[CP2K-user] [CP2K:21394] Speed comparation: the new versions of cp2k are slower than the old

[li lincai]

Hi 
  Thanks for your message!
Both CP2K versions are installed by toolchain. OpenMPI 4.14 for CP2K23.1 
and OpenMPI 5.06 for CP2K25.1.
To manage multiple versions of CP2K, each version is separated in the 
.bashrc file to avoid messes. So, I can promise that the OpenMPI versions 
and CP2K versions are strictly corresponded. 

regards 
LCL
On Monday, April 14, 2025 at 7:53:16 PM UTC+8 Jürg Hutter wrote:

> Hi
>
> thank you for the outputs.
> It looks to me that the main part of the difference comes from MPI library 
> calls.
> Check the timings for "mp_alltoall_z22v".
> Can you check what MPI is used for the different versions of CP2K?
>
> regards
> JH
>
> ________________________________________
> From: cp... at googlegroups.com <cp... at googlegroups.com> on behalf of li 
> lincai <lilin... at gmail.com>
> Sent: Friday, April 11, 2025 8:27 AM
> To: cp2k
> Subject: Re: [CP2K:21361] Speed comparation: the new versions of cp2k are 
> slower than the old
>
> I have tested two systems; one has atoms, and the other has 744 atoms. The 
> two tasks ended normally, and their log information and input files are 
> attached.
> Thanks for cp2k developers' answers.
>
> On Monday, April 7, 2025 at 3:45:01 PM UTC+8 Jürg Hutter wrote:
> Hi
>
> Yes, it is obvious, but why? Can you run the same job with both versions
> such that the jobs finish correctly? Then there will be detailed timing
> outputs that will give the necessary information to see what is going on.
>
> regards
> JH
>
> ________________________________________
> From: cp... at googlegroups.com <cp... at googlegroups.com> on behalf of li 
> lincai <lilin... at gmail.com>
> Sent: Monday, April 7, 2025 9:31 AM
> To: cp2k
> Subject: Re: [CP2K:21333] Speed comparation: the new versions of cp2k are 
> slower than the old
>
> I have tested a Pt-IL (ionic liquid) system on my workstation, and the 
> input and log file are attached as below: Is is obvious that cp2k2025.1 is 
> slower than cp2k2023.1.
>
>
> On Wednesday, February 26, 2025 at 5:46:05 PM UTC+8 Jürg Hutter wrote:
> Hi
> Can you have a look at the detailed timings at the end of your test jobs?
> Do you see a specific routine causing the slow down or are all parts 
> running slower?
> If you post that information we can have a deeper look?
> regards
> JH
>
> ________________________________________
> From: cp... at googlegroups.com <cp... at googlegroups.com> on behalf of li 
> lincai <lilin... at gmail.com>
> Sent: Tuesday, February 25, 2025 2:54 AM
> To: cp2k
> Subject: [CP2K:21204] Speed comparation: the new versions of cp2k are 
> slower than the old
>
> Dear cp2k developers:
> I found that the new versions of CP2K are slower than the old ones. I 
> installed CP2K-2023.1, 2023.2, and 2025.1 on my computer. CP2K-2025.1 is 
> much slower than CP2K-2023.1 when running the same task.
> My computer's CPU is EPYC-7V12, and my system is Rocky linux 9.2. I tested 
> at least two systems; one is Pt-water system ( about 300 atoms), and 
> another is the ionic liquid system (about 500 atoms, including C H O F S 
> N), the results are same that CP2K-2025.1 is much slower than CP2K-2023.1. 
> For example, CP2K-2025.1 need 2.7 s while CP2K-2023.1. just need 1.9 s for 
> one SCF step.
> I compiled CP2K with GCC 11 and MKL math library.
> Do these default parameters has improved their accuracy, or I wrongly 
> compiled my CP2K?
>
> --
> 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<mailto:cp2k+uns... at googlegroups.com
> >.
> To view this discussion visit 
> https://groups.google.com/d/msgid/cp2k/f7011633-4f1d-4ea3-841e-9744f37b6d3dn%40googlegroups.com
> <
> https://groups.google.com/d/msgid/cp2k/f7011633-4f1d-4ea3-841e-9744f37b6d3dn%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+uns... at googlegroups.com<mailto:cp2k+uns... at googlegroups.com
> >.
> To view this discussion visit 
> https://groups.google.com/d/msgid/cp2k/71c11b2b-4455-40fc-8138-a659187a241bn%40googlegroups.com
> <
> https://groups.google.com/d/msgid/cp2k/71c11b2b-4455-40fc-8138-a659187a241bn%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+uns... at googlegroups.com<mailto:cp2k+uns... at googlegroups.com
> >.
> To view this discussion visit 
> https://groups.google.com/d/msgid/cp2k/63d07639-49cc-46d4-8cdf-8c75e4453fe2n%40googlegroups.com
> <
> https://groups.google.com/d/msgid/cp2k/63d07639-49cc-46d4-8cdf-8c75e4453fe2n%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 visit https://groups.google.com/d/msgid/cp2k/02b509bb-b07c-4be3-a528-727f3ad82da2n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20250419/f70ea29d/attachment-0001.htm>