Skip to content

README.md

Latest commit

 

History

History
64 lines (50 loc) · 2.21 KB

README.md

File metadata and controls

64 lines (50 loc) · 2.21 KB

SONIC workflows

This repository serves to deploy and run SONIC workflows for performance tests.

Setup

source /cvmfs/cms.cern.ch/cmsset_default.sh
wget https://raw.githubusercontent.com/fastmachinelearning/sonic-workflows/CMSSW_16_0_X/setup.sh
chmod +x setup.sh
./setup.sh
voms-proxy-init --rfc --voms cms -valid 192:00
cd CMSSW_16_0_0_pre4/src/sonic-workflows
cmsenv

Notes:

  • Input AODSIM files currently hosted at Purdue

Purdue resources

Interactive CPU job:

sinteractive --account=cms --partition=hammer-g -N 1 -n1 -c4 --mem-per-cpu=2G

Interactive GPU job:

sinteractive --account=cms --partition=hammer-f --gres=gpu:1 -N 1 -n1 -c4 --mem-per-cpu=2G

Running

To see the available options:

cmsRun run.py --help

To run a workflow with the default settings:

cmsRun run.py --maxEvents 100

Driver commands

2023 miniAOD:

runTheMatrix.py -w upgrade -l 12434.21 --dryRun --command="--no_exec"

Modified commands:

./get_files_on_disk.py /TTtoLNu2Q_TuneCP5_13p6TeV_powheg-pythia8/Run3Summer23DRPremix-130X_mcRun3_2023_realistic_v14-v2/AODSIM -o files__TTtoLNu2Q_TuneCP5_13p6TeV_powheg-pythia8__Run3Summer23DRPremix-130X_mcRun3_2023_realistic_v14-v2__AODSIM.txt
cat files__TTtoLNu2Q_TuneCP5_13p6TeV_powheg-pythia8__Run3Summer23DRPremix-130X_mcRun3_2023_realistic_v14-v2__AODSIM.txt | head -n 3 > files__TTtoLNu2Q_TuneCP5_13p6TeV_powheg-pythia8__Run3Summer23DRPremix-130X_mcRun3_2023_realistic_v14-v2__AODSIM__truncated.txt
cmsDriver.py step4  -s PAT --conditions auto:phase1_2023_realistic --datatier MINIAODSIM -n 10 --eventcontent MINIAODSIM --geometry DB:Extended --era Run3_2023 --no_exec --filein filelist:files__TTtoLNu2Q_TuneCP5_13p6TeV_powheg-pythia8__Run3Summer23DRPremix-130X_mcRun3_2023_realistic_v14-v2__AODSIM__truncated.txt --fileout file:step4.root

Run3/Phase2 SONIC-enabled workflows are available from runTheMatrix.py -w upgrade -n with suffix .9001

Listing models

The following script provides a list of all models possibly used by a config:

./getModels.py --config step4_PAT

(Some of the listed models may not actually be used, depending on task and output configurations, but that can only be evaluated by fully executing the config.)