Hp XC System 2.x Software Manuel d'utilisateur Page 75
- Page / 154
- Table des matières
- MARQUE LIVRES
Noté. / 5. Basé sur avis des utilisateurs
If you specify a script at the end of the srun command line (not as an argument to -A), the
spawned shell executes that script using the allocated resources (interactively, without a queue).
See the -b option for script requirements.
If you specify no script, you can then execute other instances of srun int eract ivel y, within the
spawned subshell, to run multiple parallel jobs on the resources that you allocated to the subshell.
Resources (such as nodes) will only be freed for other jobs when you terminate the subsh ell.
-a=jobid (--attach=jobid)
The -a=jobid option attaches (or reattaches) your current srun session to the already
running job whose SLURM ID is jobid. The job to which you attach m ust have its resources
managed by SLURM, but it can be eith er interactive ("allocated," started with -A)orbatch
(started with --b). This option allows you to monitor o r intervene in previously started srun
jobs. You cannot use -a with -b or -A. Because the running job to which you attach already
has its resources specified, you cannot use -a with -n, -N,or-c. You can only attach to
jobs for which you are the authorized owner.
By default, -a attaches to the designated job read-only. stdout and stderr are copied to
the attaching srun, just as if the current srun session h ad started the job. However, signals
are not forw arded to the remote processes (and a single Ctrl/C will detach the read-only srun
from the j ob).
If you use -j (-join)or-s (-steal) alo n g with -a,yoursrun session join s the running
job and can also forward signals to it as well as receive stdout and stderr from it. If you
join a S L U RM batch (-b) job, you can send signals to its batch script. Join (-j) does not
forward stdin, but steal (-s, which closes other open sessions with the job) does forward
stdin as well as signals.
-j (--join)
The -j option joins a running SLURM job (always used only with -a op tion to specify the
jobid). This not only duplicates stdout and stderr to the attaching srun session, but it
also forwards signals to the job’s script or processes as well.
-s (--steal)
The -s option steals all connection s to a running SLURM job (always used only with -a
option to specify the jobid). -steal closes any open sessions with the specified job, then
copies stdout and stderr to the attaching srun session, and i t also forwards both signals
and stdin to the job’s script or processes.
6.4.4 srun Resource-Allocation O ptions
The srun options assign compu te resources to your parallel SLURM -m anaged job. These
options can be used alone or in combination. Also, refer to the other srun options that can
affect node management for your job, especially the control options and c onstrain t options.
-n procs (--nprocs=procs)
The -n procs option requests that srun execute procs processes. To control how these
processes are distributed amon g nodes and CPUs, combine -n with -c or -N as explained
below (default is one process per node).
-N n (--nodes=n)
The -N n option allo cates at least n nodes to this job, where n may be one of the following:
• a specific n ode count (such as -N16)
• a node count range (such as -N14-18)
Using SLURM 6-5
- HP XC System Software 1
- Contents 3
- 3 Developing Applications 4
- 4 Debugging Applications 5
- 5 Tuning Applications 5
- 6UsingSLURM 5
- 8UsingHP-MPI 6
- 7 Using LSF 6
- 9UsingHPMLIB 7
- 10 Advanced Topics 8
- AExamples 8
- About This Document 11
- HP XC Inform ation 12
- For M ore Information 13
- Supplementary Information 13
- Manpages 14
- Related Information 15
- Typographical Conventions 16
- HP Encourages Your Comments 17
- 1.1 System Architecture 19
- 1.1.3 Storage and I/O 20
- 1.1.4 File System 20
- 1.2 User Environment 22
- 1.3.1 Parallel Applications 23
- 1.3.2 Serial Applications 23
- 1.4 Run-Time E nvironment 24
- 1.4.4 HP-MPI 25
- Using the System 27
- 2.2.1 Notes on Modulefiles 28
- 2.2.2 Supplied Modulefiles 29
- 2.2.6 Loading a Modulefile 30
- 2.2.7 Unloading a M odulefile 31
- 2.2.8 Modulefile Conflicts 31
- 2.2.9 Creating a Mod u lefile 32
- 2.3.1 Introduction 33
- 2.3.5 Launching Jobs 34
- $ man sinfo 39
- $ man -k keyword 39
- Using the System 2-13 39
- Developing Applications 41
- 3.2 Using Compilers 42
- 3.4 Interrupting a Job 43
- 3-4 Developing Applications 44
- Developing Applications 3-5 45
- 3-6 Developing Applications 46
- Developing Applications 3-7 47
- $ module load mpi 48
- $ mpicc -o mycode hello.c 48
- $ mpif90 -o mycode hello.f 48
- 3-8 Developing Applications 48
- 3.8 Developing Libraries 49
- 3.9 Advanced To pics 50
- 3-14 Deve loping Applications 54
- Developing Applications 3-15 55
- Debugging Applications 57
- 4-2 D ebugging Applications 58
- $ totalview 60
- 4-4 D ebugging Applications 60
- Debugging Applications 4-5 61
- 4-6 D ebugging Applications 62
- -verbosity %V %F" 63
- Debugging Applications 4-7 63
- 4-8 D ebugging Applications 64
- Etnus TotalView Version# 65
- Debugging Applications 4-9 65
- $ mpirun -srun -n2 Psimple 66
- 4-10 Debugging Applications 66
- $ squeue 67
- $ scancel --user username 67
- Debugging Applications 4-11 67
- Tuning Applications 69
- Using SLURM 71
- $ srun -N 16 -b myscript.sh 73
- $ srun -A -N 4 myscript.sh 73
- $ srun -a 6543 -j 73
- Using SLURM 6-3 73
- 6.4.2 srun Signal Handling 74
- 6.4.3 srun Run-Mode Options 74
- Using SLURM 6-5 75
- 6-6 Using SLURM 76
- 6.4.5 srun Control Options 77
- 6.4.6 srun I/O Options 78
- Using SLURM 6-9 79
- 6.4.7 srun Constraint Options 80
- Using SLURM 6-11 81
- 6.4.9 Using srun with HP-MPI 82
- 6.4.10 Using srun with LSF 82
- 6.8 Job Accounting 84
- 6.9 Fault Tolerance 84
- 6.10 Security 84
- Using LSF 85
- 7.1.2 Topology Support 86
- 7.1.3 Notes on LSF-HPC 87
- SLURM_JOBID=53 89
- SLURM_NPROCS=4 89
- 7-6 Using LSF 90
- 7.3.1 Getting Status of LSF 91
- $ bhosts 92
- 7-8 Using LSF 92
- 7.4 Submitting Jobs 93
- 7.4.3 Submitting a Serial Job 97
- Using LSF 7-17 101
- 7-18 Using LSF 102
- Using HP-MPI 111
- 8.3.3 Using srun with HP-MPI 114
- 8.3.4 Using LSF and HP-MPI 116
- 8.3.5 MPI Versioning 117
- 8.6 Truncated Messages 118
- 8.5 32-Bit B uilds on XC4000 118
- Using HP-MPI 8-9 119
- 8.9 Environment Variables 120
- 8.9.5 MPI_PIN_PERCENTAGE 121
- 8.9.6 MPI_PAGE_ALIGN_MEM 121
- 8.9.7 MPI_MAX_WINDOW 121
- 8.9.8 MPI_ELANLOCK 121
- 8.9.9 MPI_USE_LIBELAN 121
- 8.9.10 MPI_USE_LIBELAN_SUB 122
- Work-arounds 124
- Using HP MLIB 125
- 9.1.2 MLIB and Module Files 126
- 9.2.1 Platform Sup port 126
- 9.2.2 Library Support 126
- 9.2.3 MPI Parallelism 126
- 9.2.4 Modulefiles and MLIB 127
- 9.2.6 Compiling and Linking 127
- 9.2.7 Licensing 128
- 9.2.8 MLIB Man pages 128
- 9.3.1 Platform Sup port 128
- 9.3.2 Library Support 128
- 9.3.3 MPI Parallelism 129
- 9.3.4 Modulefiles and MLIB 129
- 9.3.5 Compiling and Linking 129
- 9.3.6 Licensing 130
- 9.3.7 MLIB Man pages 130
- Advanced Topics 131
- Examples 135
- $ lshosts 140
- Show the environment: 142
- Run the job: 142
- View the running job: 142
- View the finished job: 142
- Examples A-9 143
- Glossary 145
- Glossary-2 146
- Glossary-3 147
- Glossary-4 148
- Glossary-5 149
- Glossary-6 150
Produits connexes et manuels pour Logiciel Hp XC System 2.x Software
(41 pages)
(81 pages)© 2020, manymanuals.fr. Tous droits réservés | 0.044 s |
Manymanuals.com
Manymanuals.de
Manymanuals.fr
Manymanuals.it
Manymanuals.pl
Manymanuals.cz
Manymanuals.es
Manymanuals-pt.com
Commentaires sur ces manuels