[QE-users] 答复: Huge RAM required for HSE calculation
Giuseppe Mattioli
giuseppe.mattioli at ism.cnr.it
Wed Dec 12 11:53:52 CET 2018
Dear LIANG Xiongyi
Indeed, your calculation DOES REQUIRE a huge amount of memory!!! You
may try to reduce it by:
1) Using norm conserving pseudopotential, so that you reduce ecutrho
and work with a faster and well-tested algorithm.
2) using ecutfock=1~2*ecutwfc (you can do it only with NC PPs, afaik).
This has a great impact on memory usage. You may test it with a
smaller system, but as a rule of thumb ecutfock=ecutwfc is a fast and
good choice for geometry optimizations, and ecutfock=1.5~2.0*ecutwfc
is a sensible choice for properties calculation. If you are desperate
and you want a more than decent EXX correction use ecutfock=ecutwfc
and calculate your band structure with the minimum requirement of
memory. If the calculation is unstable you can progressively raise
ecutfock.
3) it is not directly related to memory requirements, but
adaptive_thr=.true. makes the calulation of inner loops faster.
HTH
Giuseppe
Quoting LEUNG Clarence <liangxy123 at hotmail.com>:
> Dear federico,
>
> The error message is
>
> error: Exceeded job memory limit
> error: Step 5242.0 exceeded memory limit (130943908 > 128000000),
> being killed
>
> If reduce the number of Kpoint, the calculated Band structure will
> be not smooth enough.
>
> Thanks
>
> LIANG Xiongyi
> ________________________________
> 发件人: users <users-bounces at lists.quantum-espresso.org> 代表 IORI,
> Federico <federico.iori at airliquide.com>
> 发送时间: 2018年12月12日 16:44
> 收件人: users at lists.quantum-espresso.org
> 主题: Re: [QE-users] Huge RAM required for HSE calculation
>
> Hi.
>
> Which is the output error message you get?
>
>
> I am not too expert with HSE in QE, but I suppose the kpoints number
> is quite large.
> Maybe try to reduce the number of kpoints to see if u reproduce the
> REM error.
> Start with a small kpoint mesh and try to increase it gradually
> until reaching convergence, I would say.
>
> HTH
> With the best
> federico
>
>
> On Wed, Dec 12, 2018 at 9:35 AM LEUNG Clarence
> <liangxy123 at hotmail.com<mailto:liangxy123 at hotmail.com>> wrote:
> Dear QE users,
>
> Recently, I want to calculate the band structure by HSE method.
> However, the calculation will stop because huge RAM require. From
> the output file we know that:
>
> Estimated static dynamical RAM per process > 8.46 GB
>
> Estimated max dynamical RAM per process > 8.59 GB
>
> Estimated total dynamical RAM > 1099.85 GB
>
> I think 1099.85 GB RAM is so huge that it is not reasonable to afford.
>
> BTW, our server is 2 nodes, 128 threads， 256G RAM. The QE version is 6.3.
>
> And my input file is
>
>
> &CONTROL
> calculation='scf',
> etot_conv_thr = 3.5D-6 ,
> forc_conv_thr = 4.0D-4 ,
> wf_collect = .true. ,
> verbosity='high',
> /
> &SYSTEM
> ibrav=14,
> celldm(1)=36.1187133640d0, celldm(2)=0.6119121863d0,
> celldm(3)=1.3079965678d0,
> celldm(4)=0.0000000000d0, celldm(5)=0.0000000000d0,
> celldm(6)=0.0000000000d0,
> nat=63,
> ntyp=2,
> ecutwfc=50,
> ecutrho=400,
> input_dft='hse',
> occupations='smearing',
> smearing = 'gaussian' ,
> degauss = 0.002 ,
> vdw_corr = 'DFT-D3' ,
> nspin = 2 ,
> starting_magnetization(1) = 0.5 ,
