Skip to content

Latest commit

 

History

History
96 lines (79 loc) · 3.27 KB

README.md

File metadata and controls

96 lines (79 loc) · 3.27 KB
Raw

High Performance Computing (HPC) Benchmark

Your friendly neighborhood lightweight utility for monitoring resource usage of your pipelines and analyses.

Monitor metrics of any command line run. Metrics include CPU percentage, memory usage (MB), and processor information. Metrics information will be logged every second by default. You can set the sampling rate with flag --max_sampling_rate. A CSV file will output after the run is complete that contains metrics information.

Installation

Currently, this CLI tool is on TestPyPi. To pip install:

pip install -i https://test.pypi.org/simple/ hpc-benchmark

User Guide

hpc_benchmark run

hpc_benchmark run 'run command' [options] -o <output_directory>

Arguments:
    'run command'              Command line run you would like to monitor
    <output_directory>         Directory you would like to save output CSV file 

Options:
  --out_dir -o  PATH            Full path where output CSV file will be saved in
  --max_sampling_rate -msr INT  Frequency in seconds of how often to output metrics.
                                Default is about 1 second. This sampling rate is an 
                                estimate and may output +/- 2 seconds of the MSR.
  --help                        Print help

INPUTS:
- Your command should be between `' '`. This way, hpc_benchmark will run 
  whatever is between `' '` in the terminal.

OUTPUTS:
- Specify output directory with flag `-o`. hpc_benchmark will output a CSV file 
  with all the COU and memory usage informtion. 

OPTIONAL: 
- The `-msr INT` flag should be used if you want to specify the interval in which CPU 
  or memory usage gets benchmarked. It defaults to 1 second, so if you are running 
  something that takes hours, it is recommended to use `-msr 60` so that metric will 
  be meaured every minute.

EXAMPLE:

hpc_benchmark run 'docker run \
--security-opt=apparmor:unconfined \
--rm -v /home/{username}/data:/data \
-v /home/{username}/output:/output fcpindi/c-pac:latest \
/data /output participant \
--save_working_dir --skip_bids_validator \
--n_cpus 1 --mem_gb 15 \
--participant_label sub-12345 \
--preconfig abcd-options' \
-o /home/{username}/hpc_benchmark.csv -msr 60



hpc_benchmark plot

hpc_benchmark plot </home/{username}/{file}.csv>

Arguments: 
  <CSV file>                  CSV file that was the output for hpc_benchmark run

Details: Use the output CSV file from your hpc_benchmark run as the input argument for hpc_plot_metrics. This will open a new tab in your internet browser of your plot. Plot can later be saved by clicking the camera icon in the top right corner.

EXAMPLE:

hpc_benchmark plot /home/{username}/hpc_benchmark.csv

image