[Wien] WIEN2k and gfortran II

2016-12-12 Thread Martin Kroeker 
>Anyway, the openblas is really good, and if somebody would know how to 
>"vectorize" the cos, sin (exp) calls in gfortran this would be very 
>valuable.
Zhang Xianyi, the mastermind behind OpenBLAS :-) , is also working on an
"OpenVML" vector math library - but I must admit I did not get around
to trying it yet. See https://github.com/xianyi/OpenVML
-- 
Dr. Martin Kroekermar...@ruby.chemie.uni-freiburg.de
c/o Prof.Dr. Caroline Roehr
Institut fuer Anorganische und Analytische Chemie der Universitaet Freiburg

___
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:  
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html

Re: [Wien] WIEN2k and gfortran II

2016-12-12 Thread Laurence Marks 
That is quite an extraordinary time difference! The KKK/OpenBlas comparison
is very clear, but I wonder if it is just the vectorization routines for
the rest.

One observation about ifort; it is really a very good optimizer, albeit
sometimes too aggressive. There are quite a few loops in Wien2k which are
technically not the fastest, with indices the wrong way around or redundant
variables. At various times I have rewritten some of these, and seen almost
no timing change. I think this is because ifort is itself doing the
reorganization as well as others such as cache and memory optimizations.

I doubt that gfortran is as good an optimizing compiler as ifort.

---
Professor Laurence Marks
"Research is to see what everybody else has seen, and to think what nobody
else has thought", Albert Szent-Gyorgi
http://www.numis.northwestern.edu
Corrosion in 4D http://MURI4D.numis.northwestern.edu
Partner of the CFW 100% gender equity project, www.cfw.org/100-percent
Co-Editor, Acta Cryst A


On Dec 12, 2016 04:04, "Peter Blaha"  wrote:

Inspired by the recent posts about   gfortran and openblas, I made some
timing tests myself.

I was using the "Test-Case" (serial benchmark) from our website (a
complex case with NMAT=3481.

I tested it on an Intel I7-3939 (6 core) processor with either ifort+mkl
(2016.3.210)  or   gfortran+openblas.
I was using 1, 2, 4 or 6 cores (set via OMP_NUM_TRHEADS) of one PC:

1 core:
Intel TIME HAMILT (WALL) = 5.2, HNS = 4.2, DIAG =25.8
gfortran: TIME HAMILT (WALL) =36.3, HNS = 4.0, DIAG =25.0

2 cores:
Intel TIME HAMILT (WALL) = 5.3, HNS = 2.5, DIAG =14.4
gfortran: TIME HAMILT (WALL) =36.3, HNS = 2.4, DIAG =13.4

4 cores:
Intel TIME HAMILT (WALL) = 5.3, HNS = 1.7, DIAG = 7.7
gfortran: TIME HAMILT (WALL) =36.6, HNS = 1.7, DIAG = 7.9

6 cores:
Intel TIME HAMILT (WALL) = 5.3, HNS = 1.5, DIAG = 6.4
gfortran: TIME HAMILT (WALL) =36.4, HNS = 2.0, DIAG = 7.4

So obviously, the openblas is really VERY good and basically of the same
quality as the MKL (if not faster !!).

But: Setting up the eigenvalue problems (HAMILT) involves the
calculation of many cosines (exponentials) and we can use the
"vector-cosines" from the mkl. This makes ifort in this part 7 times
faster 
This can also be seen from the partial timing in case.output1 of the
hamilt-times, where phase and us are significantly faster:

ifort
Time for al,bl(hamilt, cpu/wall) :  0.3 0.3
Time for legendre (hamilt, cpu/wall) :  0.1 0.1
Time for phase(hamilt, cpu/wall) :  1.1 1.3
Time for us   (hamilt, cpu/wall) :  1.2 1.2
Time for overlaps (hamilt, cpu/wall) :  2.0 1.9
Time for distrib  (hamilt, cpu/wall) :  0.1 0.0
gfortran
Time for al,bl(hamilt, cpu/wall) :  0.2 0.3
Time for legendre (hamilt, cpu/wall) :  0.2 0.2
Time for phase(hamilt, cpu/wall) : 25.925.3
Time for us   (hamilt, cpu/wall) :  6.3 6.8
Time for overlaps (hamilt, cpu/wall) :  2.8 3.0
Time for distrib  (hamilt, cpu/wall) :  0.0 0.0

This limits gfortan significantly, making it in these tests a factor of
two (or, when using 4 cores a factor of 3) slower.

Anyway, the openblas is really good, and if somebody would know how to
"vectorize" the cos, sin (exp) calls in gfortran this would be very
valuable.

