IBM Cluster 1350
| Detail |
|---|
System Hardware Configuration 1. IBM Cluster 1350 Computing Nodes - 404 computing nodes.
Each computing node consists of :
- Model - IBM System x3550
- CPU - Intel Woodcrest 3.0 Ghz Dual-Core processor × 2
- Memory - 16 GB PC2-5300 667MHz FBD 240-pin ECC DDR2-SDRAM
- Hard Disk - 3.5" 146.8GB 10k rpm SAS HDD × 2
- network - NetXtreme 1000 T Dual Port Ethernet Adapter × 1
- interconnect - Voltair PCI-E DDR IB HCA × 1
2. IBM ClusterA Computing 1350 Nodes - 128 computing nodes.
Each computing node consists of :
- Model - IBM System x3550
- CPU - Intel X5450 3.0GHz Quad core processor × 2
- Memory - 16 GB PC2-5300 667MHz FBD 240-pin ECC DDR2-SDRAM
- Hard Disk - 3.5" 146.8GB 15k rpm SAS HDD × 2
- network - 1000 T Dual Port Ethernet Adapter × 1
3. Storage Nodes - 8 storage nodes . Each storage node consists of :
- Model - IBM System x3650
- CPU - Intel Woodcrest 3.0 Ghz Dual-Core processor × 2
- Memory - 8 GB PC2-5300 667MHz FBD 240-pin ECC DDR2-SDRAM
- Hard Disk(Internal) - 3.5" 146.8GB 10k rpm SAS HDD x 2
- HD Card - 4Gbps PCI-E Dual Port FC Adapter × 2
- Hard Disk(External) - 146.8GB × 256 (total)
- network Card - IBM 10 Gb E SR Server Adapter × 1
4. Network
- TCP/IP : Force10 E600i 7-slot Chassis Bundle Switch × 2
- interconnect : Voltaire 9288 InfiniBand Switch × 2
5. Major Directories
- /u1 : 544.9 GB
- /u2 : 1.06 TB
- /package : 544.9 GB
- /work1: 2.13 TB
- /work2: 2.13 TB
- /work3: 4.36 TB
- /work4: 4.36 TB
- /scratch : 136 GB * 404 nodes for IBM Cluster 1350
- /scratch : 136 GB * 128 nodes for IBM Cluster 1350A
6. Job Submission
- Interactive: yes (1 hrs limits)
- Batch: LoadLevel
- 1350 LoadLeveler使用簡介
- 1350 A LoadLeveler使用簡介
Software
1. System Software
- SuSE Enterprise 9.0 2CPUs HPC Client Version
- IBM Cluster System Management(CSM) for Linux
- IBM GPFS Linux V3.1
- IBM Tivoli Workload Scheduler LoadLeveler
- Intel C/C++ Compiler for Linux
- Intel Fortran Compiler for Linux
2. Application Packages
Most of application packages on system are placed in /package directory.
You may get further information for application package list at
NCHC Cluster 1350& 1350A User Computing Environment
1. Hardware configuration
2. Queue Policy
- IBM Cluster 1350
|
Queue Name
|
Max CPUs
|
Max CPUs/job
|
PRIO
|
CPU/Run time limit/job
|
對象
|
|---|---|---|---|---|---|
|
test
|
32
|
1 - 32
|
1
|
1 hour
|
一般計畫/
專用計畫 |
|
serial
|
48
|
1
|
1
|
7 days
|
|
|
8cpu
|
1048
|
2 - 8
|
2
|
4 days
|
|
|
32cpu
|
1024
|
9 - 32
|
3
|
4 days
|
|
|
64cpu
|
1024
|
33 - 64
|
4
|
2 days
|
|
|
mono
|
1536
|
1 - 128
|
5
|
2 days
|
專用計畫
|
- IBM Cluster 1350A
|
Queue Name
|
Max CPUs
|
Max CPUs/job
|
PRIO
|
CPU/Run time limit/job
|
對象
|
|---|---|---|---|---|---|
|
e8cpu
|
1024
|
1 - 8
|
1
|
5 days
|
一般計畫/
專用計畫 |
|
e64cpu
|
1024
|
9 - 64
|
2
|
3 days
|
|
|
e128cpu
|
1024
|
65 - 128
|
3
|
2 days
|
主機 Max CPUs 2048
一般卅專用 使用者有 max jobs/user = 6，max jobs/group = 20
max cpus/user = 256 限制
3. Disk usage Constrained Policy
- /tmp : clear files not accessed over 1 day.
- /work1 : clear files not accessed over 2 week.
- /work2 : clear files not accessed over 2 week.
- /work3 : clear files not accessed over 2 week.
- /work4 : clear files not accessed over 2 week.
- /scratch : clear files not accessed over 1 day.( only apply in computing node )
ATTN: due to disk space limitation, please use /work* instead of /usr/tmp or /tmp
ATTN: NCHC has no obligation to backup files under these working file systems, and reserves the right to change/reconfigure/destroy these working file systems without any backup.
4. Compilers
C/C++ Compiler
- Intel C/C++ 9.1 (icc, icpc)
- GNU GCC Collection 3.3.3 (gcc, g++)
- PGI C/C++ 6.2 (pgcc, pgCC)
Fortran
- Intel Fortran 9.1 (ifort)
- GNU GCC Collection 3.3.3 (g77)
- PGI PGF77 6.2 (pgf77,pgf90)
5. Libraries
- MPICH compiled with GCC Compiler
- MPICH compiled with Intel Compiler
- MPICH compiled with PGI Compiler
6. Queuing System
LoadLeveler 3.4.0
7. File System
IBM GPFS Linux V3.1
NCHC Consulting Window
Mr. Sheng Chung Lu (03)5776085 # 205 
a00scl00@nchc.org.tw
