[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: compiling development sources
From: |
Jaroslav Hajek |
Subject: |
Re: compiling development sources |
Date: |
Thu, 11 Feb 2010 10:21:26 +0100 |
On Thu, Feb 11, 2010 at 10:07 AM, Carlo de Falco
<address@hidden> wrote:
>
> On 11 Feb 2010, at 09:51, Jaroslav Hajek wrote:
>
>> Oh my. I should use g95 to verify the sources.
>> Please apply the following patch (only fixes test suite, so should be
>> irrelevant for Octave) and try again:
>
>
> I still get the following warning:
>
> ------------------------------------------------------------------------
> g95 -O3 -funroll-loops -c utils.f
> In file utils.f:58
>
> call drandg(2*m,n,x,2*ldx)
> 1
> In file utils.f:34
>
> subroutine drandg(m,n,x,ldx)
> 2
> Warning (155): Inconsistent types (COMPLEX(8)/REAL(8)) in actual argument
> lists at (1) and (2)
> In file utils.f:51
>
> call srandg(2*m,n,x,2*ldx)
> 1
> In file utils.f:21
>
> subroutine srandg(m,n,x,ldx)
> 2
> Warning (155): Inconsistent types (COMPLEX(4)/REAL(4)) in actual argument
> lists at (1) and (2)
> ------------------------------------------------------------------------
>
Yes, these two are intentional (Fortran guarantees this to work, and
it can't be so easily avoided because there is no CLARUV/ZLARUV in
LAPACK).
> But the tests now seem to pass (see below).
> Notice that I had to manually change permissions of the script
> test/report_results
> for 'make test' to work. Maybe you should change
>
> ./report_results $(OUTS)
>
> to
>
> $(SH) ./report_results $(OUTS)
>
> in test/Makefile ?
> c.
>
>
>
> testing QR rank-1 update routines.
> All residual errors are expected to be small.
>
> sqr1up test (full factorization):
> residual error = 0.122189521790E-05 PASS
> orth. residual error = 0.476837158203E-06 PASS
> dqr1up test (full factorization):
> residual error = 0.334454686168E-14 PASS
> orth. residual error = 0.999200722163E-15 PASS
> cqr1up test (full factorization):
> residual error = 0.133548405756E-05 PASS
> orth. residual error = 0.596046447754E-06 PASS
> zqr1up test (full factorization):
> residual error = 0.712296977776E-14 PASS
> orth. residual error = 0.177635683940E-14 PASS
> sqr1up test (economized factorization):
> residual error = 0.745058059692E-06 PASS
> orth. residual error = 0.536441802979E-06 PASS
> dqr1up test (economized factorization):
> residual error = 0.210942374679E-14 PASS
> orth. residual error = 0.888178419700E-15 PASS
> cqr1up test (economized factorization):
> residual error = 0.174107810835E-05 PASS
> orth. residual error = 0.596046447754E-06 PASS
> zqr1up test (economized factorization):
> residual error = 0.454852485457E-14 PASS
> orth. residual error = 0.144328993201E-14 PASS
> sqr1up test (rows < columns):
> residual error = 0.143446959555E-05 PASS
> orth. residual error = 0.476837158203E-06 PASS
> dqr1up test (rows < columns):
> residual error = 0.155431223448E-14 PASS
> orth. residual error = 0.111022302463E-14 PASS
> cqr1up test (rows < columns):
> residual error = 0.175077479980E-05 PASS
> orth. residual error = 0.476837158203E-06 PASS
> zqr1up test (rows < columns):
> residual error = 0.280501353073E-14 PASS
> orth. residual error = 0.133226762955E-14 PASS
> ----------------------------------------------------------------------
> total: PASSED 24 FAILED 0
>
>
> testing QR column insert routines.
> All residual errors are expected to be small.