Peter Blaha

On 12/08/2016 01:51 PM, John Rundgren wrote:
> Dear Arthur,
>
> "Linker Flags" and "R_LIB" are found by consulting google on
> "xianyi-openblas user manual".
>
> The "include" flag is necessary, otherwise there is a conflict with
> /usr/link/ld.
>
> Xianyi recommends -lopenblas and adds -lpthread -lgfortran with
> motivations understood by wise Linuxers. They have not done any harm.
>
> Could you improve calculation time ...? In a previous wien-bounces you
> find a test where gfortran+openblas is fully competitive with intel+mkl.
> A try is worthwhile.
>
> Best regards / John
>
>
> John Rundgren
> Department of Theoretical Physics, KTH Royal Institute of Technology
>
>
> ___
> Wien mailing list
> Wien@zeus.theochem.tuwien.ac.at
> https://urldefense.proofpoint.com/v2/url?u=http-3A__zeus.
theochem.tuwien.ac.at_mailman_listinfo_wien&d=CwICAg&c=
yHlS04HhBraes5BQ9ueu5zKhE7rtNXt_d012z2PA6ws&r=U_
T4PL6jwANfAy4rnxTj8IUxm818jnvqKFdqWLwmqg0&m=LEs97us_bzM_-
ekk7jHI23IDX5u8HNB0WsYy5TlAROU&s=4ZF5koixKVtEe8uFZ3WxwMYft6mbBO
A_NeMfSH4_eyA&e=
> SEARCH the MAILING-LIST at:  https://urldefense.proofpoint.
com/v2/url?u=http-3A__www.mail-2Darchive.com_wien-
40zeus.theochem.tuwien.ac.at_index.html&d=CwICAg&c=
yHlS04HhBraes5BQ9ueu5zKhE7rtNXt_d012z2PA6ws&r=U_
T4PL6jwANfAy4rnxTj8IUxm818jnvqKFdqWLwmqg0&m=LEs97us_bzM_-
ekk7jHI23IDX5u8HNB0WsYy5TlAROU&s=PkWwLIVnknBQProsZ2RGefHAlq2CYo
v5A2ip53hpcqg&e=
>

--

Re: [Wien] WIEN2k and gfortran II

2016-12-12 Thread Peter Blaha 
Inspired by the recent posts about   gfortran and openblas, I made some 
timing tests myself.


I was using the "Test-Case" (serial benchmark) from our website (a 
complex case with NMAT=3481.


I tested it on an Intel I7-3939 (6 core) processor with either ifort+mkl 
(2016.3.210)  or   gfortran+openblas.

I was using 1, 2, 4 or 6 cores (set via OMP_NUM_TRHEADS) of one PC:

1 core:
Intel TIME HAMILT (WALL) = 5.2, HNS = 4.2, DIAG =25.8
gfortran: TIME HAMILT (WALL) =36.3, HNS = 4.0, DIAG =25.0

2 cores:
Intel TIME HAMILT (WALL) = 5.3, HNS = 2.5, DIAG =14.4
gfortran: TIME HAMILT (WALL) =36.3, HNS = 2.4, DIAG =13.4

4 cores:
Intel TIME HAMILT (WALL) = 5.3, HNS = 1.7, DIAG = 7.7
gfortran: TIME HAMILT (WALL) =36.6, HNS = 1.7, DIAG = 7.9

6 cores:
Intel TIME HAMILT (WALL) = 5.3, HNS = 1.5, DIAG = 6.4
gfortran: TIME HAMILT (WALL) =36.4, HNS = 2.0, DIAG = 7.4

So obviously, the openblas is really VERY good and basically of the same 
quality as the MKL (if not faster !!).


But: Setting up the eigenvalue problems (HAMILT) involves the 
calculation of many cosines (exponentials) and we can use the 
"vector-cosines" from the mkl. This makes ifort in this part 7 times 
faster 
This can also be seen from the partial timing in case.output1 of the 
hamilt-times, where phase and us are significantly faster:


ifort
Time for al,bl(hamilt, cpu/wall) :  0.3 0.3
Time for legendre (hamilt, cpu/wall) :  0.1 0.1
Time for phase(hamilt, cpu/wall) :  1.1 1.3
Time for us   (hamilt, cpu/wall) :  1.2 1.2
Time for overlaps (hamilt, cpu/wall) :  2.0 1.9
Time for distrib  (hamilt, cpu/wall) :  0.1 0.0
gfortran
Time for al,bl(hamilt, cpu/wall) :  0.2 0.3
Time for legendre (hamilt, cpu/wall) :  0.2 0.2
Time for phase(hamilt, cpu/wall) : 25.925.3
Time for us   (hamilt, cpu/wall) :  6.3 6.8
Time for overlaps (hamilt, cpu/wall) :  2.8 3.0
Time for distrib  (hamilt, cpu/wall) :  0.0 0.0

This limits gfortan significantly, making it in these tests a factor of 
two (or, when using 4 cores a factor of 3) slower.


Anyway, the openblas is really good, and if somebody would know how to 
"vectorize" the cos, sin (exp) calls in gfortran this would be very 
valuable.


Peter Blaha

On 12/08/2016 01:51 PM, John Rundgren wrote:

Dear Arthur,

"Linker Flags" and "R_LIB" are found by consulting google on
"xianyi-openblas user manual".

The "include" flag is necessary, otherwise there is a conflict with
/usr/link/ld.

Xianyi recommends -lopenblas and adds -lpthread -lgfortran with
motivations understood by wise Linuxers. They have not done any harm.

