[gmx-developers] GROMACS uses cases with 50-2K atoms

[Szilárd Páll]

Hi Guido,

On Fri, Jun 11, 2021 at 3:16 PM Guido Giuntoli <guido.giuntoli at huawei.com>
wrote:

> Thanks Berk and Szilard,
>
>
>
> I am trying to achieve a small system that stills being representative on
> a single core. If I understood correctly from:
> https://mailman-1.sys.kth.se/pipermail/gromacs.org_gmx-developers/2021-February/011004.html
> good scaling is achieved typically with setups of 50-1000 atoms/core in
> order to fit the problem in the LLC.
>

*At least in LLC (on CPU-only), in some cases even second-to-last level
caches.

>
> I am wondering if the water boxes are still representative of this
> situation without including the bonded forces. Do they use the nbnxn_kernel
> and the pme computations in a similar way as a normal use case?
>

Yes. The only thing different compared to some of the more real-world
biomolecular use-cases is the lack of bonded interactions and to some
degree a more homogeneous workload distribution (so less tendency for
workload-density related imbalance). The latter is hard to "emulate" in
systems generated by stacking, but the former would be addressed by either
using stacked villin (or to a great extent by using simpler systems like
ethanol boxes I mentioned before).


>
>
> Should I use
> https://manual.gromacs.org/documentation/2018/onlinehelp/gmx-solvate.html
> for generating water boxes?
>

That or just grab a set from here:
https://ftp.gromacs.org/benchmarks/water_bare_hbonds.tar.gz

Cheers,
--
Szilárd


>
>
> Thanks a lot, Guido.
>
>
>
> *From:* gromacs.org_gmx-developers-bounces at maillist.sys.kth.se [mailto:
> gromacs.org_gmx-developers-bounces at maillist.sys.kth.se] *On Behalf Of *Szilárd
> Páll
> *Sent:* Friday, June 11, 2021 2:55 PM
> *To:* Discussion list for GROMACS development <gmx-developers at gromacs.org>
> *Subject:* Re: [gmx-developers] GROMACS uses cases with 50-2K atoms
>
>
>
> Hi,
>
>
>
> RNAse is 18k/24k (dodec/cubic setup), otherwise you can stack the 5k
> villin and get 10k, 20k,... systems (using "gmx genconf -nbox N M K").
>
>
>
> Additionally to water boxes, I do have some water + ethanol boxes too
> which do have bonded interactions, but that will only be useful if you
> really want the granularity of system sizes beyond what the villing
> stacking allows. I can share these if you are interested.
>
>
>
> Cheers,
>
> --
> Szilárd
>
>
>
>
>
> On Fri, Jun 11, 2021 at 10:54 AM Berk Hess <hess at kth.se> wrote:
>
> Hi,
>
> I don't think we have much else in that range.
>
> I would use water boxes of different size. They don't have bonded
> interactions, so you don't test everything, but it is anyhow unrealistic to
> use a single thread, as synchronization between threads and cache sharing
> have large effects on the performance of GROMACS.
>
> Cheers,
>
> Berk
>
> On 6/9/21 5:37 PM, Guido Giuntoli wrote:
>
> Hello,
>
>
>
> I would like to know if there are GROMACS use cases/inputs that simulate
> about 50 – 2K atoms.
>
>
>
> Is there any benchmark hard coded in GROMACS that allows parametrization
> and test different problem sizes?
>
>
>
> The nonbonded-benchmark helps but I need something that executes all the
> typical GROMACS algorithms (PME, NxN computations, etc.). I need such a
> small case for a single-core simulator where I don’t have any threading
> facility.
>
>
>
> The smallest case (ready to use) that I found up to now is called “villin”
> with 5K atoms.
>
>
>
> Thanks a lot, Guido.
>
>
>
> *Mobile: +49 15904451229*
>
> *Email:    guido.giuntoli at huawei.com <guido.giuntoli at huawei.com>*
>
>
>
> [image: HW_POS_RBG_Vertical-300ppi]
>
> HUAWEI TECHNOLOGIES Duesseldorf GmbH
> Hansaallee 205, 40549 Dusseldorf, Germany, *www.huawei.com*
> <http://www.huawei.com/>
> Registered Office: Düsseldorf, Register Court Düsseldorf, HRB 56063,
> Managing Director: Li Peng, Li Jian, Shi Yanli
>
> Sitz der Gesellschaft: Düsseldorf, Amtsgericht Düsseldorf, HRB 56063,
> Geschäftsführer: Li Peng, Li Jian, Shi Yanli
>
>
> *-----------------------------------------------------------------------------------------------*
>
> *This e-mail and its attachments contain confidential information from
> HUAWEI, which is intended only for the person or entity whose address is
> listed above. Any use of the information contained herein in any way
> (including, but not limited to, total or partial disclosure, reproduction,
> or dissemination) by persons other than the intended recipient(s) is
> prohibited. If you receive this e-mail in error, please notify the sender
> by phone or email immediately and delete it!*
>
>
>
>
>
>
>
> --
> Gromacs Developers mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-developers
> or send a mail to gmx-developers-request at gromacs.org.
>
> --
> Gromacs Developers mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-developers
> or send a mail to gmx-developers-request at gromacs.org.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-developers/attachments/20210614/16631ddd/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image002.png
Type: image/png
Size: 15784 bytes
Desc: not available
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-developers/attachments/20210614/16631ddd/attachment-0001.png>