360 likes | 534 Views
Using NPACI Systems Part I. NPACI Environment ( http://www.npaci.edu/Resources ). NPACI Parallel Computing Institute August 13- August 17, 2001 San Diego Supercomputer Center. NPACI Environment Outline:. Logging in ssh, startup scripts File System $HOME, $WORK, HPSS, HSI Accounting
E N D
Using NPACI SystemsPart I. NPACI Environment(http://www.npaci.edu/Resources) NPACI Parallel Computing Institute August 13- August 17, 2001 San Diego Supercomputer Center
NPACI Environment Outline: • Logging in • ssh, startup scripts • File System • $HOME, $WORK, HPSS, HSI • Accounting • reslist
NPACI environment: logging in • ssh horizon.npaci.edu -l username • ssh (secure shell) required for secure access • see: http://security.sdsc.edu/
NPACI environment: unix startup files • .cshrc, .login, and .profile NPACI provided • set up programming environment • set up unix environment • execute COPYDEFAULTS to restore startup files /usr/local/paci/shellrc/bin/COPYDEFAULTS (then source .cshrc)
Filesystem Structure • $HOME/: 50-100 MB/user of backed up storage used for storage of unix essentials, etc. • $WORK/: >200 GB of purged (96 hours) disk storage primary workspace • Archival Storage: HPSS(SDSC)/DMF(TACC) • Disk/tape system providing TBytes of storage
HPSS - Archival Storage (SDSC) • Interactive execution: “hsi” is recommended HPSS interface utility; can also use pftp - Beginning Interactive Execution: tf004i% pftp Saving a file in HPSS: pftp> put filename Retrieving a file from HPSS: pftp> get filename Ending Interactive Execution: pftp> quit
HPSS - Archival Storage (SDSC) • Single line execution: • Saving a file in HPSS: tf004i% pftp put filename • Retrieving a file from HPSS: tf004i% pftp get filename • For tips and examples of using HPSS, see http://www.npaci.edu/HPSS http://www.sdsc.edu/Storage/hsi/
NPACI environment: accounting • Normally updated every 24 hours • beware of going over your allocation • Commands: • reslist -u username (-a accountname) • resalloc
Using NPACI SystemsPart II. Blue Horizon(http://www.npaci.edu/Horizon) NPACI Parallel Computing Institute August 13- August 17, 2001 San Diego Supercomputer Center
Overview • Hardware overview • Compiling programs • How CPUs are requested • Running parallel jobs • Interactive • Loadleveler scripts
Configuration • 144 IBM SP-High Nodes • 8 shared memory processors/node for 1152 processors • 4 Gbytes/node • Power3 processors • 375 MHz • 4 floating point operations per cycle • 1.5 GFLOPS/processor • Nodes connected together with IBM switch • 350 MB/second
Programming Environment • Compilers • Fortran : xlf, xlf90, mpxlf, mpxlf90 • C/C++ : xlc, xlC, mpcc, mpCC, gcc(GNU c/c++) • Tools • Debuggers • Profilers • Parallel Environment (PE) and Parallel Operating Environment (POE) • Enables running parallel jobs • Message passing libraries: LAPI, MPL, MPI, MPICH
Compiling C • C • xlc -qtune=pwr3 -qarch=pwr3 -O3program.c • C MPI • mpcc -qtune=pwr3 -qarch=pwr3 -O3program.c • C OpenMP • cc_r -qsmp=noauto:omp program.c • C MPI and OpenMP • mpcc_r -qsmp=noauto:omp program.c • C++ • xlC mpCC xlC_r mpCC_r
Compiling Fortran • Fortran • xlf -qtune=pwr3 -qarch=pwr3 -O3program.f • Fortran MPI • mpxlf -qtune=pwr3 -qarch=pwr3 -O3program.f • Fortran OpenMP • xlf_r -qsmp=noauto:omp program.f • Fortran MPI and OpenMP • mxlf_r program -qsmp=noauto:omp program.f • Fortran 90 • xlf90, mpxlf90 xlf90_r mpxlf90_r
Laura C. Nett: Describe the break down of the machine: Describe how to request processors or CPUs. Note that # of nodes and tasks are requested via the POE or LL environment where # of threads is set via environmental variable. Current hardware is limited to only using 4 MPI tasks per node meaning that 4 PEs will be idle ONLY if using fast communications...don’t worry because you aren’t charged for them. New colony switch will change that. Note when using threads: set the number of threads to spawn per MPI task. How CPUs are Requested • Blue Horizon has 144 Nodes with 8 CPUs on each node. 144*8=1152 CPUs • Request: # of Nodes # of tasks per Node # of threads per task (hybrid code only) • # of CPUs requested = nodes*tasks*threads • Hybrid code - spawns threads per MPI task
