Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 25 Jan 2007 09:25:47 -0600
From:      Rick Voland <rpvoland@facstaff.wisc.edu>
To:        vittorio <vdemart1@tin.it>
Cc:        freebsd-ports@freebsd.org
Subject:   Re: math/R eats up all.
Message-ID:  <45B8CBFB.6090001@facstaff.wisc.edu>
In-Reply-To: <200701241944.53346.vdemart1@tin.it>
References:  <20070117.180355.02238085.chat95@mac.com> <20070124.094023.04505493.chat95@mac.com> <20070124.103041.34613013.chat95@mac.com> <200701241944.53346.vdemart1@tin.it>

next in thread | previous in thread | raw e-mail | index | archive | help
Can anyone else confirm a problem or lack of problem building math/R 
with the build option for math/atlas using the new gcc42 ?

Last night, I rebuilt math/blas and math/atlas successfully using their 
default settings.  I have had no problem with them.

Then, I built math/R selecting all the options in the options window 
(letterpaper, dvi manuals, atlas).  I know from previous attempts that 
math/R builds successfully on this machine using the latest gcc42 with 
all options except when I select math/atlas.  When I delete 
/var/db/ports/R/options (to reset the options), math/atlas is not 
automatically selected.

math/R built up to the point of starting lapack stuff (right after 
graphics devices), and then stops building, but R continues to use 100% 
CPU cycles.  I allowed it go for two hours with no more progress.  R 
(without atlas) would complete in maybe 20 minutes.  The computer 
remained responsive, but I had to kill R.

math/R built successfully with math/atlas on this computer before the 
change to gcc42.  In my make.conf, I identify the CPU as Pentium-M, but 
I have no optimization settings in make.conf.

Thanks,

Rick Voland
rpvoland@facstaff.wisc.edu


vittorio wrote:
> I helped the mantainer to update  R to both  2.4.0 and to 2.4.1 (ports/107638, 
> still waiting to be committed, long overdue).
> 
> I can't understand the point! 
> On a pentium 4 notebook I hadn't the slightest problem in compiling the latest  
> R, atlas, blas and lapack.  
> As a matter of fact R requires 'blas' as a dependency as well as 'atlas'. So 
> if you compile/update  atlas (i did it yesterday!) by default blas should be 
> compiled and installed.
> 
> I suggest  to install the following packages with the default configuration in 
> this order: 
> 1)  blas
> 2) atlas
> 3) R
> 
> Let me know....
> 
> Ciao
> Vittorio
> 
>  
> Alle 01:30, mercoledì 24 gennaio 2007, NAKATA Maho ha scritto:
>> Hi,
>>
>> I recieved some message that we cannot build math/R.
>>
>> If one use ATLAS instead of BLAS/LAPACK, build of math/R stops at:
>>> mkdir ../../../../library/grDevices/libs
>> and then hangs with 100% CPU usage.
>>
>> I don't know how I fix it sorry.
>> -- NAKATA, Maho (maho@FreeBSD.org)
>> _______________________________________________
>> freebsd-ports@freebsd.org mailing list
>> http://lists.freebsd.org/mailman/listinfo/freebsd-ports
>> To unsubscribe, send any mail to "freebsd-ports-unsubscribe@freebsd.org"
> _______________________________________________
> freebsd-ports@freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-ports
> To unsubscribe, send any mail to "freebsd-ports-unsubscribe@freebsd.org"
> 


-- 
Rick Voland
rpvoland@facstaff.wisc.edu



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?45B8CBFB.6090001>