[Wien] RAM use during band structure calculation

[pluto]

Dear All,

I am calculating 50x50 k-point mesh of a big slab. This takes several 
days on the i7 13700 desktop with 128 GB of RAM (would be faster on the 
cluster, of course).

During the lapw1 step the RAM use increases, as indicated in top or 
htop, from something like 4GB per process into something like 30GB per 
process (I run 4x k-paralell with additional OMP=2). Info from top is 
pasted below.

Is this normal? I would assume k-points are calculated one after the 
other, so the RAM use should not change.

Best,
Lukasz



Tasks: 524 total,   4 running, 520 sleeping,   0 stopped,   0 zombie
%Cpu(s): 14.0 us,  0.1 sy,  0.0 ni, 85.7 id,  0.1 wa,  0.0 hi,  0.0 si,  
0.0 st
MiB Mem : 128044.0 total,   1229.5 free, 106096.5 used,  22624.1 
buff/cache
MiB Swap:  32088.0 total,      0.1 free,  32087.9 used.  21947.6 avail 
Mem

     PID USER      PR  NI    VIRT    RES    SHR S  %CPU  %MEM     TIME+ 
COMMAND
1081393 lplucin   20   0   42.4g  31.9g   9852 R 100.0  25.5   4036:02 
lapw1
1081437 lplucin   20   0   42.3g  31.9g  10824 R  99.7  25.5   4044:16 
lapw1
1081439 lplucin   20   0   42.2g  30.4g  10076 R  99.7  24.3   4022:12 
lapw1