MG5 multi-threading option?? - TOO SLOW on cluster **
Hi,
I have set 34 jobs on the HEP cluster at my institution (scripts cycling through sequentially ’switching on’ EFT parameter values one at a time) for two given models.
Usually, the jobs seem to go quite fast but I just checked and they haven’t progressed past the first cycle of the loop, still running the Madgraph / Rivet processes but very slow … !
but some of jobs have spawned a very high number of threads each - is madevent running in some kind of multiprocessing mode?
How does one fix this i.e. stop the multi-threading?
I attach an example script I am using that works just fine but the jobs needs to be sped-up (this is not a cluster hardware performance issue):
#! /bin/bash
#PBS -o /unix/bsmxsec/
#PBS -e /unix/bsmxsec/
#Specify name of parameter:
par_name=cW
cd /unix/bsmxsec/
n=0
mkdir "${par_name}"
cd ${par_name}
#for i in -1.0 -0.75 -0.50 -0.25 1.00000e-99 0.25 0.50 0.75 1.0
for i in -0.25 -0.2 -0.15 -0.1 -0.05 0.05 0.1 0.15 0.2 0.25
do
cp /home/smustafa/
find /home/smustafa/
#######
~/SFNP/
import model SMEFTsim_
generate p p > Z Z j j QCD==0
add process p p > Z Z j j QCD==2
output EWK_QCD_
launch
shower = pythia8
0
set nevents 100000
set ebeam1 7000
set ebeam2 7000
0
exit
EOF
cd $PWD/EWK_
gunzip unweighted_
~/MadGraph5_
1
unweighted_
decayed.lhe
z
4
EOF
mv unweighted_
mv decayed.lhe unweighted_
gzip unweighted_
mkfifo tag_1_pythia8_
sleep 1s
rm tag_1_pythia8.cmd
cp /home/smustafa/
sleep 1s
cp ~/Analysis/
./run_shower.sh &
#######
export ATLAS_LOCAL_
sleep 2s
rivet -a ATLAS_2019_I00001 --pwd tag_1_pythia8_
n=$((n+1))
cd ../../../
done
Question information
- Language:
- English Edit question
- Status:
- Solved
- Assignee:
- No assignee Edit question
- Solved by:
- Olivier Mattelaer
- Solved:
- Last query:
- Last reply: