Skip to content

Files

Failed to load latest commit information.

Latest commit

 Cannot retrieve latest commit at this time.

History

History

simulator

README.md

Simulator

0. Content

  • simulator/ contains code for simulation and is adapted from Tiresias.
    • cluster_spec/ contains configuration files for cluster, e.g., the number of nodes, the number of GPU per node.
    • trace-data/ contains traces for simulation evaluation.
    • calc.py computes metrics, e.g., avg. JCT, Makespan, and 99th JCT.
    • cluster.py, switch.py, and node.py contain implementations of the cluster.
    • jobs.py and model.py contain information of the jobs.
    • flags.py contains the argument definition method.
    • log.py and utils.py contain auxiliary functions.
    • matching.py contains the implementation of the matching algorithm for Muri.
    • run_sim.py contains the implementation of different scheduling policies.

1. Environment config

Step 1: create conda environment

# create conda env
conda create -n muri python=3.8
conda activate muri

Step 2: install python dependencies

conda install numpy
conda install -c conda-forge cvxpy

2. Reproduce simulation results (for SIGCOMM'22 artifact evaluation)

  • cd <repo>/simulator
  • Figure 9: bash sim_fig9.sh (takes about 2 days)
  • Figure 10: bash sim_fig10.sh (takes about 4 days)
  • Figure 11: bash sim_fig11.sh (takes about 2 days)
  • Figure 12: bash sim_fig12.sh (takes about 2 days)
  • Figure 13: bash sim_fig13.sh (takes about 1 days)
  • Note: The detailed results will be stored in <repo>/simulator/results/.
  • To generate the figure shown in our paper, please change the raw data in draw_fig9-10.py and draw_fig11-13.py to the test results (JCT, Makespan, and/or 99th JCT), and run python draw_fig9-10.py and python draw_fig11-13.py.