[Wien] Problem compiling Wien2k_07.3

[Peter Blaha]

Maybe the suggestions of the present siteconfig are not completely clear.
Is this more clear ?

....
Thus set NMATMAX to  sqrt(MEMORY/10)  (MEMORY in Bytes)

NMATMAX=10000 ==>   1GB (real) (==> cells with about 50-150 atoms/unitcell)
     ==> for lapw1c:    NMATMAX will be reduced internally to NMATMAX/sqrt2
     ==> for lapw1_mpi: NMATMAX will be increased internally to NMATMAX*sqrt(NP)

NUME determines the number of states to output. As a rule of thumb one can
estimate 100 basis functions per atom in the cell and 10 occupied states per
atom, so set    NUME=NMATMAX/10.

---------
Sorry, I did not mention, that there is only a param.inc (no param.inc_r and 
_c) file anymore, thus only ONE NMATMAX can be set (for the real case). For 
complex or mpi calculations NMATMAX will be scaled automatically.

NUME: The suggestion of NMATMAX/30 is a bit too restricted (it does NOT cause 
much memory anyway). I suggest NMATMAX/10 as default

David Holec schrieb:
> Dear Peter and all Wien users,
> 
> Sorry for the confusion: the problems were caused by unintentional mixing of
> different Wien2k versions: I simply extracted the new version into a
> directory with older version and this confused the compiler. All was set
> right once I had deleted the old version and proceed the installation from
> scratch. Thus, it was 'yet another user-induced error'...
> 
> However, I still needed to change the switch -lmkl_lapack64 to -lmkl_lapack
> to make it work.
> 
> On a different note: during the installation I need to set up NMATMAX and
> NUME parameters. I have 4GB of RAM in total, I'm running in k-parallel mode
> (two processors - a dual core machine) and the calculations run in complex
> mode. Thus I'm setting NMATMAX = SQRT (4 * 1024^3 / (2*20)) ~ 10000. Using
> the rule of thumb suggested in siteconfig script I then set NUME to
> 10000/30~345. However, what puzzles me is that the default values
> (NMATMAX=130000 & NUME = 1000) don't follow this rule of thumb. Why?
> 
> All the best,
> David
> 
> 
> __________________________________________________
> 
> David Holec               
> Department of Materials Science & Metallurgy
> University of Cambridge
> Pembroke Street
> CB2 3QZ, U.K.   
> 
> Tel: +44 (0) 1223 334404
> Email: dh331 at cam.ac.uk
> __________________________________________________
> 
> 
> 
> 
> 
> -----Original Message-----
> From: wien-bounces at zeus.theochem.tuwien.ac.at
> [mailto:wien-bounces at zeus.theochem.tuwien.ac.at] On Behalf Of Peter Blaha
> Sent: Friday, August 17, 2007 10:37 AM
> To: A Mailing list for WIEN2k users
> Subject: Re: [Wien] Problem compiling Wien2k_07.3
> 
> I don't have mkl9.1 nor ifort 10.0 yet.
> Library names must be adapted to the specific mkl version. That should be
> save
> 
> The error in setwar.F is unclear to me.
> 
> Please change into SRC_lapw1.
> ls -als setwar.*
> 
> do you see anything else than setwar.F and setwar.o ?? It should NOT!
> 
> Try   make     and   make complex
> 
> and check the output on the screen, in particular when setwar is compiled.
> 
>> I'm compiling the newest version of Wien2k. My system is Suse 10.1, 
>> ifor 10.0.025, mkl 9.1.023.
>>
>> First, I used the standard option in siteconfig (Linux (Intel ifort 9 
>> or 10 compiler + mkl 9.0)). I corrected the paths to my libraries (-L 
>> option in LDFLAGS and R_LIBS) and got some errors. Those were resolved 
>> by changing the option -lmkl_lapack64 (default) to -lmkl_lapack (as 
>> suggested by Gerhard Fecher in his notes) in R_LIBS options. Is this
> correct work-about?
>> Is it going to influence my calculations?
>>
>> After that, I'm getting "only" following errors in compilation of LAPW1:
>> Compile time errors (if any) were:
>>
>> SRC_lapw1/compile.msg:fortcom: Error: setwar.F, line 96: This label 
>> has not been SRC_lapw1/compile.msg:fortcom: Error: setwar.F, line 96: 
>> This label is undefined SRC_lapw1/compile.msg:make[1]: *** [setwar.o] 
>> Error 1
>> SRC_lapw1/compile.msg:make: *** [real] Error 2
>> SRC_lapw1/compile.msg:fortcom: Error: setwar.F, line 96: This label 
>> has not been SRC_lapw1/compile.msg:fortcom: Error: setwar.F, line 96: 
>> This label is undefined SRC_lapw1/compile.msg:make[1]: *** [setwar.o] 
>> Error 1
>> SRC_lapw1/compile.msg:make: *** [complex] Error 2
>>
>> My understanding is that the appropriate lines are not uncommented 
>> during compilations or something similar is going on. Any idea how to 
>> resolve this?
>>
>> Thanks in advance.
>> David Holec
>> _______________________________________________
>> Wien mailing list
>> Wien at zeus.theochem.tuwien.ac.at
>> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> 

-- 

                                       P.Blaha
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-15671             FAX: +43-1-58801-15698
Email: blaha at theochem.tuwien.ac.at    WWW: http://info.tuwien.ac.at/theochem/
--------------------------------------------------------------------------