This is a python package to allow usage of hadd (from ROOT CERN) in parallel. It allows adding histogram and .root files. The addition will occur in blocks instead of one at all. It was designed to work with limited memory (local development primilary). The addition will use multiprocessing if avilable in your python enviroment.
The minimum supported python version is 3.6.
| Python version | Supported |
|---|---|
| 2.x | NO |
| 3.1-3.5 | NO |
| 3.6 | YES |
| 3.7 | YES |
| 3.8 | YES |
| 3.9 | YES |
| 3.10 | YES |
| 3.11 | YES |
you can install the package directly from pypi repository
pip install parallel-hadd
- clone this repository
git clone https://github.com/MohamedElashri/hadd-parallel
- build the package from source using
pip
pip3 install ./hadd-parallel
The command to use the package is
phadd
you can use it as the following (with default arguments)
phadd out.root *.root
where out.root will be the output file the contains all the *.root files.
more information about how to customize the settings are available on program help page
phadd -h
usage: phadd [-h] [-t TMPDIR] [-j NUM_JOBS] [-n NUM_FILES] [-f] [-s] [-l {DEBUG,INFO,WARNING,ERROR,CRITICAL}] [-v]
output_file input_files [input_files ...]
Hadd ROOT histograms in parallel
positional arguments:
output_file the output file
input_files one or more input files
optional arguments:
-h, --help show this help message and exit
-t TMPDIR, --tmpdir TMPDIR
the temporary directory to store intermediate files
-j NUM_JOBS, --jobs NUM_JOBS
the number of jobs to run in parallel , [default cpu_count * 2 = {N_JOBS}
-n NUM_FILES, --num_files NUM_FILES
the number of files to hadd at once
-f, --force_overwrite
force overwrite of output file
-s, --save-tmp save the intermediate files, otherwise they will be deleted (which is the default)
-l {DEBUG,INFO,WARNING,ERROR,CRITICAL}, --log {DEBUG,INFO,WARNING,ERROR,CRITICAL}
the log level, [default Warning]
-v, --version show program's version number and exit