[CP2K-user] [CP2K:20432] My ifort classic built of v2024.1 gives segfaults with TC-LRC + ADMM.

[Krack Matthias]

Hi Martin

The new Intel Fortran compiler ifx (instead of the “classic” ifort) is not yet working with CP2K (see this issue<https://github.com/cp2k/cp2k/pull/3479>, icx/ifort works but not icx/ifx). We hope that an upcoming oneAPI release (after 2024.2.0) will resolve the ICE.
I still suspect some kind of memory problem (“ulimit -s unlimited” is set?) in your case.

Best

Matthias

From: cp2k at googlegroups.com <cp2k at googlegroups.com> on behalf of Martin Konôpka <konopka2010 at gmail.com>
Date: Tuesday, 9 July 2024 at 17:14
To: cp2k <cp2k at googlegroups.com>
Subject: Re: [CP2K:20432] My ifort classic built of v2024.1 gives segfaults with TC-LRC + ADMM.
Hi Matthias,
Thanks. Unfortunately, the job crashes quite in the same way also after I have set the OMP_STACKSIZE variable as you suggested. I think I will try to use the non-classic intel compilers (although I have no idea about differences) and perhaps will omit some libraries from the built.

Best
Martin.

Dátum: utorok 9. júla 2024, čas: 16:41:40 UTC+2, odosielateľ: Krack Matthias
Hi Martin

It seems that you are using the default OMP_STACKSIZE. I suggest to increase that value to at least 16M or more, e.g. by using “export OMP_STACKSIZE=64M”.
CP2K v2024.1 uses much more OpenMP constructs than v9.1 which causes that need for a larger OMP_STACKSIZE.

HTH

Matthias

From: cp... at googlegroups.com <cp... at googlegroups.com> on behalf of Martin Konôpka <konop... at gmail.com>
Date: Tuesday, 9 July 2024 at 16:28
To: cp2k <cp... at googlegroups.com>
Subject: [CP2K:20430] My ifort classic built of v2024.1 gives segfaults with TC-LRC + ADMM.
Dear CP2K experts,

I compiled CP2K 2024.1 with Intel OneAPI tools, using the CP2K toolchain. (I acknowledge that the toolchain is now quite easily usable compared to those some years ago. I prefer to use the toolchain for it provides flexibility.)  Resulting executable cp2k.popt passed successfully a test with the PBE functional. However, it crashed upon a segmentation fault when I used PBE0 with the TC-LRC and ADMM techniques. The same input file (an oxygen atom) was processed cleanly on the same machine using CP2K 9.1 installed through the OS package manager . For each of the tests I cautiously set the environments so that v9.1 used the "factory" tools (installed with the ubuntu apt tool) and v2024.1 used the Intel tools from /opt/intel/oneapi.

I would be grateful for your advices on what should I do in a different way.
Some additional information:

OS: lubuntu 22.04, quite a standard installation
256 GB RAM in the machine

I attach:
* a file with my command to the toolchain
* the arch file local.psmp
* output from the command ldd /dalsie/cp2k-2024.1/exe/local/cp2k.popt
* cp2k.inp
* cp2k.out_v9.1
* cp2k.out_v2024.1
* stderr_v2024.1

If more info is needed from my side, just let me know. Thanks.

Best regards
Martin Konôpka.
--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns... at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/d04cd014-76bd-4a28-903a-6ea76b5ab085n%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/d04cd014-76bd-4a28-903a-6ea76b5ab085n%40googlegroups.com?utm_medium=email&utm_source=footer>.
--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com<mailto:cp2k+unsubscribe at googlegroups.com>.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/b87b83d9-857a-4425-8be1-1a2e79468a04n%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/b87b83d9-857a-4425-8be1-1a2e79468a04n%40googlegroups.com?utm_medium=email&utm_source=footer>.

-- 
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/ZRAP278MB0827FA85819D735594F20A3DF4DB2%40ZRAP278MB0827.CHEP278.PROD.OUTLOOK.COM.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20240709/40454dbf/attachment-0001.htm>