ABINIT Discussion Forums

Hello Dear users
i use ABINIT for my structures.
my structure is almost big. and when I run the ABINIT for my program, for example number of atoms 20, ntypat 3, acell 34.072254 35.222133 20.043059.
i see this error at first.
emana : ERROR -
Test failed to allocate 4712.049 Mbytes
It is not worth to continue
Action : modify input variable to fit the available memory.
or increase limit on available memory.
How do i deactivate this error in my PC?
My operating system is:
linux ubuntu 12.04
ABINIT last version
PC: RAM 8GB, CPU 3.7 GHZ , HARD 500 GB, Core I8
If it is possible answer to my question

as alternative:

may suit

Howether, if getwfk -1 is used (it strongly speeds up the calculation), then maybe you should read in the internet about how to in increase the size (4Gb!) of folder in ubuntu .
But if the problem is that the system does not have enough RAM, the problem is more difficult - how did you compiled ABINIT? (maybe try this: viewtopic.php?f=2&t=1391 )

Best wishes,
Maxim

maxim wrote:as alternative:

Code: Select all

prtden 0 #to reduce output size
prtwf 0 #to reduce output size
prteig 0 #to reduce output size

may suit

Howether, if getwfk -1 (this greatly speeds up the calculation) is used, then maybe you should read in the internet how to in increase (4Gb!) the size of a folder in ubuntu.
Though if the problem is that the system does not have enough RAM, the problem is more difficult - how do you compiled (maybe try viewtopic.php?f=2&t=1391 )?

Best wishes,
Maxim

Hi prof. maxim
thanks
But it doesn't work!!

Maybe it means that your calculation is too difficult for your PC,
If you give me your input file, I can look at it and recommend something else.

Best wishes,
Maxim

Your unit cell is huge for a PC. Just reduce it ... or buy 8Gb more of memory for your PC if you still want to do the calculation with it.

If you cannot reduce the cell size, please consider using a cluster or applying for time in a supercomputing center.

I also have a question - how about using a nanohub? Are these opportunities for performing the calculation free of charge (16 cores AMD opteron)?

maxim wrote:I also have a question - how about using a nanohub? Are these opportunities for performing the calculation free of charge (16 cores AMD opteron)?

Hi Prof. Maxim
I didn't use nanohub.
I send my input file

ndtset 2
#spin related quantities (only second dataset)
spinat2
0.0 0.0 4.0
0.0 0.0 4.0
0.0 0.0 4.0
0.0 0.0 4.0
0.0 0.0 4.0
0.0 0.0 4.0
0.0 0.0 4.0
0.0 0.0 4.0
0.0 0.0 4.0
0.0 0.0 4.0
0.0 0.0 4.0
0.0 0.0 4.0
0.0 0.0 4.0
0.0 0.0 4.0
0.0 0.0 4.0
0.0 0.0 4.0
0.0 0.0 4.0
0.0 0.0 4.0
0.0 0.0 4.0
0.0 0.0 4.0
nsppol2 2
#generate the total density of states in both cases
prtdos 1
#Rather standard input for a bcc metal
iscf 5
prtden 1
tolvrs 1.0d-10
acell 34.072254 35.222133 20.043059
ecut 20
natom 20
nband 70
ngkpt 6 6 6
nshiftk 1

nstep 100
ntypat 3
occopt 3
rprim 0.5 -0.5 0
0.5 0.5 0
0 0 1
shiftk 0.5 0.5 0.5
tsmear 0.01
typat 1 1 1 1 2 2 2 2 3 3 3 3 3 3 3 3 3 3 3 3
xred
0.13600000 0.12600000 0.78200000
0.86400000 0.87400000 0.21800000
0.86400000 0.37400000 0.78200000
0.13600000 0.62600000 0.21800000
0.00000000 0.75000000 0.78900000
0.00000000 0.25000000 0.21100000
0.25000000 0.50000000 0.75000000
0.75000000 0.00000000 0.75000000
0.09900000 0.58100000 0.82770000
0.90100000 0.41900000 0.17230000
0.90100000 0.91900000 0.82770000
0.09900000 0.08100000 0.17230000
0.14500000 0.36700000 0.42860000
0.85500000 0.63300000 0.57140000
0.85500000 0.13300000 0.42860000
0.14500000 0.86700000 0.57140000
0.35400000 0.16400000 0.45200000
0.64600000 0.83600000 0.54800000
0.64600000 0.33600000 0.45200000
0.35400000 0.66400000 0.54800000
znucl 83 25 8
# add to conserve old < 6.7.2 behavior for calculating forces at each SCF step
optforces 1
If it is possible answer my question
Best Wishes

Hi jahangirii!

Your cell is big - try to upload your job at nanohub http://nanohub.org/ (though I do not completely shure whether it is free - as they say, it is free). Reduce your number of k-points (using prtkpt 1 and kptrlen). Maybe you can reduce number of bands. Maybe you should buy RAM as pouillon advised.

Hope this help
Maxim

maxim wrote:Hi jahangirii!

Your cell is big - try to upload your job at nanohub http://nanohub.org/ (though I do not completely shure whether it is free - as they say, it is free). Reduce your number of k-points (using prtkpt 1 and kptrlen). Maybe you can reduce number of bands. Maybe you should buy RAM as pouillon advised.

Hope this help
Maxim

Hi Prof. Maxim
I am extremely grateful for your response & i use your recommendation.
With Best