R devel - Nov 2001 - R-devel & ATLAS generates Dr. Watson on NT (was RE: Look, Wa tson! La.svd & ATLAS)

Prof. Bates & R-devel,

I've done more test with the following results:

I have two versions of ATLAS 3.2.1.  One was compiled on my old Thinkpad
600E (PII), the other was compiled on my new Thinkpad T22 (PIIISSE1).

I compiled R-devel dated 10/31, 11/01 and 11/04, linked against either of
the two ATLAS libs.  All gave Dr. Watson when given this code:

  La.svd(matrix(runif(1e5), 1e3, 1e2))

(There's no problem with smaller matrices, say 100x100.)

When compiled without ATLAS, the above code ran fine in all three R-devel's.

I tried R-1.3.1 with both ATLAS libs, and both ran fine.

Any ideas on what could be wrong?  Seems like R-devel doesn't like ATLAS, at
least on WinNT...

Regards,
Andy


-----Original Message-----
From: Douglas Bates [mailto:bates@stat.wisc.edu]
Sent: Friday, November 02, 2001 9:27 AM
To: Liaw, Andy
Cc: 'r-devel@stat.math.ethz.ch'
Subject: Re: [Rd] Look, Watson! La.svd & ATLAS


"Liaw, Andy" <andy_liaw@merck.com> writes:
> Dear R-devel,
> 
> I had attempted to compile r-devel (dated Oct. 31, 2001) on WinNT with
link
> to ATLAS, with mostly success.  However, when I tried the following, I got
a
> visit from Dr. Watson:
> 
> 
> R : Copyright 2001, The R Development Core Team
> Version 1.4.0 Under development (unstable) (2001-10-31)
> 
> R is free software and comes with ABSOLUTELY NO WARRANTY.
> You are welcome to redistribute it under certain conditions.
> Type `license()' or `licence()' for distribution details.
> 
> R is a collaborative project with many contributors.
> Type `contributors()' for more information.
> 
> Type `demo()' for some demos, `help()' for on-line help, or
> `help.start()' for a HTML browser interface to help.
> Type `q()' to quit R.
> 
> > set.seed(1)
> > x <- matrix(runif(1e4), 1e2, 1e2)
> > invisible(La.svd(x))                         # This ran fine.
> > x <- matrix(runif(1e5), 1e3, 1e2)
> > invisible(La.svd(x))
> 
> (Dr. Watson says there's an access violation.)
I was able to run those examples without ill effects on an R-1.4.0
(development) (2001-10-30) with Atlas 3.2.1 on a Linux system (Debian
testing)
> set.seed(1)
> x <- matrix(runif(1e4), 1e2, 1e2)
> invisible(La.svd(x))
> x <- matrix(runif(1e5), 1e3, 1e2)
> invisible(La.svd(x))
> x <- matrix(runif(1e4), 1e2, 1e2)
> system.time(invisible(La.svd(x)))
[1] 0.04 0.00 0.04 0.00 0.00
> x <- matrix(runif(1e5), 1e3, 1e2)
> system.time(invisible(La.svd(x)))
[1] 1.15 0.01 2.39 0.00 0.00
> system.time(La.svd(x, nu = 0, nv = 0))
[1] 0.33 0.00 0.63 0.00 0.00
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.
-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._.
_._

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

On Tue, 6 Nov 2001, Liaw, Andy wrote:
> I've done more test with the following results:
>
> I have two versions of ATLAS 3.2.1.  One was compiled on my old Thinkpad
> 600E (PII), the other was compiled on my new Thinkpad T22 (PIIISSE1).
>
> I compiled R-devel dated 10/31, 11/01 and 11/04, linked against either of
> the two ATLAS libs.  All gave Dr. Watson when given this code:
>
>   La.svd(matrix(runif(1e5), 1e3, 1e2))
>
> (There's no problem with smaller matrices, say 100x100.)
>
> When compiled without ATLAS, the above code ran fine in all three
R-devel's.
>
> I tried R-1.3.1 with both ATLAS libs, and both ran fine.
>
> Any ideas on what could be wrong?  Seems like R-devel doesn't like
ATLAS, at
> least on WinNT...
I have no idea.  Quite possible though.  As we no longer have an active
Windows maintainer, it's down to you.  (Very little testing has been done
on Windows since 1.3.1 was released.)  You need to build a debug version
and look at the dump with gdb to start to debug the problem.

It's not realistic to expect R-devel to work on minority platforms all the
time.  These days it is unrealistic to expect much support on Windows
at all.

-- 
Brian D. Ripley,                  ripley@stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272860 (secr)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._