[CP2K-user] [CP2K:20674] LiCl is running slower than KCl

[Hung Nguyen]

Dear Prof. Hutter,

Thank you so much for pointing it out! I have just figured out that in the 
HFX data, I showed in the prior message, I did use a different amount of 
memory. My sincere apologies for the confusion!

I have been running a few systems. Some of them have memory calculated on 
the fly, and some of them not. The following is the result I obtain for 
LiCl and KCl when I use a lot of memory to make sure that no ERI is 
calculated on the fly:

For LiCl
     1 OT CG       0.80E-01 1919.1     0.00842150     -4418.1754324748 
-4.42E+03
     2 OT LS       0.13E+00  447.5                    -4456.4755208811

For KCl
     1 OT CG       0.80E-01  687.7     0.00530909     -8394.4355734905 
-8.39E+03
     2 OT LS       0.32E+00  197.3                    -8440.0336970755


Also, the following is result when I used less memory in KCl and ended up 
having some ERI calculated on the fly:

     1 OT CG       0.80E-01  683.5     0.01650856     -6317.2855863300 
-6.32E+03
     2 OT LS       0.32E+00  375.8                    -6856.0434752195
     3 OT CG       0.32E+00  400.3     0.00885142     -7865.8954808822 
-1.55E+03

Best,

Hung
On Monday, September 9, 2024 at 11:04:08 AM UTC-5 Jürg Hutter wrote:

> Hi
>
> please quantify "much slower".
>
> Odd thing is that according to your output for LiCl you store all 
> integrals incore:
> LiCl
> HFX_MEM_INFO| Number of sph. ERI's calculated on the fly: 0
> KCl
> HFX_MEM_INFO| Number of sph. ERI's calculated on the fly: 178229986639
>
> However, the total amount of integrals is much larger for LiCl (Li has a 
> more difuse basis than K).
>
> regards
> JH
>
> ________________________________________
> From: cp... at googlegroups.com <cp... at googlegroups.com> on behalf of Hung 
> Nguyen <hungngu... at gmail.com>
> Sent: Monday, September 9, 2024 5:46 PM
> To: cp2k
> Subject: [CP2K:20669] LiCl is running slower than KCl
>
> Dear CP2K community,
>
> I am running PBE0-D3 (50% HFX exchange) on several molten salt systems and 
> I found that LiCl system is running much slower than KCl (I used identical 
> input files for them, except coordinate-related information).
> The following is HFX information the log file reports:
> For LiCl:
>
> HFX_INFO| Replica ID: 1
>
> HFX_INFO| FRACTION: 0.5000000000
>
> HFX_INFO| Interaction Potential: TRUNCATED
>
> HFX_INFO| Cutoff Radius [angstrom]: 6.0000000000
>
> HFX_INFO| EPS_SCHWARZ: 1.0E-06
>
> HFX_INFO| EPS_SCHWARZ_FORCES 1.0E-06
>
> HFX_INFO| EPS_STORAGE_SCALING: 1.0E+00
>
> HFX_INFO| NBINS: 64
>
> HFX_INFO| BLOCK_SIZE: 4
>
> HFX_INFO| NUMBER_OF_SHELLS: AUTO
>
> HFX_INFO| Number of periodic shells considered: -1
>
> HFX_INFO| Number of periodic cells considered: 27
>
> HFX_MEM_INFO| Est. max. program size before HFX [MiB]: 973
>
> HFX_MEM_INFO| Number of cart. primitive ERI's calculated: 1492589796730
>
> HFX_MEM_INFO| Number of sph. ERI's calculated: 1180851126208
>
> HFX_MEM_INFO| Number of sph. ERI's stored in-core: 1180851126208
>
> HFX_MEM_INFO| Number of sph. ERI's stored on disk: 0
>
> HFX_MEM_INFO| Number of sph. ERI's calculated on the fly: 0
>
> HFX_MEM_INFO| Total memory consumption ERI's RAM [MiB]: 891545
>
> HFX_MEM_INFO| Whereof max-vals [MiB]: 30785
>
> HFX_MEM_INFO| Total compression factor ERI's RAM: 10.11
>
> HFX_MEM_INFO| Total memory consumption ERI's disk [MiB]: 0
>
> HFX_MEM_INFO| Total compression factor ERI's disk: 0.00
>
> HFX_MEM_INFO| Size of density/Fock matrix [MiB]: 372
>
> HFX_MEM_INFO| Size of buffers [MiB]: 78
>
> HFX_MEM_INFO| Number of periodic image cells considered: 27
>
> HFX_MEM_INFO| Est. max. program size after HFX [MiB]: 12886
>
> For KCl:
>
> HFX_INFO| Replica ID: 1
>
> HFX_INFO| FRACTION: 0.5000000000
>
> HFX_INFO| Interaction Potential: TRUNCATED
>
> HFX_INFO| Cutoff Radius [angstrom]: 6.0000000000
>
> HFX_INFO| EPS_SCHWARZ: 1.0E-06
>
> HFX_INFO| EPS_SCHWARZ_FORCES 1.0E-06
>
> HFX_INFO| EPS_STORAGE_SCALING: 1.0E-01
>
> HFX_INFO| NBINS: 64
>
> HFX_INFO| BLOCK_SIZE: 4
>
> HFX_INFO| NUMBER_OF_SHELLS: AUTO
>
> HFX_INFO| Number of periodic shells considered: -1
>
> HFX_INFO| Number of periodic cells considered: 27
>
> HFX_MEM_INFO| Est. max. program size before HFX [MiB]: 1489
>
> HFX_MEM_INFO| Number of cart. primitive ERI's calculated: 986338569460
>
> HFX_MEM_INFO| Number of sph. ERI's calculated: 678423895453
>
> HFX_MEM_INFO| Number of sph. ERI's stored in-core: 500193908814
>
> HFX_MEM_INFO| Number of sph. ERI's stored on disk: 0
>
> HFX_MEM_INFO| Number of sph. ERI's calculated on the fly: 178229986639
>
> HFX_MEM_INFO| Total memory consumption ERI's RAM [MiB]: 608162
>
> HFX_MEM_INFO| Whereof max-vals [MiB]: 6047
>
> HFX_MEM_INFO| Total compression factor ERI's RAM: 6.27
>
> HFX_MEM_INFO| Total memory consumption ERI's disk [MiB]: 0
>
> HFX_MEM_INFO| Total compression factor ERI's disk: 0.00
>
> HFX_MEM_INFO| Size of density/Fock matrix [MiB]: 396
>
> HFX_MEM_INFO| Size of buffers [MiB]: 90
>
> HFX_MEM_INFO| Number of periodic image cells considered: 27
>
> HFX_MEM_INFO| Est. max. program size after HFX [MiB]: 9582
>
> I would appreciate it if someone could help me to point out why LiCl is 
> running much slower KCl.
>
> Best,
>
> Hung
>
> --
> 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 on the web visit 
> https://groups.google.com/d/msgid/cp2k/db784987-f0fe-4f12-88ed-13d1b0c4568dn%40googlegroups.com
> <
> https://groups.google.com/d/msgid/cp2k/db784987-f0fe-4f12-88ed-13d1b0c4568dn%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/21cceb29-0daf-429d-8aa7-5f6163c1b7bcn%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20240909/c5157536/attachment-0001.htm>