[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: 20041124: Ack! netcdf-3.6.0_beta6 fails on ordinary RedHat system w/PGI

"David Pierce" <address@hidden> writes:

> >> I'm trying to upgrade to netcdf-3.6 because I need the large file
> >> support,
> >> but it failed to build on my system (a fairly ordinary dual-xeon
> >> workstation with RH WS 3.1 or so, and PGI compilers).  I'm attaching the
> >> entire output of the process of configure, make, make test.  The C and
> >> f77
> >> tests seemed to pass, while the pgf90 test failed.  I see from your
> >> collected output of builds on various platforms that a very similar
> >> system
> >> (sunshine) did pass this test.  Any idea what version of the PGI
> >> compilers
> >> that machine has?
> >>
> >> The versions of the PGI compilers I am using are 5.1-3.
> 
> make[2]: Entering directory 
> `/home/pierce/src/netcdf/netcdf-3_6_0-beta6/src/f90'
> pgf90 -o netcdf_test   netcdf_test.f90 netcdf.o typeSizes.o 
> ../libsrc/libnetcdf.a 
> netcdf_test.f90:
> ./netcdf_test
> example_good.cdl example.cdl differ: byte 881, line 37
> *** Failure ***
> *** example_good.cdl  2000-04-05 19:33:14.000000000 +0000
> --- example.cdl       2004-11-24 20:37:58.000000000 +0000
> ***************
> *** 34,43 ****
>     953, 954, 955,
>     956, 957, 958,
>     959, 960, 961,
> !   962, 963, 964,
> !   965, 966, 967,
> !   968, 969, 970,
> !   971, 972, 973 ;
>   
>    lat = -90, -87.5, -85, -82.5 ;
>   
> --- 34,43 ----
>     953, 954, 955,
>     956, 957, 958,
>     959, 960, 961,
> !   950, 951, 952,
> !   953, 954, 955,
> !   956, 957, 958,
> !   959, 960, 961 ;
>   
>    lat = -90, -87.5, -85, -82.5 ;

This problem is caused by a bug in the Portland Group compiler. It is
fixed in version 5.2-2, which is what we are running.

I believe that they also offer a patch o 5.1, but I've never tried
that. Presumably if they fixed the bug, which has to do with array
slices, then it should work too.

So either upgrade or get the patch, and try again (make sure you do a
make clean first). Let me know if that doesn't work for you.

Ed Hartnett
Unidata