>
> sqrinc test (full factorization):
> residual error = 0.834465026855E-06 PASS
> orth. residual error = 0.476837158203E-06 PASS
> dqrinc test (full factorization):
> residual error = 0.198452365652E-14 PASS
> orth. residual error = 0.888178419700E-15 PASS
> cqrinc test (full factorization):
> residual error = 0.954721485869E-06 PASS
> orth. residual error = 0.476837158203E-06 PASS
> zqrinc test (full factorization):
> residual error = 0.633102584584E-14 PASS
> orth. residual error = 0.133226762955E-14 PASS
> sqrinc test (economized factorization):
> residual error = 0.908970832825E-06 PASS
> orth. residual error = 0.417232513428E-06 PASS
> dqrinc test (economized factorization):
> residual error = 0.205391259556E-14 PASS
> orth. residual error = 0.999200722163E-15 PASS
> cqrinc test (economized factorization):
> residual error = 0.230040586757E-05 PASS
> orth. residual error = 0.683888231379E-06 PASS
> zqrinc test (economized factorization):
> residual error = 0.469955767903E-14 PASS
> orth. residual error = 0.111022302463E-14 PASS
> ----------------------------------------------------------------------
> total: PASSED 16 FAILED 0
>
>
> testing QR column delete routines.
> All residual errors are expected to be small.
>
> sqrdec test (full factorization):
> residual error = 0.834465026855E-06 PASS
> orth. residual error = 0.476837158203E-06 PASS
> dqrdec test (full factorization):
> residual error = 0.179023462721E-14 PASS
> orth. residual error = 0.888178419700E-15 PASS
> cqrdec test (full factorization):
> residual error = 0.213074417843E-05 PASS
> orth. residual error = 0.476837158203E-06 PASS
> zqrdec test (full factorization):
> residual error = 0.638465723566E-14 PASS
> orth. residual error = 0.111022302463E-14 PASS
> sqrdec test (economized factorization):
> residual error = 0.566244125366E-06 PASS
> orth. residual error = 0.357627868652E-06 PASS
> dqrdec test (economized factorization):
> residual error = 0.380251385934E-14 PASS
> orth. residual error = 0.888178419700E-15 PASS
> cqrdec test (economized factorization):
> residual error = 0.145934416196E-05 PASS
> orth. residual error = 0.596046447754E-06 PASS
> zqrdec test (economized factorization):
> residual error = 0.508800871843E-14 PASS
> orth. residual error = 0.133226762955E-14 PASS
> ----------------------------------------------------------------------
> total: PASSED 16 FAILED 0
>
>
> testing QR column shift routines.
> All residual errors are expected to be small.
>
> sqrshc test (left shift, full factorization):
> residual error = 0.834465026855E-06 PASS
> orth. residual error = 0.357627868652E-06 PASS
> dqrshc test (left shift, full factorization):
> residual error = 0.149880108324E-14 PASS
> orth. residual error = 0.122124532709E-14 PASS
> cqrshc test (left shift, full factorization):
> residual error = 0.149257743942E-05 PASS
> orth. residual error = 0.476837158203E-06 PASS
> zqrshc test (left shift, full factorization):
> residual error = 0.595205264017E-14 PASS
> orth. residual error = 0.133226762955E-14 PASS
> sqrshc test (right shift, economized factorization):
> residual error = 0.923871994019E-06 PASS
> orth. residual error = 0.476837158203E-06 PASS
> dqrshc test (right shift, economized factorization):
> residual error = 0.346944695195E-14 PASS
> orth. residual error = 0.777156117238E-15 PASS
> cqrshc test (right shift, economized factorization):
> residual error = 0.127628334212E-05 PASS
> orth. residual error = 0.476837158203E-06 PASS
> zqrshc test (right shift, economized factorization):
> residual error = 0.497870862620E-14 PASS
> orth. residual error = 0.888178419700E-15 PASS
> ----------------------------------------------------------------------
> total: PASSED 16 FAILED 0
>
>
> testing QR row insert routines.
> All residual errors are expected to be small.
