lapackers

[julie langou <julie@cs.utk.edu>]

Please find the message below...

> 

> Dear lapackers,
> It is time to freeze what will go in the LAPACK and ScaLAPACK releases as 
> we need to prepare the posters for SC11.
> Below is a draft of the Release Notes for LAPACK 3.4.0. This contains what 
> is already in.
> Please let me know if you have some codes that will be ready for mid 
> November:
> - [UCD] status of LAPACK QZ, ScaLAPACK QR, ScaLAPACK CAQR
> - [UC Berkley] status of Craig's algo integration
> - [somebody?] ScaLAPACK LDLT
> 
> For ScaLAPACK, Rodney, I would need a detailed section on the change for 
> ScaLAPACK 2.0. Also we need to coordinate the update of the ScaLAPACK 
> installer and the creation of a new CMAKE build system for ScaLAPACK. I am 
> fine working on it, but just let me know.
> Again for Rodney, we need some performance on MRRR. Did you have time to 
> get some, if not, same, I can take care of it.
> 
> Regarding LAPACKE, INTEL did not send us yet the new version. I will 
> contact them.
> Let me know if you have any questions about the next LAPACK and ScaLAPACK 
> releases.
> Regards,
> Julie
> 
> 
> LAPACK 3.4.0
> ============
> 
> Release date:  11/XX/2011.
> 
> This material is based upon work supported by the National Science 
> Foundation
> and the Department of Energy under Grant No. NSF-OCI-1032861 (2010-2013 
> est.),
> NSF-CCF-00444486 (2005-2008 est.), NSF-CNS-0325873 (2004-2009 est.),
> NSF-EIA-0122599 (2001-2004 est.), NSF-ACI-0090127, DOE-DE-FC02-01ER25478,
> DOE-DE-FC02-06ER25768.
> 
> LAPACK is a software package provided by Univ. of Tennessee, Univ. of
> California Berkeley, Univ. of Colorado Denver and NAG  Ltd.
> 
> 1. Support and questions:
> -------------------------
> 
> http://icl.cs.utk.edu/lapack-forum/
> 
> 2. LAPACK 3.4.0: What's new
> ---------------------------
> 
>  1. *FULL Standard C language APIs for LAPACK:* The standard interfaces 
> include
>  support for Fortran and C data formats (column-major and row-major) to ease
>  interoperability with, and migration of, Fortran code. In LAPACK 3.3.0, 
> only
>  LAPACK driver were provided, now 100% of LAPACK functionalities are 
> accessible
>  in a C-friendly manner.  These new interfaces were also just released in 
> Intel
>  MKL XXX product. PLASMA XXX relies on it.
> 
>  2. *New kernels for communication avoiding QR* 
> 
>  3. *New TESTING output analysis*:
> 
>  4. *Doxygen generated documentation*
> 
> 
> 3. External Contributors
> ------------------------
> 
>    * Intel MKL team
> 
> 4. Thanks
> ---------
> 
> Thanks for bug-report/patches/suggestions to:
> 
>    * Sebastien Fabbro
>    * Tobias Burnus
>    * Hatem Ltaief
>    * Yana Gurieva (Intel)
>    * 
> 
> 5. Developer list
> -----------------
> 
> .Principal Investigators
> 
>    * Jim Demmel (University of California at Berkeley, USA)
>    * Jack Dongarra (University of Tennessee and ORNL, USA)
>    * Julien Langou (University of Colorado Denver, USA)
> 
> .LAPACK developers involved in this release
> 
>    * Rodney James (University of Colorado Denver, USA) 
>    * Igor Kozachenko (University of California at Berkeley, USA)
>    * Julie Langou (University of Tennessee, USA)
>    * Benjamin Lipshitz (University of California at Berkeley, USA)
> 
> 6. More details
> ----------------
>                                                           
> 6.1. FULL LAPACKE
> ~~~~~~~~~~~~~~~~~
> 
> .contributor
> 
>    INTEL
>    
> .contribution
> 
> 
> 6.2.  New kernels for communication avoiding QR
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> 
> .lapacker
> 
>    Rodney James (University of Colorado Denver, USA) 
> 
> .contribution
> 
> QR factorization for triangular-pentagonal matrices
> (generalization of triangle-over-square and triangle-over-triangle)
> 
> QRT (QR factorization with compact WY representation for Q) routines for 
> general matrices
> 
> 
> 
> 6.3. New TESTING output analysis
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> 
> .lapackers
> 
>     Julie Langou (University of Tennessee, USA)
> 
> .contribution
> 
> A new python file lapack_testing.sh has been added to the LAPACK folder.
> This script allows you to get a quick overview of the output of the LAPACK 
> testing.
> This script is run after the TESTING are compiled and run, that is after 
> you type either make or make lapack_testing.
> The LAPACK testing should give depending on your machine between 200 and 
> 1000 numerical errors.
> To get a short summary of the TESTING, you can use.
> ./lapack_testing.py -s
> More detailed usage of lapack_testing.py below.
> 
> [literal]
> ./lapack_testing.py [-h|--help] [-d dir |--dir dir] [-s |--short] [-r 
> |--run] [-e |--error] [-p p |--prec p] [-t test |--test test] [-n | 
> --number]
>     - h is to print this message
>     - r is to use to run the LAPACK tests then analyse the output (.out 
