problem with accessing and saving data into a data file
MMT
I had a problem with accessing and saving data into a data file when I
am running a big parallelised MPI code on supercomputer (Blue C). I am
using MPI and Fortran90 in a windows environment.
I wrote very short samples from my code which needs to read coil data,
volume etc. and write huge sensitivity matrix data and eddy current
data as output.
My code parallelised well in cluster computers and gave correct
results. I moved them to supercomputer and compiled using Makefile and
created an executive file. And then I submitted the task using
Loadleveler script. But the output result is always 0 and data file
that I intended to save was not changed at all.
I saved the data in a MODULE as below:
CHARACTER (LEN = 105), PARAMETER ::
FileNameVolume='/hpc/home/yasheng/ParaJJCoil/Results/Volume.dat'
CHARACTER (LEN = 105), PARAMETER ::
FileNameSensMatrix='/hpc/home/yasheng/ParaJJCoil/Results/SensMatrix.dat'
In the PROGRAM and SUBPROGRAM, I read the data and saved it as below
OPEN(8,file=FileNameSensMatrix , form='formatted',status='unknown')
WRITE(8,300)(Sensmatrix2)
CLOSE(8)
I tried to search information from internet, but I could not find a
clue about it.
I would appreciate very much if you could tell me how it would be a
correct way of reading and writing data when running the code in a
Supercomputer environment?
Thank you very much,
Yasheng
Joachim Worringen
> I would appreciate very much if you could tell me how it would be a
> correct way of reading and writing data when running the code in a
> Supercomputer environment?
MPI-IO usually scales better than Fortran I/O for large codes/machines/data. Not
that it would be easier to use, though...
--
Joachim - reply to joachim at domain ccrl-nece dot de
Opinion expressed is personal and does not constitute
an opinion or statement of NEC Laboratories.
Joachim Worringen
> MMT wrote:
>
>> I would appreciate very much if you could tell me how it would be a
>> correct way of reading and writing data when running the code in a
>> Supercomputer environment?
>
> MPI-IO usually scales better than Fortran I/O for large
> codes/machines/data. Not that it would be easier to use, though...
But you can learn something about it i.e. at http://www.pvmmpi06.org/index.php?id=25