Could you improve calculation time ...? In a previous wien-bounces you
find a test where gfortran+openblas is fully competitive with intel+mkl.
A try is worthwhile.

Best regards / John


John Rundgren
Department of Theoretical Physics, KTH Royal Institute of Technology


___
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:  
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html



--

  P.Blaha
--
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-165300 FAX: +43-1-58801-165982
Email: bl...@theochem.tuwien.ac.atWIEN2k: http://www.wien2k.at
WWW:   http://www.imc.tuwien.ac.at/TC_Blaha
--
___
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:  
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html

Re: [Wien] WIEN2k and gfortran II

2016-12-08 Thread "Niedermayr, Arthur" 
Hello John,

thank you very much for the useful information!

Best regards
Arthur






 Original Message 
Subject: [Wien] WIEN2k and gfortran II
Local Time: 8. Dezember 2016 1:51 PM
UTC Time: 8. Dezember 2016 12:51
From: j...@kth.se
To: wien@zeus.theochem.tuwien.ac.at 




Dear Arthur,

"Linker Flags" and "R_LIB" are found by consulting google on "xianyi-openblas 
user manual".

The "include" flag is necessary, otherwise there is a conflict with 
/usr/link/ld.

Xianyi recommends -lopenblas and adds -lpthread -lgfortran with motivations 
understood by wise Linuxers. They have not done any harm.

Could you improve calculation time ...? In a previous wien-bounces you find a 
test where gfortran+openblas is fully competitive with intel+mkl. A try is 
worthwhile.

Best regards / John









John Rundgren
Department of Theoretical Physics, KTH Royal Institute of Technology___
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:  
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html

[Wien] WIEN2k and gfortran II

2016-12-08 Thread John Rundgren 
Dear Arthur,

"Linker Flags" and "R_LIB" are found by consulting google on "xianyi-openblas 
user manual".

The "include" flag is necessary, otherwise there is a conflict with 
/usr/link/ld.

Xianyi recommends -lopenblas and adds -lpthread -lgfortran with motivations 
understood by wise Linuxers. They have not done any harm.

Could you improve calculation time ...? In a previous wien-bounces you find a 
test where gfortran+openblas is fully competitive with intel+mkl. A try is 
worthwhile.

Best regards / John



John Rundgren
Department of Theoretical Physics, KTH Royal Institute of Technology
___
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:  
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html

Re: [Wien] WIEN2k and gfortran

2016-12-08 Thread John Rundgren 
Dear Peter,
about the speed of gfortran+OpenBLAS and ifort+mkl:

In my previous message I used the 8-thread processors i7 and xeon, 
respectively. When the same test is repeated on the same xeon machine, 
computing time is 132 min and 138 min, respectively.

An amazing and hopeful result for poor researchers.

Most of the computing time is consumed in lapw0, lapw1, and lapw2. The 
following table shows the storage occupied by the corresponding binary files,

  binary   gfortran+OpenBLAS   ifort+mkl
  lapw0568 kB  2.2 MB
  lapw1344 kB  1.9 MB
  lapw2490 kB  2.2 MB

The present test seems to indicate that the code refinement of ifort relative 
to gfortran is counterproductive for WIEN2k. But is the test result generally 
valid? More WIEN2k experience than mine is required for a response.

Best regards / John




From: Wien  on behalf of "Niedermayr, 
Arthur" 
Sent: Wednesday, December 7, 2016 2:34 PM
To: A Mailing list for WIEN2k users
Subject: Re: [Wien] WIEN2k and gfortran

Hello Mr. Rundgren,

I wanted to ask you, why do I need the link /opt/OpenBLAS/include in linker 
flags?
And what about -lgfortran, -lpthread and "-I" option? What are they doing?


I am also using OpenBLAS with following options (without compilation errors):
L Linker Flags: $(FOPT) -L../SRC_lib -L/opt/OpenBLAS/lib
R R_LIB (LAPACK+BLAS):  -llapack_lapw -lopenblas

Could I improve the calculation time of my program if I use your options?


Best regards
Arthur Niedermayr


Ps. I thought that the OpenBLAS library only replaces the BLAS library, but not 
the LAPACK library.




 Original Message --------
Subject: [Wien] WIEN2k and gfortran
Local Time: 5. Dezember 2016 11:54 AM
UTC Time: 5. Dezember 2016 10:54
From: j...@kth.se
To: wien@zeus.theochem.tuwien.ac.at 



Dear WIEN2k team,

siteconfig and User's Guide can be updated concerning gfortran and BLAS:

 L Linker Flags: $(FOPT) -I /opt/OpenBLAS/include -L/opt/OpenBLAS/lib
 R R_LIB (LAPACK+BLAS):  -lopenblas -lpthread -lgfortran

OpenBLAS being available from the sites

www.openblas.net and https://github.com/xianyi/OpenBLAS/wiki/Installation-Guide

In a test taking 2.5 h/iteration gfortran/ifort time was 1.3. A current test 
got interesting when one of my disks containing ifort crashed.

