KEMBAR78
Home Linux Lab Systems | PDF | Matlab | Usb
0% found this document useful (0 votes)
5 views13 pages

Home Linux Lab Systems

The document outlines the setup for a home lab using ECC-Linux and Raspberry Pi systems, detailing software tools like MobaXterm and MATLAB for development and mathematical computations. It provides instructions for installing Raspbian, configuring SSH and VNC access, and building a Raspberry Pi cluster for parallel processing. Additionally, it mentions the use of NVIDIA Jetson Nano for GPU computing and resources for accessing parallel systems through NSF ACCESS.

Uploaded by

samuel.siewert
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
5 views13 pages

Home Linux Lab Systems

The document outlines the setup for a home lab using ECC-Linux and Raspberry Pi systems, detailing software tools like MobaXterm and MATLAB for development and mathematical computations. It provides instructions for installing Raspbian, configuring SSH and VNC access, and building a Raspberry Pi cluster for parallel processing. Additionally, it mentions the use of NVIDIA Jetson Nano for GPU computing and resources for accessing parallel systems through NSF ACCESS.

Uploaded by

samuel.siewert
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 13

CSCI 551 - Home Lab Setup

ECC-Linux and CSCI Cluster Access

Alternative to ECC Systems for Multi-Core


Threading and Vector Instructions

MATLAB and other Tools to Check your Math

 Sam Siewert 1
ECC-Linux: Using MobaXterm from Windows
Download MobaXterm for Windows - https://mobaxterm.mobatek.net/download.html
• Best option for Windows host (compared to MS Powershell)
• Use VS-code, geany, vim, vi, etc. on R-Pi for graphical development tools
• Make sure you enabled SSH
• Note your IP address with “ifconfig –a” and set up MobaXterm SSH session

Download Cyberduck for Mac or Windows - https://cyberduck.io/download/


Advantages of MobaXterm
• Simple X-windows display
on your host
• Lower overhead than VNC
• Easy to use

Not available for Macintosh Host


Alternatives to MobaXterm
• Putty – simple download for
Windows or Mac
• VNC – full desktop display
from R-Pi
• Must enable VNC on R-Pi
for VNC
• Cyberduck, etc.

 Sam Siewert 2
MATLAB from CSU Portal
https://www.mathworks.com/academia/tah-
portal/california-state-university-chico-31511335.html

 Sam Siewert 3
Install MATLAB to Check you Math (Windows)
MATLAB Treats everything like a Vector or Matrix
Interpreted Programming Language – for Math and Vector/Matrix,
Linear Algebra, Simulation, etc.
Now available at CSU if you are interested – CSU Page, Quick Start
MATLAB
Toolboxes
“ver” to list

Computer Vision
GPU Coder
Image Processing
Parallel Computing
DSP System
Statistics and Machine
Learning

Symbolic Math

Help Center

https://www.mathworks.com/help/matlab/matlab_prog/matlab-operators-and-special-characters.html
 Sam Siewert 4
RT Embedded Linux – Raspberry Pi 4 (3b+ or Better)
70% of work on R-Pi possible Cost for home lab - $80 total
– Final project on R-Pi https://www.adafruit.com/product/3775

– R-Pi can be clustered but ECC-


Linux cluster or CSCI clusters
recommend instead! https://www.ideasonboard.org/uvc/

Raspberry Pi 3b+ Meets Course Requirements for Shared Memory


Multi-core, Vector instructions
• Does not support Intel PS XE, MPI or OpenMP
• Can use for final project and all POSIX threading
assignments
• Use ECC for MPI and OpenMP exercises
Recommend Canakit or similar with all components needed - $80
• Micro SD card, heat sinks, case, power cord with switch
• NOOB image on SD card – just enable wireless or wired
• Turn-key start-up
Quick start with Raspbian install (sufficient for course)
Simple UVC compatible USB-2 web camera (machine vision)
Set up with Keyboard, HDMI monitor and mouse one time!
Use with SSH Terminal like MobaXterm or Putty on home router
Alternatively access with VNC (higher overhead)
All POSIX thread starter code tested on R-Pi 3b+ and 4

https://www.canakit.com/raspberry-pi-4-starter-kit.html
https://www.canakit.com/raspberry-pi-3-model-b-plus-starter-kit.html
Home Lab – Physical Set-up
R-Pi 4 1. Wall socket 100-240 VAC for AC/DC adapter, 5V, 2.5 amp out
2. Cat-6 Ethernet cable to home router (wired configuration)
recommended 3. Wireless home router (wireless use)
(Specs) 4. Terminal for bring-up only - USB keyboard and mouse, HDMI monitor