Laura C. Nett: Interactive jobs run similar to SP2 except you need to define the number of CPUs by defining # of nodes and tasks. if you want threads to be spawned define this before running poe. The interactive CPUs are set aside for use during the day and unlike the SP2 they are not shared. There are various environmental variables that you may want to set and they get defined using all lower case. Remember that the tasks per node is a number between 1-8 the number of CPUs on a node but for straight MPI code using fast communications the max you can set this to is 4. Running Interactive Jobs • Interactive Jobs: poe a.outarg_list -nodes n -tasks_per_node m -rmpool 1 • Interactive Debugging: pedb a.outarg_list -nodes n -tasks_per_node m -rmpool 1 Alternatively, use environment variables: setenv MP_NODES n setenv MP_TASKS_PER_NODE m setenv MP_RMPOOL 1 tasks_per_node: 1-8 (number of CPUs on each node)
Batch Jobs with LoadLeveler • Develop a program and make sure it runs correctly • Write a script file that has information about your job and the nodes/tasks/threads you want to run it on • Submit the script file to the LoadLeveler • Check the status of your job • Batch Script Generator (sample) http://www.sdsc.edu/~lnett/Tools/batchform.html
LoadLeveler Scripts: Example 1 #!/bin/ksh # @ environment = MP_EUILIB=us; MP_SHARED_MEMORY=YES; MP_PULSE=0; MP_INTRDELAY=100; # @ class = high # @ job_type = parallel # @ node=16 # @ tasks_per_node=8 # @ node_usage=not_shared # @ network.MPI=css0,not_shared,US # @ wall_clock_limit =3:30:00 # @ input = /dev/null # @ output = LL_out.$(jobid) # @ error = LL_err.$(jobid) # @ initialdir = /work/login-name/mydir # @ notify_user = login-name@npaci.edu # @ notification = always # @ queue poe a.out or poe myscript myscript: #!/bin/ksh a.out
LoadLeveler Scripts: Example 24CPUs per node PLUS 2 threads per task #!/bin/ksh # @ environment = MP_EUILIB=us; MP_SHARED_MEMORY=YES; MP_PULSE=0;MP_INTRDELAY=100; # @ class = high # @ job_type = parallel # @ node=16 # @ tasks_per_node=4 # @ node_usage=not_shared # @ network.MPI=css0,not_shared,US # @ wall_clock_limit =3:30:00 # @ input = /dev/null # @ output = LL_out.$(jobid) # @ error = LL_err.$(jobid) # @ initialdir = /work/login-name/mydir # @ notify_user = login-name@npaci.edu # @ notification = always # @ queue export OMP_NUM_THREADS=2 export SPINS=0 export YIELDS=0 export SPINLOOPTIME=5000 poe a.out or poe myscript myscript: #!/bin/ksh export OMP_NUM_THREADS=2 export SPINS=0 export YIELDS=0 export SPINLOOPTIME=5000 a.out
Keywords in LoadLeveler Scripts arguments - arguments to pass to the executable input - file to use as stdin output - file to use as stdout initialdir - initial working directory for job executable - executable or script to run job_type - can be parallel or serial notify_user - user to send email to regarding this job, keywords are case insensitive
Keywords for LoadLeveler Scripts(cont.) notification - e-mail to notify_user (always, error, start, never, complete) node - how many nodes do you want tasks_per_node - MPI tasks/node (max of 4 for now) wall_clock_limit - maximum wall clock time to use for the job class - class (queue) to run in (low,normal,high,express) queue - puts the job in the queue
LoadLeveler Commands • Job Submission llsubmit filename submit job filename to LoadLeveler llcancel [-q] jobid cancel a job with the given id llq [–s job_number] list job [job number] • Job Status showq show job queue showbf show backfill
