[gmx-users] Laptop features for MD simulations - recommendations

Wed Feb 3 22:25:23 CET 2016

Hi,

Well if you already have a laptop with a decent cpu you could consider an
external gpu. There are companies that sell that kind of hardware
solutions, but, they are expensive. On the other hand, if you have a DIYer
vein, you could build an eGPU solution. This would require a desktop gpu
and power supply, and the connector with the pci express interface. There
are some tutorials on the web.
Of course this would kill mobility, but the bang for the buck when compared
to a enthusiast level laptop would be higher. Moreover, this kind of system
is less prone to overheat, as the gpu's tend to run hot, and cramming
powerhouses in tiny cases doesn't help cooling at all.

Cheers

Davide Cruz
On 3 Feb 2016 8:25 p.m., "Szilárd Páll" <pall.szilard at gmail.com> wrote:

> On Wed, Feb 3, 2016 at 8:46 PM, Dries Van Rompaey
> <dries.vanrompaey at gmail.com> wrote:
> > Hi,
> >
> > Just to expand a bit on Szilárd’s excellent answer, you can find a list
> of benchmarks for mobile cards here:
> >
> http://www.notebookcheck.net/Mobile-Graphics-Cards-Benchmark-List.844.0.html
>
> Note: even though you see quite high on that list GTX 7xx series GPUs,
> I strongly suggest to go with 9xx instead. Given the perf/W increase
> Kepler to Maxwell (for measured values see our recent paper on GROMACS
> hardware benchmarking), putting everything but performance aside, with
> the amount of heat that can be extracted from the laptop enclosure
> you'll get a lot more performance with the latter.
> Same goes for AMD, the cards are decent with our OpenCL support, but
> can only about match the Kepler-gen cards.
>
> > Kind regards
> >
> > Dries
> >
> >
> >> On 03 Feb 2016, at 15:14, Szilárd Páll <pall.szilard at gmail.com> wrote:
> >>
> >> Hi,
> >>
> >> What exactly do you want from the laptop besides performance?
> >> Autonomy, weight, size restrictions?
> >>
> >> If you want something with a decent amount of autonomy, just get a
> >> high-end Skylake laptop and do your runs on your workstation or
> >> cluster. If you care about battery time, you may as well forget about
> >> a powerful system (even a standalone GPU may not be worth it unless
> >> you can get switchable graphics) - or at least plan to put effort into
> >> avoiding high CPU/GPU load when on battery.
> >>
> >> If you want max performance, get a beast that's often referred to as
> >> "portable workstation" with a high-performance CPU (not 15W part) and
> >> a GTX 970M or 980M. AFAIK NVIDIAs site has a section where they
> >> advertise gaming laptops with these high-end mobile chips.
> >>
> >> Unless you want to suffer with Windows, I'd also recommend looking
> >> around before buying and making sure that you don't end up with some
> >> niche hardware that does not run well under GNU/Linux. I can highly
> >> recommend System76, they have some monsters too - I have even seen
> >> recently a rave review about a portable workstation they offer with a
> >> desktop CPU.
> >>
> >> Cheers,
> >> --
> >> Szilárd
> >>
> >>
> >>
> >> On Wed, Feb 3, 2016 at 11:17 AM, Yasser Almeida Hernández
> >> <yasser.almeida.hernandez at chemie.uni-hamburg.de> wrote:
> >>> Dear Joao,
> >>>
> >>> I am aware that laptops are not the proper hardware to run MD
> simulations. I
> >>> do my runs in workstations designed for that purpose. The hardware I
> want to
> >>> buy is not just for simulations, but for general private/work purpose,
> but I
> >>> would like to have a device robust enough to run calculations. My
> question
> >>> was to get some thoughts about an ideal/optimal hardware.
> >>>
> >>> Best
> >>>
> >>> Yasser
> >>>
> >>> ------------------------
> >>> Message: 6
> >>> Date: Wed, 3 Feb 2016 10:53:31 +0100
> >>> From: Jo?o Henriques <joao.henriques.32353 at gmail.com>
> >>> To: Discussion list for GROMACS users <gmx-users at gromacs.org>
> >>> Subject: Re: [gmx-users] Laptop features for MD simulations -
> >>>        recommendations
> >>> Message-ID:
> >>>        <
> CALC+HGT7zSMBkwitU11dprvH4m_qYW2BTKgnsq2Tvb70O71kwQ at mail.gmail.com>
> >>> Content-Type: text/plain; charset=UTF-8
> >>>
> >>> Dear Yasser,
> >>>
> >>> I am by no means an expert on hardware, but I would strongly advise
> against
> >>> buying a laptop for the specific purpose of running (any and all) MD
> >>> simulations. A desktop is much better suited platform for such a task,
> as
> >>> you should be able to get a much bigger bang for your buck (this is the
> >>> main reason, but there are more cons).
> >>>
> >>> Notice that I am not saying that you cannot run MD simulations on a
> laptop,
> >>> I'm just advising against it. In fact, you could probably run MD
> >>> simulations on a smartphone if you wanted to, but it doesn't sound
> like a
> >>> very good investment of your time and money. In my *personal opinion*,
> the
> >>> same holds true for the laptop.
> >>>
> >>> Best regards,
> >>> /J
> >>>
> >>> On Wed, Feb 3, 2016 at 10:30 AM, Yasser Almeida Hern?ndez <
> >>> yasser.almeida.hernandez at chemie.uni-hamburg.de> wrote:
> >>>
> >>>> Hi all,
> >>>>
> >>>> I want to buy a laptop suitable and powerful enough for MD simulations
> >>>> with GROMACS (mostly membrane/protein systems). Could you recommend
> >>>> optimal
> >>>> features for this goal?
> >>>>
> >>>> Thanks
> >>>>
> >>>> Yasser
> >>>>
> >>>> --
> >>>> Yasser Almeida Hern?ndez
> >>>> PhD student
> >>>> Institute of Biochemistry and Molecular Biology
> >>>> Department of Chemistry
> >>>> University of Hamburg
> >>>> Martin-Luther-King-Platz 6
> >>>> 20146 Hamburg
> >>>> Germany
> >>>> +49 40 42838 2845
> >>>> yasser.almeida.hernandez at chemie.uni-hamburg.de
> >>>> office: Grindelallee 117, room 250c
> >>>>
> >>>> --
> >>>> Gromacs Users mailing list
> >>>>
> >>>> * Please search the archive at
> >>>> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_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-users or
> >>>> send a mail to gmx-users-request at gromacs.org.
> >>>
> >>>
> >>>
> >>> ------------------------------
> >>>
> >>> --
> >>> Gromacs Users mailing list
> >>>
> >>> * Please search the archive at
> >>> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_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-users or
> send a
> >>> mail to gmx-users-request at gromacs.org.
> >>>
> >>> End of gromacs.org_gmx-users Digest, Vol 142, Issue 14
> >>> ******************************************************
> >>>
> >>>
> >>>
> >>>
> >>> --
> >>> Gromacs Users mailing list
> >>>
> >>> * Please search the archive at
> >>> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_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-users or
> send a
> >>> mail to gmx-users-request at gromacs.org.
> >> --
> >> Gromacs Users mailing list
> >>
> >> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_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-users or
> send a mail to gmx-users-request at gromacs.org.
> >
> > --
> > Gromacs Users mailing list
> >
> > * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_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-users or
> send a mail to gmx-users-request at gromacs.org.
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_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-users or
> send a mail to gmx-users-request at gromacs.org.