Armstrong, Whit
2006-Oct-19 22:50 UTC
[Rd] lapack tests fail -- is this something to worry about?
I see the comment here about some 64bit combinations failing. Well, my configuration fails. Is this really something to worry about? for R-patched: fi1->cat SVN-REVISION Revision: 39576 Last Changed Date: 2006-10-03 fi1->> ## failed for some 64bit-Lapack-gcc combinations: > sm <- cbind(1, 3:1, 1:3) > eigenok(sm, eigen(sm))Error: abs(A %*% V - V %*% diag(lam)) < Eps is not all TRUE> eigenok(sm, eigen(sm, sym=FALSE))Error: abs(A %*% V - V %*% diag(lam)) < Eps is not all TRUE>I'm not using a nonstandard blas or lapack, just the ones included in R. Here is my system information: fi1->gcc --version gcc (GCC) 3.2.3 20030502 (Red Hat Linux 3.2.3-47) Copyright (C) 2002 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. fi1->uname -a Linux fi1.nyc.office.hcmny.com 2.4.21-32.ELsmp #1 SMP Fri Apr 15 21:03:28 EDT 2005 x86_64 x86_64 x86_64 GNU/Linux fi1-> This is RHEL v3. I know gcc 3.2.3 is old, but I don't have root, and our admins are not crazy about changing 1 system to accommodate 1 user, so I'm happy to ignore the test failure if it's not something to worry about. I'm not using lapack routines in my code, but I don't know to what extent R makes use of lapack internally. Can anyone give me the skinny on whether this test failure really matters? Thanks, Whit This e-mail message is intended only for the named recipient(s) above. It may contain confidential information. If you are not the intended recipient you are hereby notified that any dissemination, distribution or copying of this e-mail and any attachment(s) is strictly prohibited. If you have received this e-mail in error, please immediately notify the sender by replying to this e-mail and delete the message and any attachment(s) from your system. Thank you.
Prof Brian Ripley
2006-Oct-20 11:57 UTC
[Rd] lapack tests fail -- is this something to worry about?
I think it is just that you should bear in mind that your system is less accurate than a more modern one (and lots of compiler bugs have been fixed since gcc 3.2.3). [I am surprised your sysadmins are happy running an OS that is lacking so many of the security features of current Linuxen. Ours are not!] On Thu, 19 Oct 2006, Armstrong, Whit wrote:> I see the comment here about some 64bit combinations failing. > > Well, my configuration fails. Is this really something to worry about? > > for R-patched: > fi1->cat SVN-REVISION > Revision: 39576 > Last Changed Date: 2006-10-03 > fi1-> > > >> ## failed for some 64bit-Lapack-gcc combinations: >> sm <- cbind(1, 3:1, 1:3) >> eigenok(sm, eigen(sm)) > Error: abs(A %*% V - V %*% diag(lam)) < Eps is not all TRUE >> eigenok(sm, eigen(sm, sym=FALSE)) > Error: abs(A %*% V - V %*% diag(lam)) < Eps is not all TRUE >> > > > I'm not using a nonstandard blas or lapack, just the ones included in R. > > > Here is my system information: > > fi1->gcc --version > gcc (GCC) 3.2.3 20030502 (Red Hat Linux 3.2.3-47) > Copyright (C) 2002 Free Software Foundation, Inc. > This is free software; see the source for copying conditions. There is > NO > warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR > PURPOSE. > > fi1->uname -a > Linux fi1.nyc.office.hcmny.com 2.4.21-32.ELsmp #1 SMP Fri Apr 15 > 21:03:28 EDT 2005 x86_64 x86_64 x86_64 GNU/Linux > fi1-> > > > This is RHEL v3. I know gcc 3.2.3 is old, but I don't have root, and > our admins are not crazy about changing 1 system to accommodate 1 user, > so I'm happy to ignore the test failure if it's not something to worry > about. I'm not using lapack routines in my code, but I don't know to > what extent R makes use of lapack internally. > > Can anyone give me the skinny on whether this test failure really > matters? > > Thanks, > Whit > > > > > > This e-mail message is intended only for the named recipient(s) above. It may contain confidential information. If you are not the intended recipient you are hereby notified that any dissemination, distribution or copying of this e-mail and any attachment(s) is strictly prohibited. If you have received this e-mail in error, please immediately notify the sender by replying to this e-mail and delete the message and any attachment(s) from your system. Thank you. > >-- Brian D. Ripley, ripley at 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 272866 (PA) Oxford OX1 3TG, UK Fax: +44 1865 272595