Re: NEC-LIST: Help with GNEC error message.

Hi,

I am not familiar with GNEC but in general if it is a Method of
Moments code, here are my 2 cents worth.

Have you been monitoring the hard disk <-> processor traffic and your
% free space after setting up the problem?

Looks like there is a possibility that (if it is NOT a Beowulf/Avalon
type of computer ) the hard disk ran out of temporary storage for
those intermediate iterations while solving the matrix.

Can anyone help me figure this out ....
5826 segments means 5826 X 3 = 17478 variables (assuming Ex Ey & Ez
and the segments is not is a dielectric medium, which would require
double the number of variables... ) and so the matrix would take
17478^2 = 3.05480484E+8 matrix location and each location, say if it
is work out with 4 bytes arithmetic, hence the memory requirement is
... (I suppose) 1.221921936E+9 bytes. Then I should say that the hard
disk should have swap space for at least 1.34411413E+9 (1281 MB) based
on a 10% allowance for the operating system.

With 'tight-budget' technology ( not the state of the art, I must add
... ) , I should say that this problem is 'doable' with a 1 GB (ECC)
desktop computer type computer with say a 1GHz uP, with a reasonably
fast front side bus (133 MHz SDRAM should be OK, DDR and RDRAM is a
bit stiff at this moment though) a 20 GB ATA/66 (or if there is
avaliable ATA/100 controller for the ) hard disk.

Tim Foo <fwj_at_np.edu.sg>
ECE Dept, http://www.np.edu.sg/ece/
Ngee Ann Polytechnic,
535 Clementi Road,
Singapore 599489

Tel: + 65 460 6143
Fax: + 65 467 1730

"Richeson, Peter D" <Peter.Richeson_at_OKC.Boeing.com>@ece.ubc.ca on
08/04/2000 08:51:56 PM wrote
>
>
> I am trying to run a large model, 5826 segments in GNEC. I ran it
> over night because I knew it would take several hours to run. When I
> got in this AM, I found the following error:
>
> RECOT : Error writing file 11
> Program Record NO 9 Logical record 13052 I1, I2 =13122736
> Trace write from CMSET... <snip>
Received on Mon Aug 07 2000 - 07:49:31 EDT