Re: Compile NAMD on Bluegene

From: Dong Luo (us917_at_yahoo.com)
Date: Thu Jun 15 2006 - 14:16:30 CDT

When I type charmrun, it says:
you must set BL_HOME environment variable.
Setting BL_HOME to any value including
/bgl/BlueLight/ppcfloor/, the result changed to:
charmrun: line 39: [: -nt: unary operator expected
Running on 1 processors:
charmrun: line 47: /BlueLight/bin/CIORUN: No such file
or directory
charmrun: line 47: /BlueLight/bin/CIORUN: cannot
execute: No such file or directory

I googled CIORUN, it seems belong to BGLsim. Since I
am running program on real BG/L, why should it still
try to use CIORUN?

NAMD compilation is finished after removing -ldl from
BlueGeneL.tcl and using fftw within machine. (Newly
compiled fftw gives an error: undefined reference to
`__ctype_b_loc') Howerver, when I test with just
typing namd2, it tells me: illegal instruction.

Will this comilation be really useable?

--- Gengbin Zheng <gzheng_at_ks.uiuc.edu> wrote:

>
> I don't remember we used BGL_HOME, I thought it
> should be BL_INSTALL
> it should be the path prefix that has bglsys/include
> and bglsys/lib,
> typically it is /bgl/BlueLight/ppcfloor/
>
> Gengbin
>
> Dong Luo wrote:
>
> >This Blue Gene machine is located at Boston
> >University. As far as I know, NAMD is not
> installed.
> >Since NAMD requires single precision fftw to do the
> >PME calculation and I found the preinstalled fftw
> with
> >the machine has this version, I will use that fftw
> >first. It it failed again, I will try to compile it
> as
> >Gengbin suggested.
> >I have another question as mentioned in previous
> post,
> >ie, what BL_HOME variable should I define to be
> able
> >to use charm++?
> >
> >--- Gengbin Zheng <gzheng_at_ks.uiuc.edu> wrote:
> >
> >
> >
> >>I think most Blue Gene machines already have NAMD
> >>installed, you may
> >>want to find out.
> >>
> >>To compile your own version, please check out the
> >>following web site for
> >>how to compile fftw for NAMD
> >>
> >>
> >>
>
>http://www.ks.uiuc.edu/Research/namd/wiki/index.cgi?NamdCompile
> >
> >
> >>Gengbin
> >>
> >>Mgr. Lubos Vrbka wrote:
> >>
> >>
> >>>hi,
> >>>
> >>>
> >>>
> >>>>This time, after compiled for a while, the new
> >>>>
> >>>>
> >>problem
> >>
> >>
> >>>>came out as:
> >>>>"src/ComputePme.C", line 8.10: 1540-0836 (S) The
> >>>>#include file <sfftw.h> is not found.
> >>>>"src/ComputePme.C", line 9.10: 1540-0836 (S) The
> >>>>#include file <srfftw.h> is not found.
> >>>>I checked the fftw directory and find out there
> >>>>
> >>>>
> >>are
> >>
> >>
> >>>>only fftw.h and rfftw.h in the include
> directory.
> >>>>As the fftw libary are downloaded from the
> >>>>
> >>>>
> >>website as
> >>
> >>
> >>>>suggested (I downloaded the file
> >>>>fftw-bluegene.tar.gz), I now have no idea how to
> >>>>
> >>>>
> >>fix
> >>
> >>
> >>>>this problem.
> >>>>Please give some hints. Thanks.
> >>>>
> >>>>
> >>>i don't have any experience with namd
> compilation,
> >>>
> >>>
> >>but this error
> >>
> >>
> >>>means that you are trying to use single precision
> >>>
> >>>
> >>version of the fftw
> >>
> >>
> >>>binary that is apparently missing on your
> >>>
> >>>
> >>computer. maybe there is
> >>
> >>
> >>>some switch you can use to specify that you want
> >>>
> >>>
> >>to use double
> >>
> >>
> >>>precision fftw instead of single precision (in
> >>>
> >>>
> >>configure scripts the
> >>
> >>
> >>>switch is usually --enable-float/--disable
> float).
> >>>
> >>>
> >>i don't know what
> >>
> >>
> >>>is the situation for namd, though...
> >>>
> >>>regards,
> >>>lubos
> >>>
> >>>
> >>>
> >
> >
> >__________________________________________________
> >Do You Yahoo!?
> >Tired of spam? Yahoo! Mail has the best spam
> protection around
> >http://mail.yahoo.com
> >
> >
>
>

__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:43:43 CST