Use Rack Mate T1 Build My Own RPi Cluster and RPi5 AI Agent.

by yoyojacky in Workshop > Home Improvement

32 Views, 0 Favorites, 0 Comments

Use Rack Mate T1 Build My Own RPi Cluster and RPi5 AI Agent.

DP-0022-A+-01.jpg
DP-0022-13.png

DeskPi Rack Mate T1

DeskPi Rack Mate T1 is Aluminum Alloy and Acrylic Frame Mini Chassis which you can setup your own cluster or home assistant server farm into the Mini Chassis. 

This chassis is made of high-strength aluminum alloy and transparent acrylic materials, providing a stable and durable structure. It can be used to support and protect various small-sized devices or mechanical components. The chassis features a frame structure composed of multiple aluminum alloy beams and columns. The acrylic side panels and cover parts offer transparency and an aesthetically pleasing appearance.

Supplies

DP-0022-05.jpg
jetsonnano.png
rpi5.png

you may need some screw drivers and Raspberry Pi, jetson nano development boards

here is my list:

  • 4 x Raspberry Pi 4B
  • 2 x Raspberry Pi 5
  • 1 x Jetson nano
  • 1 x 12 port switch
  • 1 x DeskPi pro
  • 1 x Super6c with (6 Raspberry Pi CM4 module)
  • 6 x M.2 NVMe SSD (256GB)
  • 1 x M2.5 screw driver
  • 1 x Hexgon wranch
  • 1 x 12 inch Fan
  • 1 x Rack Mate T1 mini chassis
  • 1 x Mini tower kit

Purchse URL:

Assemble Mini Chassis

DP-0022-09.png
DP-0022-08.png
DP-0022-07.png
DP-0022-06.png
DP-0022-05.png
DP-0022-15.png
DP-0022-15.png
DP-0022-09.png
DP-0022-02.jpg
DP-0022-03.jpg
DP-0022-11.png
DP-0022-10.png
DP-0022-12.png
DP-0022-13.png
DP-0022-14.png
DP-0022-08.png
DP-0022-10.png
DP-0022-11.png
DP-0022-12.png

It is very esay to assemble it, only one part you need to pay attension. the expansion board need to be insert into Raspberry Pi 4's HDMI ports and USB-C port, please insert it gentelly, and the copper pillar's size you may need to pay attention, it is different on expansion board and Raspberry Pi.

Put Everything Into Chassis

Rackmate_t1_03.jpg
rackmate_t1-01.jpg
rackmate_t1_status.jpg
rackmate_t1_OUTLOOK.jpg

Finally, I have build my own cluster on desktop, and I have alreay configured my raspberry Pi 5 as an AI agent for my cluster by installing ollama framework.

Downloads

Intall Ollama Framework

github_ollama.png
ollama_logo.png
001.png
002.png

open a terminal on my Raspberry Pi 5 and typing following command :

curl -fsSL https://ollama.ai/install.sh | sh
ollama run phi3 --verbose

Downloads

Pull Down the Models

images.png
016.png
015.png
014.png
013.png
012.png
011.png
010.png
009.png
008.png
007.png
006.png
005.png
004.png
003.png

if you want , you can pull down the models if you like :

for model in llama3 llama2 phi3 mistral qwen qwen2 qwen:4b qwen:7b
do
ollama pull $model
done

and it may take a while and about 40GB disk space at least.


Chat With Ollama

just chat with your AI model in the command line and you can install a webUI for your ollama.