CS417SP file1.txt file2.txt
Basic validation is done to ensure the given input file(s) exist and it is assumed if the file(s) exist they are formatted correctly.
The program assumes the input data takes the form of temperatures in a txt file as folows:
+61.0°C +63.0°C +50.0°C +58.0°C
+80.0°C +81.0°C +68.0°C +77.0°C
+62.0°C +63.0°C +52.0°C +60.0°C
+83.0°C +82.0°C +70.0°C +79.0°C
+68.0°C +69.0°C +58.0°C +65.0°C
Characters not in the language {0-9;. } are removed as the file is parsed.
The software creates one output file per core per dataset that follow the following nomenclature:
{epoch_time}-Temperature-Data-{file number}-Core-{core number}
The output is formatted as follows
xk <= x < xk+1; yi = c0 + c1x ; type
Where:
- xk and xk+1 are the domain in which yk is applicable
- yk is the kth function
- type is either least-squares or interpolation
- Cody N
- The Eigen Project
- Thomas Kennedy Ph.D - Old Dominion University