High performance stack

Requirements

• Install vagrant (https://www.vagrantup.com/downloads.html)
• Add vagrant command to PATH variable (if need)

WorkFlow

vagrant up  # starts and provisions the vagrant environment
vagrant ssh  # connects to machine via SSH

vagrant destroy  # stops and deletes all traces of the vagrant machine

imm.py

Simple automation tool for compiling and running your programs on IMM URAN supercomputer.

Also, you can use it for local running.

1. Setup you environment

   def local_env():
       env.user = 'vagrant'
       env.password = 'vagrant'
       env.host_string = 'localhost:22'
   
   
   def cluster_env():
       env.user = 's0*****'
       env.password = 'Vjk******'
       env.host_string = 'umt.imm.uran.ru:22'
   

2. Write your compiling and running commands

   def main():
       # put files to remote server
       _put('kmeans.cpp')  
       _put('data-gen.cpp')
   
       # compiling command && run
       run('g++ data-gen.cpp -o data-gen && ./data-gen 3 10 10 data-gen.txt')
       run('g++ kmeans.cpp -o kmeans -fopenmp && ./kmeans 10 data-gen.txt kmeans.txt')
   
       # get files from remote server
       _get('data-gen.txt')
       _get('kmeans.txt')
   
   if __name__ == "__main__":
       local_env()  # activate local environment
       #supercomputer_env()  # or activate supercomputer environment
       main()  # running main :) 
   

3. Up container (vagrant up)
4. Connect via SSH to container (vagrant ssh)
5. Run imm.py (python imm.py）

from https://github.com/pahaz/high-performance-computing-stack