[Wien] lapw0 coredump on Opensuse Tumbleweed (Kernel 6.13.8-1-default (64-bit)) with Intel OneApi ifx / icx Compiler

Tue Apr 8 10:06:19 CEST 2025

Hello Gavin and Gerhard,

thanks for the reply. The core dump is due to a segmentation fault. I 
also saw the Intel avx error in the core dump report and compiled lapw0 
without the -AVX2 flag. Tumbleweed has now a new major kernel version 
(6.14.0-1-default (64-bit)) and some other updates. I compiled all again 
(same flags as before) and now lapw0 is running without any problems.

Thanks again for your help!

Best regards,

Michael

Am 07.04.2025 um 14:24 schrieb Gavin Abo:
>
> Michael,
>
> In your stack trace [1], it seems like lapw0 could be calling a AVX 
> routine:
>
> Stack trace of thread 
> 397829:                                                    #0 
> 0x000000000103f03a __intel_avx_rep_memset (/usr/local/WIEN2k/lapw0 + 
> 0xc3f03a)
>
> If you haven't already, you might try to compile without the -xAVX2 to 
> see if it changes or not the error.
>
> Of note, I don't see Opensuse in the list of Intel tested operating 
> systems for ifx 2025.1 [2], which might cause it to be more 
> susceptible to bugs with the Intel compiler, but SuSE LINUX Enterprise 
> Server 15 in the list may be the closet to it.
>
> [1] 
> https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg23657.html
> [2] 
> https://www.intel.com/content/www/us/en/developer/articles/system-requirements/oneapi-fortran-compiler-system-requirements.html
>
> Kind Regards,
>
> Gavin
> WIEN2k user
>
> On 4/7/2025 1:02 AM, Fecher, Gerhard wrote:
>> Hallo Michael,
>> Meanwhile I installed also OneAPI 2025.1 (with  Leap, kernelversion 6.4, not Tumbleweed)
>> Wien2k and all my other programs run without any problems, so far. There is no difference to 2025.0 together with a 5.14. kernel.
>> I tested Wien2k with fftw3 from mkl (needs to change the effected makefiles by hand) and also with a self compiled version.
>>
>> My OPTIONS are
>> -free -w -O3 -axCORE-AVX512 -fp-model=precise -DINTEL_VML -traceback -assume buffered_io -I$(MKLROOT)/include
>> or with fftw from mkl
>> -free -w -O3 -axCORE-AVX512 -fp-model=precise -DINTEL_VML -traceback -assume buffered_io -I$(MKLROOT)/include  -I$(MKLROOT)/include/fftw
>> the test was on a machine with two AMD EPYCs
>>
>> What was triggering the core dump in your case? was it a segmentation fault or something else ?
>>
>> I guess there is some incompatibility between differen libraries rather a problem with the kernel
>> did you check that there is no mixture of versions.
>>
>> Some years ago we decided not to use rolling update versions anymore
>>
>> Ciao
>> Gerhard
>>
>> DEEP THOUGHT in D. Adams; Hitchhikers Guide to the Galaxy:
>> "I think the problem, to be quite honest with you,
>> is that you have never actually known what the question is."
>>
>> ====================================
>> Dr. Gerhard H. Fecher
>> Institut of Physics
>> Johannes Gutenberg - University
>> 55099 Mainz
>> ________________________________________
>> Von: Wien [wien-bounces at zeus.theochem.tuwien.ac.at] im Auftrag von Michael Fechtelkord via Wien [wien at zeus.theochem.tuwien.ac.at]
>> Gesendet: Samstag, 29. März 2025 17:42
>> An: A Mailing list for WIEN2k users
>> Cc: Michael Fechtelkord
>> Betreff: Re: [Wien] lapw0 coredump on Opensuse Tumbleweed (Kernel 6.13.8-1-default (64-bit)) with Intel OneApi ifx / icx Compiler
>>
>> Hello Gerhard,
>>
>> - Compiler switches were -O3 -xAVX2 -FR -mp1 -w -prec_div -pc80 -pad -ip
>> -DINTEL_VML -traceback -assume buffered_io -I$(MKLROOT)/include
>> - I used also O2 instead of O3 - makes no difference (crashes also)
