LAPACK Archives

[Lapack] [lapack] 32-bit GCC errors

MARK,
I did run by hand on a similar machine with gfortran 4.1.2 and my output is the 
same than you.
I looked at the error closer and this INFO ERROR is returned as there is a 
numerical problem, so this can be considered also a numerical error.

Here is our nightly built system :http://my.cdash.org/index.php?project=LAPACK
I am going to add a 32 built on a 64 bits machine.

I will update installation.hints after the next release so that user can 
compare their output to the one expected.

Regards,
Julie

On Mar 31, 2011, at 5:36 PM, Mark Dixon wrote:

On Wed, 23 Mar 2011, julie langou wrote:

Dear Mark,
We are testing extensively the LAPACK library and we know more or less the 
behavior on most popular machines.
Usually the problem comes from the compilers flags or BLAS library. That's 
why I always advice to go with the Reference BLAS first.
We have some (tortured) tests that will never pass on some machines. I 
agree, we need to take a closer look on those.
We are working on changing those LAPACK testings as they are confusing to 
users.
Here would be my guidelines
1) Test completed successfully without seg fault
2) No illegal error
3) No info error
4) No severe error
5) Between 200-500 tests failing (usually around 280 is the minimum [this is 
exactly what you have])

and that's it.
If you have a serious problem, failing tests will go above 1000.
Hope it helps
Julie

Thanks Julie, your guidance is much appreciated.

I'm having a bit of trouble with 32-bit GNU GCC (64-bit is ok). Despite 
compiling with no optimisations (I explicitly set flags for gfortran to 
'-O0'), your test post-processing script is logging two "info error"s.

To recap, I'm:

a) Building reference Netlib BLAS
b) Check that BLAS passes its tests
c) Building reference Netlib LAPACK 3.3.0

I've tried each of the cases:

1) 32-bit GCC 4.1.2 (RHEL5 4.1.2-48) on a 32-bit platform
2) 32-bit GCC 4.1.2 (RHEL5 4.1.2-50) on a 64-bit platform
3) 32-bit GCC 4.2.3 on a 64-bit platform
4) 32-bit GCC 4.4.5 on a 64-bit platform

Using example (1), the full output of your script is appended to this mail. 
Looking a little closer:

ded.out:

N=    5, IWK= 2, seed= 100,2082,  33, 613, type 17, test( 7)= 0.450E+16
DES:    1 out of  3270 tests failed to pass the threshold
*** Error code from DGEES =    6

zgg.out:

Matrix order=    5, type=16, seed=3956,2454,2676, 301, result   1 is 4.504D+15
ZGG drivers:      1 out of   1268 tests failed to pass the threshold
*** Error code from ZDRVGG =    5


The ded.out error appears in the installation.hints page as being reported in 
1999. The ZDRVGG error doesn't.


Is this a problem?

Thanks,

Mark
-- 
-----------------------------------------------------------------
Mark Dixon                       Email    : m.c.dixon@Domain.Removed
HPC/Grid Systems Support         Tel (int): 35429
Information Systems Services     Tel (ext): +44(0)113 343 5429
University of Leeds, LS2 9JT, UK
-----------------------------------------------------------------

******************************************************
RESULTS
    - numerical failure:
CST drivers:      1 out of  11664 tests failed to pass the threshold
DXV drivers:    200 out of   5000 tests failed to pass the threshold
DST drivers:      1 out of  14256 tests failed to pass the threshold
SXV drivers:     37 out of   5000 tests failed to pass the threshold
ZXV drivers:     24 out of   5000 tests failed to pass the threshold
ZGG drivers:      1 out of   1268 tests failed to pass the threshold
ZST drivers:      1 out of  11664 tests failed to pass the threshold
CST:    1 out of  4662 tests failed to pass the threshold
CST:    1 out of  4662 tests failed to pass the threshold
DES:    1 out of  3270 tests failed to pass the threshold
DSX:    1 out of  3500 tests failed to pass the threshold
DST:    1 out of  4662 tests failed to pass the threshold
DST:    1 out of  4662 tests failed to pass the threshold
ZST:    1 out of  4662 tests failed to pass the threshold
ZST:    1 out of  4662 tests failed to pass the threshold
value of largest test error             =    0.189E+01
value of largest test error             =    0.160D+01
value of largest test error             =    0.197E+01
value of largest test error             =    0.189D+01
value of largest test error                  =   0.796D+04
    - illegal error:     0
    - info error:        2
    - severe error       0



<Prev in Thread] Current Thread [Next in Thread>


For additional information you may use the LAPACK/ScaLAPACK Forum.
Or one of the mailing lists, or