[Clas_offline] Decommissioning of 32 bit interactive farms

Thu Mar 10 12:59:30 EST 2011

Hi,
  I have the whole clas package and geant4  complied with -m32 in ifarm64
ever since
the time ifarm64 was set up for test.  I have been running the gsim and
user_ana
in the 64-bit batch farm for years.

In order to make that happen I have modified the makefile structure and add
a few
scripts into the $CLAS_CMS.  I use one env variable 'COMPILER_BIT' to force
the
compiler to add '-m32' or '-m64' into each CFLAGS.  Quite a few program can
not
be compiled but I succeed to modify the alternate makefile or source code or
install the
missing libraries and finally make them work.
The CLASPACK I am using is release-4-15.  A few months ago  I have also
compiled
the release_f8_2  for a student using the same makefile structure.

Please note that I am not using scon because scon showed up too late.
If anyone want to use my make structure, you can simply do the following:

//get into your build dir, which is the one contains packages
cd  $CLAS_PACK/..

//setup the env, you 'd better look into this file to get the idea and set
it up the way you like
source /u/home/jixie/clas/builds/release-4-15/SetCLASEnv_x86_64_m32.csh

//reset CLAS_CMS to use my compile strcture
setenv CLAS_CMS /u/home/jixie/clas/builds/release-4-15/packages/cms

//then compile it, you may need to install missing libs or change codes for
some programs
//check the error message and then compare to mine
cd $CLAS_PACK; make lib

//add one line into your .cshrc
setenv COMPILER_BIT=32

Good luck.

Jixie

On Thu, Mar 10, 2011 at 7:51 AM, Johann Goetz <jgoetz at ucla.edu> wrote:

> I have been using gsim on 64 machines (compiled using the -m32 flag) for
> about a year now. Many months ago, I tested it on 32-bit machine and it gave
> the same results. This is not an issue.
>
> On Thu, Mar 10, 2011 at 12:29 AM, Dennis Weygand <weygand at jlab.org> wrote:
>
>> Yes- I can comment....
>> The plan has been in place for a long time, and not all the CLAS subgroups
>> bothered to do anything about it, in spite of consistent warnings.
>>
>> I would just keep one machine around till we are current.
>>
>> Arman: You can play a key role here- what is the status? Also, the biggest
>> worry is GSIM, which as far as I know was never able to make the transition
>> to 64 bit.
>> Johann, can you update me on the status if GSIM? Isn't it easy to do a 32
>> bit build on a 64 bit machine?
>>
>> Dennis
>>
>>
>>
>>
>> On Mar 9, 2011, at 1:19 PM, Heyes Graham wrote:
>>
>> The plan to decommission the 32-bit nodes has been in place for a long
>> time and I have had no feedback at all from any  of the groups indicating
>> anything that would prevent this. Of course Dennis is at the meeting in
>> Paris this week and not able to comment.
>>
>> If one node was to be kept up how long will it be needed for? I don't yet
>> know if it is possible but if it is that node must only be used as a check
>> of the 64-bit code.
>>
>> Regards to all,
>>  Graham
>>
>>
>> On Mar 9, 2011, at 12:30 pm, Alex Kubarovsky wrote:
>>
>> Hello,
>>
>> I completely agree with Valery and Ivan.
>>
>> Regards, Alex
>>
>> On Wed, Mar 9, 2011 at 11:49 AM, Valery Kubarovsky <vpk at jlab.org> wrote:
>>
>>> Hi All:
>>> I completely support the idea to keep one of our ifarms (3,4,5) alive for
>>> some time.
>>> It will give us the possibility to check the code on 64bits machines.
>>> Again, we are not ready for the complete transition due to missing
>>> programs at new platform.
>>> Valery
>>>
>>>
>>> On Wed, Mar 9, 2011 at 11:36 AM, Ivan Bedlinskiy <bedlinsk at jlab.org>
>>> wrote:
>>>
>>>> Hi Dennis,
>>>>
>>>> According to login notices on ifarms all 32bit interactive farms will be
>>>> decommissioned on Monday, March 14.
>>>>
>>>> Is it possible to keep at least one such farm in production?
>>>>
>>>> The reasons:
>>>>
>>>> 1) The migration to 64 bit systems is not complete. At least some of my
>>>> code is not linking properly on ifarml6. I believe some other people
>>>> will have similar problems.
>>>>
>>>> 2) As far as I know certain CLAS software like CED in not compiling on
>>>> ifarml6
>>>>
>>>> 3) Its better to have backup option and be able to check certain old
>>>> sources in the old environment. When migration from RHEL3 to RHFC8
>>>> occurred it was very inconvenient the old environment was not available.
>>>>
>>>> Regards
>>>> Ivan
>>>>
>>>> _______________________________________________
>>>> Clas_offline mailing list
>>>> Clas_offline at jlab.org
>>>> https://mailman.jlab.org/mailman/listinfo/clas_offline
>>>>
>>>
>>>
>>> _______________________________________________
>>> Clas_offline mailing list
>>> Clas_offline at jlab.org
>>> https://mailman.jlab.org/mailman/listinfo/clas_offline
>>>
>>
>> _______________________________________________
>> Clas_offline mailing list
>> Clas_offline at jlab.org
>> https://mailman.jlab.org/mailman/listinfo/clas_offline
>>
>>
>> _______________________________________________
>> Clas_offline mailing list
>> Clas_offline at jlab.org
>> https://mailman.jlab.org/mailman/listinfo/clas_offline
>>
>>
>>  --
>> Dennis Weygand
>> weygand at jlab.org
>> (757) 269-5926
>>
>>
>>
>>
>
>
> --
> Johann T. Goetz, PhD.
> jgoetz at ucla.edu
> Nefkens Group, UCLA Dept. of Physics & Astronomy
> Hall-B, Jefferson Lab, Newport News, VA
> Office: 757-269-5465 (CEBAF Center F-335)
> Mobile: 757-768-9999
>
>
> _______________________________________________
> Clas_offline mailing list
> Clas_offline at jlab.org
> https://mailman.jlab.org/mailman/listinfo/clas_offline
>

-- 
Best Regards,
Jixie Zhang
Jefferson Lab - Hall A - A112,
757-269-5352
_______________________________________________
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailman.jlab.org/pipermail/clas_offline/attachments/20110310/2e3a8f34/attachment.html 