Regards,
John Rundgren




___
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:  
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html

[Wien] WIEN2k and gfortran

2016-12-07 Thread Martin Kroeker 
While it is true that MKL is in most but not all cases still a bit
faster than OpenBLAS, there are a few other points to consider:
- availability for platform where MKL is (at least to my knowledge) not 
available
- no license fee 
- open source, so anybody can check verify the code (and fix any bugs
  they find) rather than just hope for the best and follow Intel errata 
The drawback, as with many volunteer projects, is an unpredictable rate of 
progress
especially in areas that require specialist skills (such as writing optimal 
assembly code for new processors).
Incidentally OpenBLAS is the successor to the old GotoBLAS (whose author joined
the Intel MKL team after leaving academia as far as I know) which is still
mentioned in siteconfig and the Installation section of the usersguide.

-- 
Dr. Martin Kroekermar...@ruby.chemie.uni-freiburg.de
c/o Prof.Dr. Caroline Roehr
Institut fuer Anorganische und Analytische Chemie der Universitaet Freiburg

___
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:  
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html

Re: [Wien] WIEN2k and gfortran

2016-12-07 Thread "Niedermayr, Arthur" 
Hello Mr. Rundgren,

I wanted to ask you, why do I need the link /opt/OpenBLAS/include in linker 
flags?
And what about -lgfortran, -lpthread and "-I" option? What are they doing?


I am also using OpenBLAS with following options (without compilation errors):
L Linker Flags: $(FOPT) -L../SRC_lib -L/opt/OpenBLAS/lib
R R_LIB (LAPACK+BLAS): -llapack_lapw -lopenblas

Could I improve the calculation time of my program if I use your options?


Best regards
Arthur Niedermayr


Ps. I thought that the OpenBLAS library only replaces the BLAS library, but not 
the LAPACK library.






 Original Message 
Subject: [Wien] WIEN2k and gfortran
Local Time: 5. Dezember 2016 11:54 AM
UTC Time: 5. Dezember 2016 10:54
From: j...@kth.se
To: wien@zeus.theochem.tuwien.ac.at 




Dear WIEN2k team,

siteconfig and User's Guide can be updated concerning gfortran and BLAS:

L Linker Flags: $(FOPT) -I /opt/OpenBLAS/include -L/opt/OpenBLAS/lib
R R_LIB (LAPACK+BLAS): -lopenblas -lpthread -lgfortran

OpenBLAS being available from the sites

www.openblas.net and https://github.com/xianyi/OpenBLAS/wiki/Installation-Guide

In a test taking 2.5 h/iteration gfortran/ifort time was 1.3. A current test 
got interesting when one of my disks containing ifort crashed.

Regards,
John Rundgren___
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:  
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html

Re: [Wien] WIEN2k and gfortran

2016-12-06 Thread Pavel Ondračka 
Just out of curiosity... have you done also some tests with ifort/openblas 
or gfortran/mkl combination? I have done few tests about two years ago with 
gfortran, ifort, openblas, mkl and also different optimisation flags 
including some advanced ones such as link time optimizations and I have come
to the conclusion that the only thing that really matters are the 
libraries... eg. the only think you actually need to get the best 
performance is mkl and the compiler choice is irrelevant.

Best regards
Pavel

"To clarify: in my test gfortran ran 1.3 slower than ifort. / Regards 

John Rundgren 

 
From: Wien  on behalf of Peter 
Blaha  
Sent: Tuesday, December 6, 2016 12:20 PM 
To: A Mailing list for WIEN2k users 
Subject: Re: [Wien] WIEN2k and gfortran 

Thank you very much for the information. I have not used gfortran for 
long time. 

Just on question: I did not fully understand the last sentence: 

> In a test taking 2.5 h/iteration gfortran/ifort time was 1.3. A current 
> test got interesting when one of my disks containing ifort crashed. 

Does it say with gfortran/OpenBlas it takes 2.5h, with ifort/mkl 1.3h ? 

Or does it mean that the ratio of the two timings is 1.3, 
i,.e. gfortran runs 1.3 times slower than ifort ? 

Regards 

On 12/05/2016 11:54 AM, John Rundgren wrote: 
> Dear WIEN2k team, 
> 
> siteconfig and User's Guide can be updated concerning gfortran and BLAS: 
> 
> L Linker Flags: $(FOPT) -I /opt/OpenBLAS/include 
> -L/opt/OpenBLAS/lib 
> R R_LIB (LAPACK+BLAS): -lopenblas -lpthread -lgfortran 
> 
> OpenBLAS being available from the sites 
> 
> www.openblas.net and 
> https://github.com/xianyi/OpenBLAS/wiki/Installation-Guide 
> 
> In a test taking 2.5 h/iteration gfortran/ifort time was 1.3. A current 
> test got interesting when one of my disks containing ifort crashed. 
> 
> Regards, 
> John Rundgren 
> 
> 
> 
> 
> ___ 
> Wien mailing list 
> Wien@zeus.theochem.tuwien.ac.at 
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien 
> SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.
theochem.tuwien.ac.at/index.html 
> 