>>
>> - single processing instead of parallel led also to a core dump
>>
>> - I compiled also with ifort and icx .. that worked fine ..no crash
>>
>> It seems that the combination of the current ifx compiler together with
>> the current Tumbleweed Kernel is responsible for the crashes.
>>
>> LEAP 16 will switch a container-based ALP architecture. It is not sure
>> if LEAP 16 will be available in fall 2025 and if the migration from
>> 15.6. to 16 (upgrade) can be handled properly. I just wanted to check
>> for an alternative.
>>
>>
>> Best regards,
>>
>> Michael
>>
>>
>>
>> Am 29.03.2025 um 16:28 schrieb Fecher, Gerhard:
>>> Not enaugh information.
>>> What Compilerswitches ? Did you try the most optimisation right from beginning or did you start with „smooth“ switches?
>>> Does the serial version work ?
>>> What do you expect to go wrong with the next Leap version ?
>>>
>>>
>>>
>>>
>>> Ciao
>>> Gerhard
>>>
>>> DEEP THOUGHT in D. Adams; Hitchhikers Guide to the Galaxy:
>>> "I think the problem, to be quite honest with you,
>>> is that you have never actually known what the question is."
>>>
>>> ====================================
>>> Dr. Gerhard H. Fecher
>>> Institut of Physics
>>> Johannes Gutenberg - University
>>> 55099 Mainz
>>> ________________________________________
>>> Von: Wien [wien-bounces at zeus.theochem.tuwien.ac.at] im Auftrag von Michael Fechtelkord via Wien [wien at zeus.theochem.tuwien.ac.at]
>>> Gesendet: Samstag, 29. März 2025 12:01
>>> An: A Mailing list for WIEN2k users
>>> Cc: Michael Fechtelkord
>>> Betreff: [Wien] lapw0 coredump on Opensuse Tumbleweed (Kernel 6.13.8-1-default (64-bit)) with Intel OneApi ifx / icx Compiler
>>>
>>> Hello all,
>>>
>>>
>>> due to the upcoming change of opensuse leap 15.6 to 16.0, I am trying if
>>> opensuse Tumbleweed  (Kernel 6.13.8-1-default (64-bit)) works better for
>>> me than the new Leap 16 architecture. Compilation of elpa 2025, fftw
>>> 3.3.10, libxc 7.0.0 and mpich 4.3.0 and WIEN2k works all fine under
>>> oneapi ifx / icx 2025.1.0.
>>>
>>> However lapw0 crashes on all structures I tried to calculate. The
>>> systemd-coredump shows the following messages:
>>>
>>> systemd-coredump[397837]: [🡕] Process 397828 (lapw0) of user 1000
>> ....
>>> Any ideas would be welcomed.
>>>
>>> Best regards,
>>>
>>> Michael
>>>
>>> --
>>> Dr. Michael Fechtelkord
>>>
>>> Institut für Geologie, Mineralogie und Geophysik
>>> Ruhr-Universität Bochum
>>> Universitätsstr. 150
>>> D-44780 Bochum
>>>
>>> Phone: +49 (234) 32-24380
>>> Fax:  +49 (234) 32-04380
>>> Email:Michael.Fechtelkord at ruhr-uni-bochum.de
>>> Web Page:https://www.ruhr-uni-bochum.de/kristallographie/kc/mitarbeiter/fechtelkord/
>>>
>>> --
>>> Dr. Michael Fechtelkord
>>>
>>> Institut für Geologie, Mineralogie und Geophysik
>>> Ruhr-Universität Bochum
>>> Universitätsstr. 150
>>> D-44780 Bochum
>>>
>>> Phone: +49 (234) 32-24380
>>> Fax:  +49 (234) 32-04380
>>> Email:Michael.Fechtelkord at ruhr-uni-bochum.de
>>> Web Page:https://www.ruhr-uni-bochum.de/kristallographie/kc/mitarbeiter/fechtelkord/
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> SEARCH the MAILING-LIST at:http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html

-- 
Dr. Michael Fechtelkord

Institut für Geologie, Mineralogie und Geophysik
Ruhr-Universität Bochum
Universitätsstr. 150
D-44780 Bochum

Phone: +49 (234) 32-24380
Fax:  +49 (234) 32-04380
Email:Michael.Fechtelkord at ruhr-uni-bochum.de
Web Page:https://www.ruhr-uni-bochum.de/kristallographie/kc/mitarbeiter/fechtelkord/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20250408/a1326ce2/attachment.htm>