> files). By default, the script will not run all the LAPACK tests
>     - d [dir] is to indicate where is the LAPACK testing directory (.out 
> files). By default, the script will use .
> LEVEL OF OUTPUT
>     - x is to print a detailed summary
>     - e is to print only the error summary
>     - s is to print a short summary
>     - n is to print the numbers of failing tests (turn on summary mode)
> SECLECTION OF TESTS:
>     - p [s/c/d/z/x] is to indicate the PRECISION to run:
>            s=single
>            d=double
>            sd=single/double
>            c=complex
>            z=double complex
>            cz=complex/double complex
>            x=all [DEFAULT]
>     - t [lin/eig/mixed/rfp/all] is to indicate which TEST FAMILY to run:
>            lin=Linear Equation
>            eig=Eigen Problems
>            mixed=mixed-precision
>            rfp=rfp format
>            all=all tests [DEFAULT]
> EXAMPLES:
>     ./lapack_testing.py -n
>            Will return the numbers of failed tests by analyzing the LAPACK 
> output
>     ./lapack_testing.py -n -r -p s
>            Will return the numbers of failed tests in REAL precision by 
> running the LAPACK Tests then analyzing the output
>     ./lapack_testing.py -n -p s -t eig 
>            Will return the numbers of failed tests in REAL precision by 
> analyzing only the LAPACK output of EIGEN testings
> Written by Julie Langou (June 2011) 
> 
> 6.4. Doxygen generated documentation
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> 
> .lapackers
> 
>     Julie Langou (University of Tennessee, USA)
> 
> .contribution
> 
> The LAPACK comment layout has been updated to be compatible with the 
> Doxygen format.
> This allows a nicer html and manpages documentation.
> For html documentation, see http://www.netlib.org/lapack/explore-html/
> Also the individual routine download is now included inside Doxygen.
> The html documentation of each LAPACK routine contains a link to Download 
> LAPACK routine + dependencies [TGZ] [ZIP] [TXT]
> 
> 7. Bug Fix
> ----------
> 
> == [green]*bug0084* :: Bug Ming Gu
> * *CORRECTED*
> * bug reported by Mong Gu (UC Berkeley) through Jim Demmel
> * LAPACK driver routine infected: All SVD solvers when asked for singular 
> values only. (Pathological matrix only)
> * bug is present in LAPACK since 3.1.1 and in Matlab 2010a and 2010b
> * For very specific matrices (given by Ming Gu), the DQDS code converges 
> much slower than expected. The algorithm converges for these matrices un 
> many more iterations
> than the LAPACK maximum number of iterations allow.
> * bug fix: (1) increase the maximum number of iterations in xlasq2, (2) 
> fall back on Demmel/Kahan algorithm in case of nonconvergence of dqds.
> * corrected by Benjamin Lipschitz on Mon Oct 3rd 2011
> 
> == [green]*bug0083* :: Bug in the testing of CHERFSX function
> * *CORRECTED*
> * bug reported by omitrofa (intel team) on the LAPACK forum
> * see link:https://icl.cs.utk.edu/lapack-forum/viewtopic.php?t=2644[forum 
> topic 2644]
> * Test of CHERFSX function doesn't work because there is an error in 
> ../TESTING/LIN/cerrhex.f ( lines 284-330 ). This function uses the diagonal 
> pivoting factorization of a Hermitian indefinite matrix instead of the 
> diagonal pivoting factorization of a Hermitian indefinite packed matrix. 
> When CHERFSX is included in conditional "IF( LSAMEN( 2, C2, 'HE' ) ) THEN " 
> the test works.
> * corrected by Julie on Mon Aug 15th 2011
> 
> == [green]*bug0082* :: [CZ]laed2 Memory problems when COMPZ='I' in DSTEDC 
> * *CORRECTED*
> * bug reported by Edward Smyth
> * Added IF statement to prevent potential out of range errors with "K+1" 
> indices.
> * corrected by Rodney on Wed Jun 21th 2011
> 
> == [green]*bug0081* :: [SD]laed2 WORKSPACE size is wrong when NCVT = NRU = 
> NCC = 0 
> * *CORRECTED*
> * bug reported by Hatem
> * Update Workspace requirement to 4*N because the routine call [SD]LASCL 
> that requires a 4*N workspace.
> * corrected by Julie on Thu May 19th 2011
> 
> == [green]*bug0080* :: [CZ]BDSQR WORKSPACE size is wrong when NCVT = NRU = 
> NCC = 0 
> * *CORRECTED*
> * bug reported by Hatem
> * Update Workspace requirement to 4*N because the routine call [SD]LASCL 
> that requires a 4*N workspace.
> * corrected by Julie on Thu May 19th 2011
> 
> == [green]*bug0079* :: ILA[SDCZ]LR out of bound access in the DO LOOP for 
> array A.
> * *CORRECTED*
> * bug detected by Julie
> * Change DO LOOP condition and add a IF inside the DO LOOP
> * corrected by Rodney on Tue April 21th 2011
> 
> == [green]*bug0029* :: problem of WORKSPACE query with DSTEDC
> * *CORRECTED*
> * reported by Tiago Requeijo on Mon Mar 30 2009
> * see 
> link:http://icl.cs.utk.edu/lapack-forum/viewtopic.php?f=2&t=1389[forum 
> topic 1389]
> * see also email from Edward Smyth on June 21st 2011
> * corrected by Rodney
>