3b+ Bottlenecks Internet


Much improved • SD card (Nand flash)
• USB-2 Camera I/O
over R-Pi 3b+ • Ethernet (< 1 Gbps)
• Memory (1 GB)

HDMI
4 Improvements
3b+ will work, • 2, 4, 8GB RAM
but about the •

1.5 GHz Quad core
Gigabit Ethernet
same cost • 2 USB-3, 2 USB-2

Well-resourced
• CPU cores
Use for • Linux SMP OS
• NEON instructions
Pthreads and
Project! Keyboard, Mouse

 Sam Siewert 6
Install Raspbian with NOOB SD card …
1. Connect to HDMI Monitor and USB Keyboard and Mouse and complete install …
2. Login as “pi”, default password is “raspberry”
3. Change it!
4. Go to “Pi icon” and “Preferences”
5. Under Raspberry Pi Configuration
Enable:

SSH (required for


MobaXterm or Putty)
VNC (if desired)

Configure any other


Localization your
might want

Keep physical monitor


+ keyboard + mouse

OR use VNC if you


need a desktop
 Sam Siewert 7
VNC Access to R-Pi from Win 10
Download and install VNC client for Win 10 or Mac
https://www.realvnc.com/en/connect/download/viewer/windows/

Numerous options exist, so choose best for your Host

VNC is simple and nice, but consumes resources and bandwidth for full desktop

Ask for help with access

VNC only for R-Pi

MobaXterm for ECC or R-Pi

 Sam Siewert 8
R-Pi 4 and Jetson Clusters

Alternatives to ECC for Distributed


Memory Cluster Scaling and GP-GPU
Parallel Processing

 Sam Siewert

9
R-Pi Distributed Memory Cluster
Basic Cluster with 4 R-Pi 4’s up to 8 – Cluster @Home
– 16 cores up to 32 cores – stack groups of 8 for more scaling!
– Interconnected with Gigabit Ethernet switch
– Use Open Source MPI or Intel OpenHPC with OpenAPI software
– Requires basic kit for heat sinks and cooling – Cloudlet, Pico Cluster
– Magpi article on building – Tested all CSCI 551 MPI code!

Minimum config
of 4 R-Pi 4’s

Build and scale


from there

Load open
source OpenMPI

 Sam Siewert 10
NVIDIA Jetson Nano (2g)
Low-cost GP-GPU platform with 4 ARM CPU cores and 128
co-processors
– Useful to learn CUDA, OpenCV, and Machine Learning
– May be useful for OpenCL (if NVIDIA supports), OpenMP, Pthreads
– Can explore hybrid shared memory multi-core + co-processing
– Low-cost ($69) + SD card to get going
NVIDIA Jetson Nano - High End ARM Linux
System with 100’s of Co-Processors shown - Similar home setup to R-Pi (SSH, VNC, or monitor)
a low-cost desktop option

Use SSH as primary method to connect

Initially set up with HDMI monitor, keyboard, mouse

Install Jetpack – Archive, Current Version

Comes with CUDA, OpenCV, and Machine


Learning tools pre-installed
 Sam Siewert 11
NSF ACCESS for Parallel Systems
https://access-ci.org/ - Parallel Clusters and GP-GPU
– Register to get an account
– Pick a donated service (e.g., ACCESS CI (XSEDE))
– Start a project!

 Sam Siewert 12
Create a Profile – Get Help, Contribute

 Sam Siewert 13

You might also like