>
> sqrinr test (full factorization):
> residual error = 0.894069671631E-06 PASS
> orth. residual error = 0.476837158203E-06 PASS
> dqrinr test (full factorization):
> residual error = 0.288657986403E-14 PASS
> orth. residual error = 0.111022302463E-14 PASS
> cqrinr test (full factorization):
> residual error = 0.147320042743E-05 PASS
> orth. residual error = 0.476837158203E-06 PASS
> zqrinr test (full factorization):
> residual error = 0.513478148889E-14 PASS
> orth. residual error = 0.111022302463E-14 PASS
> ----------------------------------------------------------------------
> total: PASSED 8 FAILED 0
>
>
> testing QR row delete routines.
> All residual errors are expected to be small.
>
> sqrder test (full factorization):
> residual error = 0.923871994019E-06 PASS
> orth. residual error = 0.476837158203E-06 PASS
> dqrder test (full factorization):
> residual error = 0.170696790036E-14 PASS
> orth. residual error = 0.888178419700E-15 PASS
> cqrder test (full factorization):
> residual error = 0.200354133995E-05 PASS
> orth. residual error = 0.596046447754E-06 PASS
> zqrder test (full factorization):
> residual error = 0.659374577221E-14 PASS
> orth. residual error = 0.999200722163E-15 PASS
> ----------------------------------------------------------------------
> total: PASSED 8 FAILED 0
>
>
> testing Cholesky rank-1 update routines.
> All residual errors are expected to be small.
>
> sch1up test:
> residual error = 0.762939453125E-05 PASS
> dch1up test:
> residual error = 0.106581410364E-13 PASS
> cch1up test:
> residual error = 0.171682331711E-04 PASS
> zch1up test:
> residual error = 0.213578748579E-13 PASS
> ----------------------------------------------------------------------
> total: PASSED 4 FAILED 0
>
>
> testing Cholesky rank-1 downdate routines.
> All residual errors are expected to be small.
>
> sch1dn test:
> residual error = 0.572204589844E-05 PASS
> dch1dn test:
> residual error = 0.106581410364E-13 PASS
> cch1dn test:
> residual error = 0.133514404297E-04 PASS
> zch1dn test:
> residual error = 0.213209075002E-13 PASS
> ----------------------------------------------------------------------
> total: PASSED 4 FAILED 0
>
>
> testing Cholesky symmetric insert routines.
> All residual errors are expected to be small.
>
> schinx test:
> residual error = 0.381469726562E-05 PASS
> dchinx test:
> residual error = 0.710542735760E-14 PASS
> cchinx test:
> residual error = 0.114664217108E-04 PASS
> zchinx test:
> residual error = 0.143832810024E-13 PASS
> ----------------------------------------------------------------------
> total: PASSED 4 FAILED 0
>
>
> testing Cholesky symmetric delete routines.
> All residual errors are expected to be small.
>
> schdex test:
> residual error = 0.381469726562E-05 PASS
> dchdex test:
> residual error = 0.710542735760E-14 PASS
> cchdex test:
> residual error = 0.114465747174E-04 PASS
> zchdex test:
> residual error = 0.177691186420E-13 PASS
> ----------------------------------------------------------------------
> total: PASSED 4 FAILED 0
>
>
> testing QR column shift routines.
> All residual errors are expected to be small.
>
> schshx test (left shift):
> residual error = 0.381469726562E-05 PASS
> dchshx test (left shift):
> residual error = 0.710542735760E-14 PASS
> cchshx test (left shift):
> residual error = 0.152587890625E-04 PASS
> zchshx test (left shift):
> residual error = 0.142177919159E-13 PASS
> schshx test (right shift):
> residual error = 0.381469726562E-05 PASS
> dchshx test (right shift):
> residual error = 0.106581410364E-13 PASS
> cchshx test (right shift):
> residual error = 0.152587890625E-04 PASS
> zchshx test (right shift):
> residual error = 0.143214466922E-13 PASS
> ----------------------------------------------------------------------
> total: PASSED 8 FAILED 0
>
>
> testing LU rank-1 update routines.
> All residual errors are expected to be small.
