[gmx-users] SSE4 → AVX2

Szilárd Páll pall.szilard at gmail.com
Wed Aug 27 14:04:45 CEST 2014


--
Szilárd


On Wed, Aug 27, 2014 at 12:32 PM, nicola staffolani
<n.staffolani at unitus.it> wrote:
> Thanx Szilárd!
>
> The concept of cross compilation is not perfectly 100% clear to me :s

Briefly: given machine A and B (with different architectures), compile
binary for B on machine A (e.g. compiling for cluster nodes).

Google knows even more.

> software: GROMACS 4.6.5
> hardware: Intel® Core™ i7-4790 CPU @ 3.60GHz × 8
> message: Compiled acceleration: SSE4.1 (Gromacs could use AVX_256 on this
> machine, which is better)
>
> Have I given you the info you had requested?

If you are compiling on that machine, I suspect you have an outdated
compiler. What compiler are you using?

>
>
>
>
> 2014-08-27 12:00 GMT+02:00 Szilárd Páll <pall.szilard at gmail.com>:
>
>> Normally the highest supported SIMD instructions set, which is in your
>> case AVX2, is detected during configuration phase. The fact that you
>> are getting that message means that:
>> - you are cross-compiling on an SSE4.1 node (e.g. cluster head node)
>> for AVX2 machines;
>> - your compiler does not support AVX/AVX2 and SSE4.1 was chosen as
>> fallback; cmake should have warned about this during configuration.
>>
>> What is your hardware/software setup? Using -DGMX_SIMD=AVX2 should
>> work in the former case.
>>
>> Cheers,
>>
>> --
>> Szilárd
>>
>>
>> On Wed, Aug 27, 2014 at 10:44 AM, nicola staffolani
>> <n.staffolani at unitus.it> wrote:
>> > Dear Gromacs users & developers,
>> >
>> > my computer tells me that, if I'd use AVX_256 instead of SSE4, then my
>> > simulations would be done much faster. Which changes in which files do I
>> > have to do in order to achieve this? I have serched the mailing list but
>> > couldn't find the explicit answer: I warn you that I am not so much of a
>> > geek, therefore suggesting me, for example, to change the path, without
>> > telling me which precise actions I have to carry on (for example, as
>> > already said, which lines I have to change/add in which file) in order to
>> > achieve this, will only generate the next email from my side and then,
>> > since by then you will have already started answering me, you will lose
>> > more time answering me back than if you would have precisely described in
>> > the first reply the order of precise actions needed to be done in order
>> to
>> > achieve the goal ;)
>> >
>> > Thank you for helping and have a nice day/night,
>> >
>> > Nicola
>> > --
>> > Nicola Staffolani PhD
>> > Biophysics & Nanoscience Centre CNISM <http://www.unitus.it/biophysics>
>> > Università della Tuscia
>> > Largo dell'Università s.n.c., I-01100 Viterbo
>> > email: n.staffolani at unitus.it
>> > tel.: +39 0761 35 70 27; fax: +39 0761 35 71 36
>> > --
>> > 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.
>>
>
>
>
> --
> Nicola Staffolani PhD
> Biophysics & Nanoscience Centre CNISM <http://www.unitus.it/biophysics>
> Università della Tuscia
> Largo dell'Università s.n.c., I-01100 Viterbo
> email: n.staffolani at unitus.it
> tel.: +39 0761 35 70 27; fax: +39 0761 35 71 36
> --
> 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.


More information about the gromacs.org_gmx-users mailing list