Date: Mon, 12 Mar 2012 02:21:40 GMT From: fork <forkandwait@gmail.com> To: freebsd-gnats-submit@FreeBSD.org Subject: ports/165960: Important SVD bug in Lapack 3.4 -- PLEASE patch port Message-ID: <201203120221.q2C2LeLN093949@red.freebsd.org> Resent-Message-ID: <201203120230.q2C2UCxc058257@freefall.freebsd.org>
next in thread | raw e-mail | index | archive | help
>Number: 165960 >Category: ports >Synopsis: Important SVD bug in Lapack 3.4 -- PLEASE patch port >Confidential: no >Severity: serious >Priority: high >Responsible: freebsd-ports-bugs >State: open >Quarter: >Keywords: >Date-Required: >Class: sw-bug >Submitter-Id: current-users >Arrival-Date: Mon Mar 12 02:30:11 UTC 2012 >Closed-Date: >Last-Modified: >Originator: fork >Release: 9.0 Release >Organization: >Environment: FreeBSD mustafa 9.0-RELEASE FreeBSD 9.0-RELEASE #0: Tue Jan 3 07:15:25 UTC 2012 root@obrian.cse.buffalo.edu:/usr/obj/usr/src/sys/GENERIC i386 >Description: The latest lapack has a bug in the way it handles something related to singular value decomposition (SVD). A LOT of algorithms depend on SVD lapack downstream, so this is pretty important to fix. Especially as it impacts my dissertation code ;) Here is the Lapack bug (they might understate its importance): http://icl.cs.utk.edu/lapack-forum/archives/lapack/msg01219.html Thanks! >How-To-Repeat: I encounter the problem in Octave. So... run octave, and you get the following. You should NOT see an error in line #2: 1> aa = rand(2,4) a = 0.67848 0.87562 0.32506 0.53669 0.42031 0.70204 0.37300 0.34772 2> svd(a) ** On entry to DORGLQ parameter number 5 had an illegal value error: svd: exception encountered in Fortran subroutine dgesvd_ >Fix: >Release-Note: >Audit-Trail: >Unformatted:
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201203120221.q2C2LeLN093949>