[Wien] seg fault in lapw1 on Opteron with PGI-6.0 / acml

[Peter Blaha]

As was mentioned, dcopy is a BLAS routine and thus comes from the ACML.
When I tested the pgi compileron an Intel-Xeon system, version 5.x was fine, 
but 6.0 had several problems.
Try the libgoto library (search the web).

> We are trying to compile wien2k-5.3 on Linux-Debian running on Opterons. We
> are using PGI-6.0 and the ACML that ships with it. The code compiles, and
> init/lapw0 work fine, but lapw1 gives a segmentation fault. We have tried
> compiling with optimization set to -O3 and -O0. Turning on the debug flag -g
> and using gdb, we get the following message indicating a problem in
> "dcopy_()":
> 
> (gdb) set args lapw1.def
> (gdb) run
> Starting program: /usr/use/amd64-linux/wien2kpgi6-05.3/bin/lapw1 lapw1.def
> 
> Program received signal SIGSEGV, Segmentation fault.
> 0x0000000000448e1a in dcopy_ ()
> (gdb) where
> #0  0x0000000000448e1a in dcopy_ ()
> #1  0x000000000040c75c in dscgst (n=781, ab=0x2a96a62020, ldab=781,
> abdiag=0x815890,
>    panelb=0x2a975c4020, info=0) at dscgst.f:111
> #2  0x000000000043fda5 in seclr4 () at seclr4_tmp_.F:203
> #3  0x0000000000408f35 in calkpt () at calkpt_tmp_.F:194
> #4  0x0000000000427d93 in lapw1 () at lapw1_tmp_.F:51
> #5  0x0000000000401e7e in main ()
> (gdb) q
> 
> Has anyone experienced this problem before? What are the compiling options
> recommended for linux / opteron / PGI-6 ?
> 
> Thanks in advance,
> Olivier.
> 
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> 


                                      P.Blaha
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-15671             FAX: +43-1-58801-15698
Email: blaha at theochem.tuwien.ac.at    WWW: http://info.tuwien.ac.at/theochem/
--------------------------------------------------------------------------