-- 

P.Blaha 
-- 
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna 
Phone: +43-1-58801-165300 FAX: +43-1-58801-165982 
Email: bl...@theochem.tuwien.ac.at WIEN2k: http://www.wien2k.at 
WWW: http://www.imc.tuwien.ac.at/TC_Blaha 
-- 
___ 
Wien mailing list 
Wien@zeus.theochem.tuwien.ac.at 
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien 
SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.
tuwien.ac.at/index.html 
___ 
Wien mailing list 
Wien@zeus.theochem.tuwien.ac.at 
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien 
SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.
tuwien.ac.at/index.html 
"___
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:  
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html

Re: [Wien] WIEN2k and gfortran

2016-12-06 Thread John Rundgren 
To clarify: in my test gfortran ran 1.3 slower than ifort. / Regards

John Rundgren


From: Wien  on behalf of Peter Blaha 

Sent: Tuesday, December 6, 2016 12:20 PM
To: A Mailing list for WIEN2k users
Subject: Re: [Wien] WIEN2k and gfortran

Thank you very much for the information. I have not used gfortran for
long time.

Just on question: I did not fully understand the last sentence:

 > In a test taking 2.5 h/iteration gfortran/ifort time was 1.3. A current
 > test got interesting when one of my disks containing ifort crashed.

Does it say with gfortran/OpenBlas it takes  2.5h, with ifort/mkl 1.3h ?

Or does it mean that the ratio of the two timings is  1.3,
i,.e. gfortran runs 1.3 times slower than ifort ?

Regards

On 12/05/2016 11:54 AM, John Rundgren wrote:
> Dear WIEN2k team,
>
> siteconfig and User's Guide can be updated concerning gfortran and BLAS:
>
>  L Linker Flags: $(FOPT) -I /opt/OpenBLAS/include
> -L/opt/OpenBLAS/lib
>  R R_LIB (LAPACK+BLAS):  -lopenblas -lpthread -lgfortran
>
> OpenBLAS being available from the sites
>
> www.openblas.net and
> https://github.com/xianyi/OpenBLAS/wiki/Installation-Guide
>
> In a test taking 2.5 h/iteration gfortran/ifort time was 1.3. A current
> test got interesting when one of my disks containing ifort crashed.
>
> Regards,
> John Rundgren
>
>
>
>
> ___
> Wien mailing list
> Wien@zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> SEARCH the MAILING-LIST at:  
> http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
>

--

   P.Blaha
--
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-165300 FAX: +43-1-58801-165982
Email: bl...@theochem.tuwien.ac.atWIEN2k: http://www.wien2k.at
WWW:   http://www.imc.tuwien.ac.at/TC_Blaha
--
___
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:  
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
___
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:  
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html

Re: [Wien] WIEN2k and gfortran

2016-12-06 Thread Peter Blaha 
Thank you very much for the information. I have not used gfortran for 
long time.


Just on question: I did not fully understand the last sentence:

> In a test taking 2.5 h/iteration gfortran/ifort time was 1.3. A current
> test got interesting when one of my disks containing ifort crashed.

Does it say with gfortran/OpenBlas it takes  2.5h, with ifort/mkl 1.3h ?

Or does it mean that the ratio of the two timings is  1.3,
i,.e. gfortran runs 1.3 times slower than ifort ?

Regards

On 12/05/2016 11:54 AM, John Rundgren wrote:

Dear WIEN2k team,

siteconfig and User's Guide can be updated concerning gfortran and BLAS:

 L Linker Flags: $(FOPT) -I /opt/OpenBLAS/include
-L/opt/OpenBLAS/lib
 R R_LIB (LAPACK+BLAS):  -lopenblas -lpthread -lgfortran

OpenBLAS being available from the sites

www.openblas.net and
https://github.com/xianyi/OpenBLAS/wiki/Installation-Guide

In a test taking 2.5 h/iteration gfortran/ifort time was 1.3. A current
test got interesting when one of my disks containing ifort crashed.

Regards,
John Rundgren




___
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:  
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html



--

  P.Blaha
--
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-165300 FAX: +43-1-58801-165982
Email: bl...@theochem.tuwien.ac.atWIEN2k: http://www.wien2k.at
WWW:   http://www.imc.tuwien.ac.at/TC_Blaha
--
___
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:  
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html

[Wien] WIEN2k and gfortran

2016-12-05 Thread John Rundgren 
Dear WIEN2k team,

siteconfig and User's Guide can be updated concerning gfortran and BLAS:

 L Linker Flags: $(FOPT) -I /opt/OpenBLAS/include -L/opt/OpenBLAS/lib
 R R_LIB (LAPACK+BLAS):  -lopenblas -lpthread -lgfortran

OpenBLAS being available from the sites

www.openblas.net and https://github.com/xianyi/OpenBLAS/wiki/Installation-Guide

In a test taking 2.5 h/iteration gfortran/ifort time was 1.3. A current test 
got interesting when one of my disks containing ifort crashed.

