Nahil Sobh @ on
Meep
Would love to have access to MEEP on the back end. We are working with a number of faculty members at Illinois who are interested in using MEEP to guide the design of their experiments and their subsequent setups.
Find information on common issues.
Ask questions and find answers from other users.
Suggest a new site feature or improvement.
Check on status of your tickets.
pending
Nahil Sobh @ on
Meep
Would love to have access to MEEP on the back end. We are working with a number of faculty members at Illinois who are interested in using MEEP to guide the design of their experiments and their subsequent setups.
Steven Clark @ on
Can you provide more information about what version of MEEP you require? I believe that version 0.10.1 is available on nanoHUB if not version 0.10.0 is available. Version 1.1.1 is pending installation.
Report abuse Reply
Nahil Sobh @ on
We are using 0.20.3
Report abuse Reply
Nahil Sobh @ on
We can definitely use the latest version 1.1.1
Report abuse Reply
Gerhard Klimeck @ on
delivering the simulation for Illinois and also Berkeley is a critical item. We must be able to simulate heavy lifting runs on the backbone clusters rossmann and coates.
Report abuse Reply
Steven Clark @ on
meep-1.1.1 is now available on both the nanoHUB execution hosts and the Purdue clusters.
To access on nanoHUB issue the command $ use meep-1.1.1
For serial execution the $ meep1 INPUT
For parallel execution in a workspace or tool session $ use mpich-gnu $ mpirun -n X meep-mpi INPUT for X less than 9
The Purdue cluster executables are accessed via the submit command. $ submit meep-1.1.1 INPUT for serial mode $ submit -n CORES -w TIME meep-1.1.1-mpi INPUT for parallel mode
Report abuse Reply
Nahil Sobh @ on
meep-mpi does not exist. Please see the log on my session below: ————————————————————————————————- nanoHUB ~/meep use mpich-gnu Do you want to make ‘use mpich-gnu’ persistent? (y/n) y
nanoHUB ~/meep mpirun -n 8 meep-mpi lens_rappture.ctl Warning: Command line arguments for program should be given after the program name. Assuming that lens_rappture.ctl is a command line argument for the program. Missing: program name Program meep-mpi either does not exist, is not executable, or is an erroneous argument to mpirun. nanoHUB ~/meep
Report abuse Reply
Steven Clark @ on
OK. You need to give the full path to meep-mpi. $ mpirun -n CORES
which meep-mpiINPUTReport abuse Reply
Alexander S McLeod @ on
Steven what is the proper submit venue for the new “dedicated meep cluster”? We receive whisperings from Gerhard that we should try it out for benchmarks and I’m ready to do so with the proper destination for submit. I’m assuming I will just use “mpirun -n CORES -v VENUE
which meep-mpiINPUT” as outlined above.Report abuse Reply
Nahil Sobh @ on
Here’s how we ran meep in the workspace and on Steele-cluster using Steve’s comments:
1. Workspace – Serial use meep-1.1.1 meep1 INPUT_FILE – Parallel (limited to 8 cores) which meep-mpi
Report abuse Reply
Steven Clark @ on
There seems to be some confusion here so let me re-state the instructions for running meep.
You can execute meep on nanoHUB resources or additional resources are available through the submit command. These resources are available for serial or parallel versions of meep.
To run on nanoHUB resources do the following – $ use meep-1.1.1
For serial execution the
$ meep1 INPUT
For parallel execution in a workspace or tool session
$ use mpich-gnu $ mpirun -n X
which meep-mpiINPUT # for X less than 9For execution on Purdue clusters
$ submit meep-1.1.1 INPUT for serial mode $ submit -n CORES -w TIME meep-1.1.1-mpi INPUT for parallel mode
It is not necessary to specify a venue for the submit command.
There is no “meep” cluster. Purdue has recently installed a new cluster named rossmann. See http://www.rcac.purdue.edu/userinfo/resources/rossmann for more information. NCN contributed to the cluster purchase and as a result has some elevated priority in it’s use.
Report abuse Reply
Alexander S McLeod @ on
To capitulate, should meep only be run in parallel without venue (=rossman?) or with mpirun locally? In other words, are the other venues allowed (clusterD/F, coates), or not? Meep-mpi does not appear to be installed on these.
Report abuse Reply
Steven Clark @ on
clusterD/F are meant to be used for smallish short running parallel jobs (5 minutes or less). The preferred choice for meep is send the jobs to the Purdue clusters. Presently coates and rossmann are configured to run taking advantage of a relatively high priority hub queue. With demonstrated use it will be possible to increase the nde count serving the hub queue.
Report abuse