[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Help-gnucap] Help-gnucap Digest, Vol 88, Issue 3
From: |
al davis |
Subject: |
Re: [Help-gnucap] Help-gnucap Digest, Vol 88, Issue 3 |
Date: |
Mon, 7 Jan 2013 12:25:12 -0500 |
User-agent: |
KMail/1.13.5 (Linux/2.6.32-5-amd64; KDE/4.4.5; x86_64; ; ) |
On Monday 07 January 2013, Romain GAILLETON wrote:
> 95% of the computing times seem to come from the LU operation
> according to the "status" command.
> I am running Gnucap on Windows using the last available
> version (2009) So I tried to run it on Linux using the 0.35
> version but the improvement was not as good as expected.
> We end up with 5% gain on the computing time using the same
> parameters you gave us earlier. Would you expect this
> result too ?
The issue I mentioned that sometimes the old version is faster
has to do with a readin problem (status "get"). If status is
showing 95% in LU, that's not the problem. If the old version
is any faster in this case, it is probably random slop or
compile options, or maybe a side effect of differences in time
step control.
Since you are getting 95% in LU .. what is "density"? That's
where node ordering comes in. For a passive circuit, it is
normal that most of the time is in LU.