[mesa-users] MESA compilation on BG/P with XL compilers

[Ben Brown]

Bill, Sean, et al.,
      This problem of the head nodes and the compute nodes having different 
compile environments (sometimes even different operating systems!) is 
fairly common to supercomputers that run light-weight kernels on the 
compute nodes.  It's always been a problem with the Bluegene systems, and 
was a problem on systems like NICS/Kraken, etc.; this is one of the 
reasons that getting a SDK to work on those systems is a bear, and each 
system usually ends up with a unique and "special" solution for compiling.

Some of the more modern systems (TACC/Stampede, if I remember rightly.
And maybe NASA/Pleiades?) have full operating systems on the compute 
nodes, so for those you can compile on the head node without doing 
anything fancy.

--Ben


> Message: 5
> Date: Tue, 29 Oct 2013 13:37:03 -0700
> From: Bill Paxton <paxton at kitp.ucsb.edu>
> Subject: Re: [mesa-users] MESA compilation on BG/P with XL compilers
> To: Sean Couch <smc at flash.uchicago.edu>
> Cc: "mesa-users at lists.sourceforge.net group"
> 	<mesa-users at lists.sourceforge.net>
> Message-ID: <AC4BAAF3-4B7B-4564-B2A1-B5302C5331D1 at kitp.ucsb.edu>
> Content-Type: text/plain; charset=us-ascii
>
> Hi Sean,
>
> Josiah's solution (compile on a compute node) would certainly be the best way to go if it is possible.   But if you can't run the compiler on a compute node, then turning off the "ck" scripts during compilation will work.  If you turn off the tests during compilation, then you will of course need to run them all later on a compute node.  To do that, you could edit the "each_package_do" script to make the do_one function do the following:
> 	cd test; ./ck; check_okay
>
> -Bill
>
>
>
> On Oct 29, 2013, at 12:29 PM, Josiah Schwab wrote:
>
>> Hi Sean,
>>
>>> I am trying to compile MESA on Intrepid, a BG/P at Argonne.  For this
>>> system, the head nodes have a different architecture than the compute
>>> nodes.  I am hitting a stumbling block when the MESA compilation
>>> script attempts to run the post-build tests for the const module
>>> because, naturally, the head node cannot execute the binary which has
>>> been built for the compute nodes.
>>
>> This is not an answer to your question, but have you considered
>> submitting the MESA compilation as a job (so it runs on the compute
>> nodes)?  That might provide a quick & easy workaround.
>>
>> If that's no good, you can probably turn off the tests by dropping in
>> versions of the ck scripts in utils/test that simply don't do anything.
>>
>> Josiah
>>
>> ------------------------------------------------------------------------------
>> Android is increasing in popularity, but the open development platform that
>> developers love is also attractive to malware creators. Download this white
>> paper to learn more about secure code signing practices that can help keep
>> Android apps secure.
>> http://pubads.g.doubleclick.net/gampad/clk?id=65839951&iu=/4140/ostg.clktrk
>> _______________________________________________
>> mesa-users mailing list
>> mesa-users at lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/mesa-users
>
>
>
>
> ------------------------------
>
> ------------------------------------------------------------------------------
> Android is increasing in popularity, but the open development platform that
> developers love is also attractive to malware creators. Download this white
> paper to learn more about secure code signing practices that can help keep
> Android apps secure.
> http://pubads.g.doubleclick.net/gampad/clk?id=65839951&iu=/4140/ostg.clktrk
>
> ------------------------------
>
> _______________________________________________
> mesa-users mailing list
> mesa-users at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/mesa-users
>
>
> End of mesa-users Digest, Vol 50, Issue 79
> ******************************************
>