Re: NAMD related

From: Axel Kohlmeyer (akohlmey_at_gmail.com)
Date: Tue Jun 11 2013 - 03:01:12 CDT

On Tue, Jun 11, 2013 at 9:51 AM, Anilkumar Naik
<anilkumar.j.naik_at_gmail.com> wrote:
> Thanks for briefing it.
>
> Hope new version of NAMD come up with NAMD working on PowerPC. If this is

there was a lot of effort spent on optimizing NAMD for ppc7 CPU when
IBM was still part of the blue waters project hosted at NCSA. however,
IBM pulled out, took back all the hardware, gave back the money,
decided to re-market their CPUs for business applications and blue
waters is now an x86_64 based Cray machine with Nvidia GPUs.

you draw your own conclusions what that means about choosing ppc7 as a
platform for scientific computing...

axel.

> more or less some linking issue then will try to take some local help.
>
> Thanks for help.
>
> Regards,
> Anilkumar
>
>
>
> On Tue, Jun 11, 2013 at 12:57 PM, Axel Kohlmeyer <akohlmey_at_gmail.com> wrote:
>>
>> two comments:
>>
>> you *really* should ask for help with a local person that has some
>> knowledge about compiling software on a machine like this. 99% of the
>> problems that you have been discussing here are system/compilation
>> problems and not at all NAMD specific problems. even more so, it is
>> extremely cumbersome to debug this over e-mail. if had found a local
>> person knowledgeable in compiling applications, your issues would have
>> been resolved a long time ago.
>>
>> remember, most people on this mailing list are NAMD *users* and not
>> developers and particularly not people with experience on exotic
>> architectures.
>>
>> that being said, you guess (not unexpectedly) is wrong.
>> the error message is clearly originating from the linker and it is
>> indicating that the /usr/lib64/libtcl8.5.so file is not compatible
>> with your compilation settings. that can have two reasons:
>> a) you are compiling for 32-bit mode, but are trying to link with
>> 64-bit libraries (as the directory name hints at).
>> this can be easily validated by doing: xlc -c test.c ; file test.o
>> the output should indicate the bitness and for more details on how
>> to select compiler bitness you should read the compiler documentation.
>> b) the library is not what it should be. i.e. either it is not 64-bit
>> or it is not for ppc7
>> this can be found out using: file /usr/lib64/libtcl8.5.so
>>
>>
>> again, these kinds of issues are part of the daily work and skill set
>> of a system administrator / system manager and are trivial to verify
>> and resolve unless your local system "person" is incompetent. so
>> please, save yourself a lot of grief and get some capable local help.
>>
>> thanks,
>> axel.
>>
>>
>> On Tue, Jun 11, 2013 at 9:14 AM, Anilkumar Naik
>> <anilkumar.j.naik_at_gmail.com> wrote:
>> > Hi,
>> >
>> > I did testing with a small program:
>> > test.c:
>> > nt main(int argc, char **argv)
>> > {
>> > Tcl_Main(argc, argv, 0);
>> > return 0;
>> > }
>> >
>> > 1)
>> > xlc test.c -L/usr/lib64 -ltcl8.5 -ldl
>> >
>> > /usr/bin/ld: skipping incompatible /usr/lib64/libtcl8.5.so when
>> > searching
>> > for -ltcl8.5
>> > /usr/bin/ld: skipping incompatible /usr/lib64/libtcl8.5.so when
>> > searching
>> > for -ltcl8.5
>> > /usr/bin/ld: cannot find -ltcl8.5
>> >
>> > 2)
>> > $cc test.c -L/usr/lib64 -ltcl8.5 -ldl
>> > $ls
>> > a.out
>> >
>> > 3)
>> > gcc test.c -L/usr/lib64 -ltcl8.5 -ldl
>> > $ls
>> > a.out
>>
>> > May be xlc is skipping incompatible libtcl8.5.so?
>> >
>> > Regards,
>> > Anilkumar
>> >
>>
>>
>> --
>> Dr. Axel Kohlmeyer akohlmey_at_gmail.com http://goo.gl/1wk0
>> International Centre for Theoretical Physics, Trieste. Italy.
>
>

--
Dr. Axel Kohlmeyer  akohlmey_at_gmail.com  http://goo.gl/1wk0
International Centre for Theoretical Physics, Trieste. Italy.

This archive was generated by hypermail 2.1.6 : Tue Dec 31 2013 - 23:23:19 CST