Regards,
John Rundgren

___
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:  
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html

Re: [Wien] Wien2k and gfortran.4.8--seeking for help

2014-12-09 Thread 李志 
Dear Peter,
It works perfectly. 
Thank you very much
All the best,
Zhi Li

> -原始邮件-
> 发件人: "Peter Blaha" 
> 发送时间: 2014年12月9日 星期二
> 收件人: "A Mailing list for WIEN2k users" 
> 抄送: 
> 主题: Re: [Wien] Wien2k and gfortran.4.8--seeking for help
> 
> Please add to line 30 an "ERR=999" statement:
> 
>  READ(nfile,err=999,end=999) 
> SS_dn(1),SS_dn(2),SS_dn(3),BNAME_dn,NV_dn,NE_dn,WEIGHT_dn
> 
> and let me know if this fixes the problem for gfortran.
> 
> Thanks
> 
> On 12/09/2014 10:18 AM, 李志 wrote:
> > Dear All,
> > Do you have any experience with Wien2k compiled by gfortran 4.8?
> > The compilation finished without error. However, when I do a test, the
> > calculation stops with prompt:
> >
> > "At line 30 of file kptin_nv.F
> > Fortran runtime error: Sequential READ or WRITE not allowed after EOF
> > marker, possibly use REWIND or BACKSPACE"
> >
> > It seems the gfortran 4.8 can't interpret the I/O command in kptin_nv.F
> > correctly. I also recompiled with -fno-whole-file option. Unfortunately,
> > the problem still appears.
> > I attach part of kptin_nv.F here. Anyone can help to resolve this
> > problem?
> >
> >1   SUBROUTINE
> > KPTIN_nv(NV_up,NV_dn,ende,BNAME_up,BNAME_dn,WEIGHT_up,WEIGHT_dn,SS_up,SS_dn,NE_up,NE_dn)
> >2
> >3   USE param
> >4   USE parallel
> >5
> >6   IMPLICIT REAL*8(A-H,O-Z)
> >7
> >8 #ifdef Parallel
> >9   include 'mpif.h'
> >   10 #endif
> >   11
> >   12   integer :: NV_up,NV_dn,ende(2)
> >   13   character*10 :: BNAME_up,BNAME_dn
> >   14   real*8 :: WEIGHT_UP,WEIGHT_dn
> >   15   real*8 :: SS_up(3),SS_dn(3)
> >   16   integer :: NE_up,NE_dn
> >   17   integer :: nfile
> >   18
> >   19
> >   20   nfile=8+1
> >   21  if (MYROWHS.eq.0.and.MYCOLHS.eq.0) then
> >   22 READ(nfile,end=998)
> > SS_up(1),SS_up(2),SS_up(3),BNAME_up,NV_up,NE_up,WEIGHT_up
> >   23 goto 111
> >   24 998 ende(1)=1
> >   25 111  continue
> >   26  endif
> >   27
> >   28   nfile=8+2
> >   29  if (MYROWHS.eq.0.and.MYCOLHS.eq.0) then
> >   30 READ(nfile,end=999)
> > SS_dn(1),SS_dn(2),SS_dn(3),BNAME_dn,NV_dn,NE_dn,WEIGHT_dn
> >   31 goto 112
> >   32 999 ende(2)=1
> >   33 112  continue
> >   34  endif
> >   35
> >   36 #ifdef Parallel
> >   37   call mpi_bcast(ende,2,MPI_INTEGER,0,MPI_COMM_WORLD,ierr)
> >   38   call mpi_barrier(MPI_COMM_WORLD,ierr)
> >   39
> >   40   if (ende(1).ne.1) then
> >   41  call
> > mpi_bcast(SS_up,3,MPI_DOUBLE_PRECISION,0,MPI_COMM_WORLD,ierr)
> >   42  call
> > mpi_bcast(BNAME_up,10,MPI_CHARACTER,0,MPI_COMM_WORLD,ierr)
> >   43  call mpi_bcast(NV_up,1,MPI_INTEGER,0,MPI_COMM_WORLD,ierr)
> >   44  call mpi_bcast(NE_up,1,MPI_INTEGER,0,MPI_COMM_WORLD,ierr)
> >   45  call
> > mpi_bcast(WEIGHT_up,1,MPI_DOUBLE_PRECISION,0,MPI_COMM_WORLD,ierr)
> >   46   endif
> >
> >
> > -原始邮件-
> > *发件人:* "Meng, Qingping" 
> > *发送时间:* 2014年12月8日 星期一
> > *收件人:* "A Mailing list for WIEN2k users"
> > 
> > *抄送:*
> > *主题:* [SPAM] Re: [Wien] about EELS
> >
> > Hi Kevin,
> >
> > Thank you so much. I found case.broadspec file.
> >
> > Best wishes,
> >
> > Qingping
> >
> > *From:*wien-boun...@zeus.theochem.tuwien.ac.at
> > <mailto:wien-boun...@zeus.theochem.tuwien.ac.at>
> > [mailto:wien-boun...@zeus.theochem.tuwien.ac.at
> > <mailto:wien-boun...@zeus.theochem.tuwien.ac.at>] *On Behalf Of
> > *Kevin Jorissen
> > *Sent:* Monday, December 08, 2014 9:57 AM
> > *To:* A Mailing list for WIEN2k users
> > *Subject:* Re: [Wien] about EELS
> >
> > Hi Qingping,
> >
> > I believe it's called case.broadspec.  Does that file exist?  I've
> > changed the names of some files in the new code version I'm preparing.
> >
> > Cheers,
> >
> > Kevin
> >
> > On Mon, Dec 8, 2014 at 3:53 PM, Meng, Qingping  >