> nqx1 = 1,
> nqx2 = 1 ,
> nqx3 = 1 ,
> /
> &ELECTRONS
> conv_thr=1d-06,
> mixing_beta=0.7d0,
> mixing_mode ='local-TF',
> /
> &IONS
> /
> ATOMIC_SPECIES
> C 12.010700d0 C_pbe_v1.2.uspp.F.UPF
> P 30.973800d0 P.pbe-n-rrkjus_psl.1.0.0.UPF
> ATOMIC_POSITIONS {crystal}
> P 0.207062953 0.994563394 0.532135751
> P 0.042823633 0.994796583 0.465770116
> C 0.157559602 0.119473623 0.509211511
> C 0.092142096 0.120398444 0.486938482
> P 0.453113260 0.016237452 0.533698293
> P 0.289980002 0.986689982 0.464529400
> C 0.399646081 0.133053286 0.505994952
> C 0.335858066 0.117232823 0.482213070
> P 0.704320783 0.007706925 0.532982064
> P 0.538824580 0.008408401 0.468045855
> C 0.654037234 0.132822314 0.511474768
> C 0.587848371 0.135094109 0.489114957
> P 0.956731588 -0.000006745 0.532145072
> P 0.791554157 0.002228032 0.466394516
> C 0.907495292 0.125510515 0.509696521
> C 0.841931621 0.126642797 0.487430517
> P 0.208266817 0.243782922 0.529541853
> P 0.043965854 0.246589308 0.464392818
> C 0.158728245 0.369838186 0.509463992
> C 0.093217169 0.370873210 0.486709312
> P 0.424177849 0.286343852 0.516852430
> P 0.290089532 0.244400471 0.459582516
> C 0.349540011 0.348792825 0.487513040
> P 0.705574516 0.255089554 0.532107492
> P 0.542012221 0.260506971 0.464249905
> C 0.656129286 0.380833823 0.512136142
> C 0.590214180 0.381152001 0.490082742
> P 0.957108151 0.250073183 0.531244245
> P 0.792998637 0.251922321 0.464998669
> C 0.907321789 0.375462368 0.510050868
> C 0.841746979 0.376334296 0.487596817
> P 0.208402877 0.492352243 0.532960869
> P 0.043963546 0.496372026 0.465432581
> C 0.160144417 0.618157610 0.509696507
> C 0.094342113 0.619600026 0.487659242
> P 0.471073678 0.513328655 0.548894023
> P 0.309129724 0.486227608 0.474070256
> C 0.414925399 0.629235003 0.523346545
> C 0.350961159 0.616373438 0.499640968
> P 0.705395802 0.505478539 0.534378721
> P 0.537192856 0.506995560 0.474222102
> C 0.656825419 0.631333460 0.511436510
> C 0.590845794 0.632466501 0.490389008
> P 0.956482750 0.499986748 0.532254831
> P 0.791270143 0.501641183 0.466922839
> C 0.906923580 0.625196225 0.510129362
> C 0.841214140 0.626345388 0.488309739
> P 0.207667541 0.743674910 0.533472131
> P 0.043998636 0.744658734 0.465730512
> C 0.158052412 0.868776604 0.510578825
> C 0.092761413 0.869723388 0.488195850
> P 0.461194613 0.761061943 0.539728957
> P 0.299530254 0.737206731 0.471960596
> C 0.407654987 0.881987587 0.515312903
> C 0.343486860 0.870097791 0.492007540
> P 0.706175574 0.757137135 0.532107627
> P 0.539960743 0.756974729 0.469099032
> C 0.655720285 0.882343139 0.510585276
> C 0.589586421 0.883272199 0.489712695
> P 0.956945276 0.749651294 0.531828394
> P 0.791991567 0.752132650 0.465926241
> C 0.907222485 0.875386768 0.509910655
> C 0.841486983 0.876756958 0.488249976
>
> K_POINTS crystal
> 54
> 0 0.5447405 0.0000000 0.2222222
> 0.3333333 0 0.0000000 0.2222222
> 0.3333333 0.5447405 0.0000000 0.2222222
> -0.333333333 0.5447405000 0.0000000000 0.2222222
> 0 0 0.0000000 0.1111111
> 0.0000000000 0.0000000000 0.0000000000 0.0
> 0.0555555556 0.0000000000 0.0000000000 0.0
> 0.1111111111 0.0000000000 0.0000000000 0.0
> 0.1666666667 0.0000000000 0.0000000000 0.0