Documentation NPACI Blue Horizon documentation http://www.npaci.edu/Horizon IBM documentation http://www.rs6000.ibm.com/resource/aix_resource/sp_books
Lab session-Blue Horizon • Log into horizon ssh -l username horizon.npaci.edu • Ensure you have the default environment /usr/local/paci/shellrc/bin/COPYDEFAULTS source .cshrc • Copy sample LoadLeveler script and code cp /work/NPACI/Training/LoadLeveler/LLexample . cp /work/NPACI/Training/LoadLeveler/hello_mpi.f . • Compile sample code: mpxlf hello_mpi.f • Run code interactively: MPI ONLY: poe a.out -nodes 1 -tasks_per_node 4 -rmpool 1 MPI/OpenMP: setenv OMP_NUM_THREADS 2 poe a.out -nodes 1 -tasks_per_node 4 -rmpool 1 • Submit job to LoadLeveler batch queue: • First edit LLexample llsubmit LLexample • Check current usage reslist
Using NPACI SystemsPart III. SUN HPC 10000(http://www.npaci.edu/HPC10000) NPACI Parallel Computing Institute August 13- August 17, 2001 San Diego Supercomputer Center
Outline • System Overview • Compiling programs • Running jobs • Lab example
Sun HPC System Overview • 64 UltraSPARC II processors with 64GB RAM • Uniform Memory Access (UMA) shared memory parallelism • Each processor has 16 kB of direct-mapped data cache, 4 MB of external cache, 400 MHz clock speed, and peak performance of 800 MFLOPS.
Accessing the Sun HPC • The Sun HPC 10000 is actually a collection of systems • Compute Engine: ultra — 64 processors • for batch jobs only • Front End : gaos.sdsc.edu —12 processors • development compatible with ultra • launch all jobs here • Connecting to the System ssh gaos.sdsc.edu -l username
Compiling • Serial Programs with Forte (Workshop 6.1) compilers f77, f90, cc, CC • MPI Programs with HPC 3.1 Front End tmf77, tmf90, tmcc, tmCC (with -lmpi) • OpenMP programs using Kuck & Associates preprocessor guidef77, guidec • OpenMP programs using Sun Forte Compilers f90 -mp=openmp -xparallel –stackvar
Running Interactively • Serial gaos % myprog • OpenMP gaos % setenv OMP_NUM_THREADS 4 gaos % ompprog • MPI gaos % bsub -I -n 4 -m ultra mpiprog (Will run in LSF “hpc” interactive queue)
Batch Queue Structure • LSF batch queues • holding place for batch requests according to the supplied information • submits a job for batch execution on host(s) that satisfy the resource requirement • Batch Script Generator http://www.sdsc.edu/~lnett/Tools/batchform.html
Submitting a Batch Job #1 • Simple MPI Program gaos % bsub -q normal -m ultra -n 16 –oe out pam prog • Equivalent Batch Script gaos % cat mpiprog.bsub.1 #!/bin/csh #BSUB -q normal #BSUB -m ultra #BSUB -n 16 # total process reservations #BSUB –oe out #BSUB cd /paci/sdsc/frost/src/mydir pam prog • Submitting the Batch Script gaos % bsub < mpiprog.bsub.1
Submitting a Batch Job #2 • Batch Script for OpenMP job gaos % cat ompprog.bsub.1 #!/bin/csh #BSUB -q high #BSUB -m ultra #BSUB -o ompprog.log.1 #BSUB -e ompprog.err.1 #BSUB -u user@somewhere.net #BSUB -n 16 # total thread reservations #BSUB cd /paci/sdsc/frost/src/mydir setenv OMP_NUM_THREADS 16 ompprog • Submitting the Batch Script gaos % bsub < ompprog.bsub.1
Monitoring and Managing Jobs • Status of your job: gaos % bjobs [-l jobid] • All Jobs gaos % bjobs -u all • Job Load Per Machine gaos % bhosts • Killing a job (jobid=33443) Gaos % bkill 33443
Lab session-Ultra • Login % ssh gaos.sdsc.edu -l username • Ensure you have the default environment % /usr/local/paci/shellrc/bin/COPYDEFAULTS % source .cshrc • Check your current usage % reslist • Get job script example and sample code % cp /work/NPACI/Training/LSF/lsf_script . % cp /work/NPACI/Training/LSF/hello_mpi.f . • Compile % tmf90 hello_mpi.f –lmpi • Run job interactively % bsub -I -m ultra -n 4 a.out • Submit job to LSF batch queue • First edit the lsf_script % bsub < lsf_script