19990426: strange problem with IMGDISP and stretch tables (cont.)

>From: weather <address@hidden>
>Organization: NMSU/NSBF
>Keywords: 199904231716.LAA23829 McIDAS IMGDISP SU=

Robert,

>Well..
>
>Now it seems to work.  The first time I saw this I repeated it about
>20 times to make sure I wasn't typing H20(zero) instead of H2O.
>I was NOT typing zero, but now it works.  I can't explain it.

Gremlins :-)

>It was on our Solaris x86 machine.  I must have made some strange
>typing mistake, but I don't know how I did it 20 times.  I am glad
>you checked it our first.

Me too.

>By the way, have you started loading your Intel machines (Solaris
>or Linux) heavily with the LDM..etc.  If so what do you think?

This is coming up.  We are going to move all of our LDM related
data ingest/decoding (except the CONDUIT stuff) over to one of our
dual 450 Mhz Pentium II machines that is running Solaris x86 2.7.
It will be interesting to benchmark performance of that machine
versus our SPARC Ultra 2 and also to compare ldm-mcidas decoder
performance for executables compiled with Sun SC5.0 versus gcc/f2c
ones.  We have heard someone note that the gcc/f2c combination is
produces slower execution than the native compilers, but we don't
have any hard numbers.  Time to get some...

Later...

Tom