>
> slu1up test (rows > columns):
> residual error = 0.286102294922E-05 PASS
> dlu1up test (rows > columns):
> residual error = 0.177635683940E-14 PASS
> clu1up test (rows > columns):
> residual error = 0.139148448852E-05 PASS
> zlu1up test (rows > columns):
> residual error = 0.113438501032E-13 PASS
> slu1up test (rows < columns):
> residual error = 0.143051147461E-05 PASS
> dlu1up test (rows < columns):
> residual error = 0.111022302463E-14 PASS
> clu1up test (rows < columns):
> residual error = 0.337174788001E-05 PASS
> zlu1up test (rows < columns):
> residual error = 0.650807585519E-14 PASS
> ----------------------------------------------------------------------
> total: PASSED 8 FAILED 0
>
>
> testing pivoted LU rank-1 update routines.
> All residual errors are expected to be small.
>
> slup1up test (rows > columns):
> residual error = 0.631809234619E-05 PASS
> dlup1up test (rows > columns):
> residual error = 0.290739654574E-13 PASS
> clup1up test (rows > columns):
> residual error = 0.649834146316E-05 PASS
> zlup1up test (rows > columns):
> residual error = 0.848431244129E-14 PASS
> slup1up test (rows < columns):
> residual error = 0.594556331635E-05 PASS
> dlup1up test (rows < columns):
> residual error = 0.388578058619E-14 PASS
> clup1up test (rows < columns):
> residual error = 0.456277348349E-05 PASS
> zlup1up test (rows < columns):
> residual error = 0.145591686034E-13 PASS
> ----------------------------------------------------------------------
> total: PASSED 8 FAILED 0
>
> TOTAL: PASSED 128 FAILED 0
>
>
Great. Thanks for the impulse. You may try to vary the random seed at line 64
data seed /4*3/
-->
data seed /4*13/
the "4*" prefix is needed (it's not multiplication, just an ugly
Fortran feature), any integer will do as the second. Or you can supply
4 different integers separated by commas.
Prior to doing so, please apply this patch:
Index: test/Makefile
===================================================================
--- test/Makefile (revision 22)
+++ test/Makefile (revision 23)
@@ -31,6 +31,7 @@
./report_results $(OUTS)
$(OUTS): %.out: %
+ echo > $@
./$< | tee $@
$(PROGS): % : %.f utils.o ../libqrupdate.a
or delete the *.out files after each try, otherwise a segfault may
escape your attention.
If you can't produce a failure this way, the problem is probably not
in qrupdate.
The next step is to try to separate the Octave test that crashes.
--
RNDr. Jaroslav Hajek, PhD
computing expert & GNU Octave developer
Aeronautical Research and Test Institute (VZLU)
Prague, Czech Republic
url: www.highegg.matfyz.cz
- Re: compiling development sources, (continued)
Re: compiling development sources, Carlo de Falco, 2010/02/10
- Re: compiling development sources, Ben Abbott, 2010/02/10
- Re: compiling development sources, Jaroslav Hajek, 2010/02/11
- Re: compiling development sources, Carlo de Falco, 2010/02/11
- Re: compiling development sources, Jaroslav Hajek, 2010/02/11
- Re: compiling development sources, Carlo de Falco, 2010/02/11
- Re: compiling development sources, Jaroslav Hajek, 2010/02/11
- Re: compiling development sources, Carlo de Falco, 2010/02/11
- Re: compiling development sources,
Jaroslav Hajek <=
- Re: compiling development sources, Carlo de Falco, 2010/02/11
- Re: compiling development sources, Jaroslav Hajek, 2010/02/11
- Re: compiling development sources, Carlo de Falco, 2010/02/11
- Re: compiling development sources, Jaroslav Hajek, 2010/02/11
- Re: compiling development sources, Carlo de Falco, 2010/02/11
- Re: compiling development sources, Jaroslav Hajek, 2010/02/11
- Re: compiling development sources, Jaroslav Hajek, 2010/02/11
- Re: compiling development sources, Carlo de Falco, 2010/02/11
- Re: compiling development sources, Carlo de Falco, 2010/02/16
- Re: compiling development sources, Carlo de Falco, 2010/02/17