CFD-MHD PC-Cluster

Since 2004/05/13

Cluster CPU-Loading

http://mhda.asiaa.sinica.edu.tw/cpu-monitoring.html
The new monitoring system

CFD-MHD PC cluster Quick User Guide and MHDHELP Archive

Who am I ?

COMPUTATIONALFLUID DYNAMICS	<-- Pure Fluid
Pure Fluid + E + B -->	MAGNETOHYDRODYNAMICS

2004 pannel poster , 2005 pannel poster

What am I ?

8 single-cpu cluster + 17 dual-cpu cluster

Single cpu nodes

mhd1 : P4 2.67GHz, 2GB DDR266, 60GB+120GB+120GB HD
mhd2 : P4 2.67GHz, 2GB DDR266, 60GB+120GB+120GB HD
mhd3 : P4 2.67GHz, 2GB DDR266, 60GB+120GB+120GB HD
mhd4 : P4 2.67GHz, 2GB DDR266, 60GB+120GB+120GB HD
mhd5 : P4 2.67GHz, 2GB DDR266, 80GB+120GB+120GB HD
mhd6 : P4 2.66GHz, 2GB DDR, 80GB+120GB+120GB HD
mhd7 : P4 2.53GHz, 2GB DDR, 80GB+120GB+120GB HD
mhd8 : P4 2.53GHz, 2GB DDR, 120GB+120GB+120GB HD
mhds9 : P4 2.26GHz, 1GB DDR, 80GB HD
mhds10 : P4 64bit 3.6GHz, 3GB DDR, 80GB HD

Dual cpu nodes

mhd9 : Intel/Xeon 2.4GHz, 2GB DDR266, 120GB HD
mhd10: Intel/Xeon 2.4GHz, 2GB DDR266, 120GB HD
mhd11: Intel/Xeon 2.4GHz, 2GB DDR266, 120GB HD
mhd12: Intel/Xeon 2.4GHz, 2GB DDR266, 120GB HD
mhd13: Intel/Xeon 2.4GHz, 2GB DDR266, 120GB HD
mhd14: Intel/Xeon 2.4GHz, 2GB DDR266, 120GB HD
mhd15: Intel/Xeon 2.4GHz, 2GB DDR266, 120GB HD
mhd16: Intel/Xeon 2.4GHz, 2GB DDR266, 120GB HD
mhd17: Intel/Xeon 2.4GHz, 2GB DDR266, 120GB HD
mhd18: Intel/Xeon 2.4GHz, 2GB DDR266, 120GB HD
mhd19: AMD64/Opteron 1.8GHz, 2GB DDR400, 120GB HD
mhd20: AMD64/Opteron 1.8GHz, 2GB DDR400, 120GB HD
mhd21: AMD64/Opteron 1.8GHz, 2GB DDR400, 120GB HD
mhd22: AMD64/Opteron 1.8GHz, 2GB DDR400, 120GB HD
mhd23: AMD64/Opteron 1.8GHz, 2GB DDR400, 120GB HD
mhd24: Intel/Nocona(EM64T) 2.8GHz, 2GB DDR2, 120GB HD
mhd25: Intel/Nocona(EM64T) 2.8GHz, 2GB DDR2, 120GB HD
mhd26: AMD64/Opteron 1.8GHz, 2GB DDR400, 160GB HD
mhd27: Intel Pentium Dual Core 3GHz, 2GB DDR400, 160GB HD
mhd28: Intel Pentium Dual Core 3GHz, 2GB DDR400, 160GB HD

One 2TB IDE raid5 Disk Array ( located at /arrays/mhda of each machines )
One 2TB IDE PVFS for fast dumping simulation data ( mhd1 - mhd8 ) --> not yet
One 400GB IDE PVFS for out-of-core computation ( mhd9 - mhd18 ) --> not yet
two 24-port Gbits switches

Network topology(updated! 2004/12/16)

How am I ?

Running your sequential job on single-cpu machines( mhd1~mhd8 ) and parallel job on dual-cpu machines first !
Use bash as default. Then modify the .bashrc under your home directory! If you are using 64bit machines, use this .bashrc.
A simple way to submit your job --> $ (time -p ./a.out > log ) >& log.cpu & (The messages are saved in log and cpu time saved in log.cpu)
Netperf