> 0.2222222222 0.0000000000 0.0000000000 0.0
> 0.2777777778 0.0000000000 0.0000000000 0.0
> 0.3333333333 0.0000000000 0.0000000000 0.0
> 0.3888888889 0.0000000000 0.0000000000 0.0
> 0.4444444444 0.0000000000 0.0000000000 0.0
> 0.5000000000 0.0000000000 0.0000000000 0.0
> 0.5000000000 0.0333333333 0.0000000000 0.0
> 0.5000000000 0.0666666667 0.0000000000 0.0
> 0.5000000000 0.1000000000 0.0000000000 0.0
> 0.5000000000 0.1333333333 0.0000000000 0.0
> 0.5000000000 0.1666666667 0.0000000000 0.0
> 0.5000000000 0.2000000000 0.0000000000 0.0
> 0.5000000000 0.2333333333 0.0000000000 0.0
> 0.5000000000 0.2666666667 0.0000000000 0.0
> 0.5000000000 0.3000000000 0.0000000000 0.0
> 0.5000000000 0.3333333333 0.0000000000 0.0
> 0.5000000000 0.3666666667 0.0000000000 0.0
> 0.5000000000 0.4000000000 0.0000000000 0.0
> 0.5000000000 0.4333333333 0.0000000000 0.0
> 0.5000000000 0.4666666667 0.0000000000 0.0
> 0.5000000000 0.5000000000 0.0000000000 0.0
> 0.4444444444 0.5000000000 0.0000000000 0.0
> 0.3888888889 0.5000000000 0.0000000000 0.0
> 0.3333333333 0.5000000000 0.0000000000 0.0
> 0.2777777778 0.5000000000 0.0000000000 0.0
> 0.2222222222 0.5000000000 0.0000000000 0.0
> 0.1666666667 0.5000000000 0.0000000000 0.0
> 0.1111111111 0.5000000000 0.0000000000 0.0
> 0.0555555556 0.5000000000 0.0000000000 0.0
> 0.0000000000 0.5000000000 0.0000000000 0.0
> 0.0000000000 0.4666666667 0.0000000000 0.0
> 0.0000000000 0.4333333333 0.0000000000 0.0
> 0.0000000000 0.4000000000 0.0000000000 0.0
> 0.0000000000 0.3666666667 0.0000000000 0.0
> 0.0000000000 0.3333333333 0.0000000000 0.0
> 0.0000000000 0.3000000000 0.0000000000 0.0
> 0.0000000000 0.2666666667 0.0000000000 0.0
> 0.0000000000 0.2333333333 0.0000000000 0.0
> 0.0000000000 0.2000000000 0.0000000000 0.0
> 0.0000000000 0.1666666667 0.0000000000 0.0
> 0.0000000000 0.1333333333 0.0000000000 0.0
> 0.0000000000 0.1000000000 0.0000000000 0.0
> 0.0000000000 0.0666666667 0.0000000000 0.0
> 0.0000000000 0.0333333333 0.0000000000 0.0
> 0.0000000000 0.0000000000 0.0000000000 0.0
>
> Thanks in advance.
>
> LIANG Xiongyi
> City University of Hong Kong
>
> _______________________________________________
> users mailing list
> users at lists.quantum-espresso.org<mailto:users at lists.quantum-espresso.org>
> https://lists.quantum-espresso.org/mailman/listinfo/users
>
>
> --
> Federico IORI
>
> Computational material scientist
>
> Paris-Saclay Research Center
>
> 1 chemin de la Porte des
> Loges<https://www.google.com/maps/place/Air+Liquide/@48.8297381,2.2016685,11.75z/data=!4m5!3m4!1s0x47e67e61a4fbbdc7:0xca3bea9e80059880!8m2!3d48.7630415!4d2.1333045>
> Les Loges en Josas ? 78354 Jouy en Josas cedex
> Mail: federico.iori at airliquide.com<mailto:federico.iori at airliquide.com>
>
> Phone: +33 7 621 605 15
>
> [https://lh5.googleusercontent.com/jWge08q7BtEKOlxuBF_cj804ywzhJmkjw9rEb5dBZLKyUL_HQkVwhcwNi9vtzVTj-EiY-wQhE18VPSACNFaAFCAWvxp6bEwkTDR6-IjGBqp-Y3Afzx-OxN8OHxXYPLS0Dz9tP8fr]
GIUSEPPE MATTIOLI
CNR - ISTITUTO DI STRUTTURA DELLA MATERIA
Via Salaria Km 29,300 - C.P. 10
I-00015 - Monterotondo Scalo (RM)
Mob (*preferred*) +39 373 7305625
Tel + 39 06 90672342 - Fax +39 06 90672316
E-mail: <giuseppe.mattioli at ism.cnr.it>
More information about the users
mailing list