Re: [Wien] Wien2k and gfortran.4.8--seeking for help

2014-12-09 Thread Peter Blaha 

Please add to line 30 an "ERR=999" statement:

READ(nfile,err=999,end=999) 
SS_dn(1),SS_dn(2),SS_dn(3),BNAME_dn,NV_dn,NE_dn,WEIGHT_dn


and let me know if this fixes the problem for gfortran.

Thanks

On 12/09/2014 10:18 AM, 李志 wrote:

Dear All,
Do you have any experience with Wien2k compiled by gfortran 4.8?
The compilation finished without error. However, when I do a test, the
calculation stops with prompt:

"At line 30 of file kptin_nv.F
Fortran runtime error: Sequential READ or WRITE not allowed after EOF
marker, possibly use REWIND or BACKSPACE"

It seems the gfortran 4.8 can't interpret the I/O command in kptin_nv.F
correctly. I also recompiled with -fno-whole-file option. Unfortunately,
the problem still appears.
I attach part of kptin_nv.F here. Anyone can help to resolve this
problem?

   1   SUBROUTINE
KPTIN_nv(NV_up,NV_dn,ende,BNAME_up,BNAME_dn,WEIGHT_up,WEIGHT_dn,SS_up,SS_dn,NE_up,NE_dn)
   2
   3   USE param
   4   USE parallel
   5
   6   IMPLICIT REAL*8(A-H,O-Z)
   7
   8 #ifdef Parallel
   9   include 'mpif.h'
  10 #endif
  11
  12   integer :: NV_up,NV_dn,ende(2)
  13   character*10 :: BNAME_up,BNAME_dn
  14   real*8 :: WEIGHT_UP,WEIGHT_dn
  15   real*8 :: SS_up(3),SS_dn(3)
  16   integer :: NE_up,NE_dn
  17   integer :: nfile
  18
  19
  20   nfile=8+1
  21  if (MYROWHS.eq.0.and.MYCOLHS.eq.0) then
  22 READ(nfile,end=998)
SS_up(1),SS_up(2),SS_up(3),BNAME_up,NV_up,NE_up,WEIGHT_up
  23 goto 111
  24 998 ende(1)=1
  25 111  continue
  26  endif
  27
  28   nfile=8+2
  29  if (MYROWHS.eq.0.and.MYCOLHS.eq.0) then
  30 READ(nfile,end=999)
SS_dn(1),SS_dn(2),SS_dn(3),BNAME_dn,NV_dn,NE_dn,WEIGHT_dn
  31 goto 112
  32 999 ende(2)=1
  33 112  continue
  34  endif
  35
  36 #ifdef Parallel
  37   call mpi_bcast(ende,2,MPI_INTEGER,0,MPI_COMM_WORLD,ierr)
  38   call mpi_barrier(MPI_COMM_WORLD,ierr)
  39
  40   if (ende(1).ne.1) then
  41  call
mpi_bcast(SS_up,3,MPI_DOUBLE_PRECISION,0,MPI_COMM_WORLD,ierr)
  42  call
mpi_bcast(BNAME_up,10,MPI_CHARACTER,0,MPI_COMM_WORLD,ierr)
  43  call mpi_bcast(NV_up,1,MPI_INTEGER,0,MPI_COMM_WORLD,ierr)
  44  call mpi_bcast(NE_up,1,MPI_INTEGER,0,MPI_COMM_WORLD,ierr)
  45  call
mpi_bcast(WEIGHT_up,1,MPI_DOUBLE_PRECISION,0,MPI_COMM_WORLD,ierr)
  46   endif


-原始邮件-
*发件人:* "Meng, Qingping" 
*发送时间:* 2014年12月8日 星期一
*收件人:* "A Mailing list for WIEN2k users"

*抄送:*
*主题:* [SPAM] Re: [Wien] about EELS

Hi Kevin,

Thank you so much. I found case.broadspec file.

Best wishes,

Qingping

*From:*wien-boun...@zeus.theochem.tuwien.ac.at

[mailto:wien-boun...@zeus.theochem.tuwien.ac.at
] *On Behalf Of
*Kevin Jorissen
*Sent:* Monday, December 08, 2014 9:57 AM
*To:* A Mailing list for WIEN2k users
*Subject:* Re: [Wien] about EELS

Hi Qingping,

I believe it's called case.broadspec.  Does that file exist?  I've
changed the names of some files in the new code version I'm preparing.

Cheers,

Kevin

On Mon, Dec 8, 2014 at 3:53 PM, Meng, Qingping mailto:qm...@bnl.gov>> wrote:

Hi Kevin,

Thank you so much for your answer. I need the data after broadening
calculation. I will plot figure myself. I do not know the name of
output file after broadening. Could you tell me the file name. Thank
you.

