Profiling Benchpress benchmarks

We are using a docker container based runtime for running the benchpress benchmarks. Here we present the steps to run the apps:

Step 1: Setting up

• Setup the docker image for running the benchmarks docker-compose build

  pip install -r scripts/00_requirements.txt

• Install the applications locally (only required once)

  docker-compose run benchpress bash install_benchmarks.sh

Step 2: Run monitoring daemon

This is required for reporting the metrics

cd node-exporter
docker-compose up

Step 3: Run the benchmarks

• FIO

  docker-compose run benchpress python3 benchpress_cli.py -b benchmarks.yml -j jobs/jobs.yml run "fio aio" 
  

• Schbench

  docker-compose run benchpress python3 benchpress_cli.py -b benchmarks.yml -j jobs/jobs.yml run "schbench default"
  

• Gapbs

  docker-compose run benchpress python3 benchpress_cli.py -b benchmarks.yml -j jobs/jobs.yml run "gapbs_bc"
  

• Nginx_Wrk

  docker-compose run benchpress python3 benchpress_cli.py -b benchmarks.yml -j jobs/jobs.yml run "nginx_wrk_bench default"
  

• Graph500

  docker-compose run benchpress python3 benchpress_cli.py -b benchmarks.yml -j jobs/jobs.yml run "graph500_omp_csr"
  

• Minebench

  docker-compose run benchpress python3 benchpress_cli.py -b benchmarks.yml -j jobs/jobs.yml run "minebench_plsa default"