Best wishes,

Qingping

*From:*wien-boun...@zeus.theochem.tuwien.ac.at

[mailto:wien-boun...@zeus.theochem.tuwien.ac.at
] *On Behalf Of
*Kevin Jorissen
*Sent:* Monday, December 08, 2014 9:44 AM


*To:* A Mailing list for WIEN2k users
*Subject:* Re: [Wien] about EELS

Hi Qingping,

the unbroadened eels is in the case.elnes file.  The broadening is
done in a separate step, by the program "broadening".  You can call
it by typing "x broadening".  It uses the file "case.inb".  This
file has an ugly input format, but the parameters are explained in
the manual.  The file is written automatically by the TELNES
program.  You can change the parameters, e.g. the spectrometer
broadening (Gaussian).  You can also change the lifetime (final
state) broadening. Sometimes it is too aggressive and then it can be
better to just turn it off.

If you are using w2web, then there should be a button for broadening
in the TELNES workflow page.

The corehole lifetime broadening is taken from a table.  The
spectromete

[Wien] Wien2k and gfortran.4.8--seeking for help

2014-12-09 Thread 李志 
Dear All,
Do you have any experience with Wien2k compiled by gfortran 4.8?
The compilation finished without error. However, when I do a test, the 
calculation stops with prompt: 

"At line 30 of file kptin_nv.F
Fortran runtime error: Sequential READ or WRITE not allowed after EOF marker, 
possibly use REWIND or BACKSPACE"

It seems the gfortran 4.8 can't interpret the I/O command in kptin_nv.F 
correctly. I also recompiled with -fno-whole-file option. Unfortunately, the 
problem still appears.
   I attach part of kptin_nv.F here. Anyone can help to resolve this problem?

  1   SUBROUTINE 
KPTIN_nv(NV_up,NV_dn,ende,BNAME_up,BNAME_dn,WEIGHT_up,WEIGHT_dn,SS_up,SS_dn,NE_up,NE_dn)
  2
  3   USE param
  4   USE parallel
  5
  6   IMPLICIT REAL*8(A-H,O-Z)
  7
  8 #ifdef Parallel
  9   include 'mpif.h'
 10 #endif
 11
 12   integer :: NV_up,NV_dn,ende(2)
 13   character*10 :: BNAME_up,BNAME_dn
 14   real*8 :: WEIGHT_UP,WEIGHT_dn
 15   real*8 :: SS_up(3),SS_dn(3)
 16   integer :: NE_up,NE_dn
 17   integer :: nfile
 18
 19
 20   nfile=8+1
 21  if (MYROWHS.eq.0.and.MYCOLHS.eq.0) then
 22 READ(nfile,end=998) 
SS_up(1),SS_up(2),SS_up(3),BNAME_up,NV_up,NE_up,WEIGHT_up
 23 goto 111
 24 998 ende(1)=1
 25 111  continue
 26  endif
 27
 28   nfile=8+2
 29  if (MYROWHS.eq.0.and.MYCOLHS.eq.0) then
 30 READ(nfile,end=999) 
SS_dn(1),SS_dn(2),SS_dn(3),BNAME_dn,NV_dn,NE_dn,WEIGHT_dn
 31 goto 112
 32 999 ende(2)=1
 33 112  continue
 34  endif
 35
 36 #ifdef Parallel
 37   call mpi_bcast(ende,2,MPI_INTEGER,0,MPI_COMM_WORLD,ierr)
 38   call mpi_barrier(MPI_COMM_WORLD,ierr)
 39
 40   if (ende(1).ne.1) then
 41  call 
mpi_bcast(SS_up,3,MPI_DOUBLE_PRECISION,0,MPI_COMM_WORLD,ierr)
 42  call mpi_bcast(BNAME_up,10,MPI_CHARACTER,0,MPI_COMM_WORLD,ierr)
 43  call mpi_bcast(NV_up,1,MPI_INTEGER,0,MPI_COMM_WORLD,ierr)
 44  call mpi_bcast(NE_up,1,MPI_INTEGER,0,MPI_COMM_WORLD,ierr)
 45  call 
mpi_bcast(WEIGHT_up,1,MPI_DOUBLE_PRECISION,0,MPI_COMM_WORLD,ierr)
 46   endif


-原始邮件-
发件人: "Meng, Qingping" 
发送时间: 2014年12月8日 星期一
收件人: "A Mailing list for WIEN2k users" 
抄送:
主题: [SPAM] Re: [Wien] about EELS



Hi Kevin,

Thank you so much. I found case.broadspec file.

Best wishes,

Qingping

 

From:wien-boun...@zeus.theochem.tuwien.ac.at 
[mailto:wien-boun...@zeus.theochem.tuwien.ac.at] On Behalf Of Kevin Jorissen
Sent: Monday, December 08, 2014 9:57 AM
To: A Mailing list for WIEN2k users
Subject: Re: [Wien] about EELS

 

Hi Qingping,

 

I believe it's called case.broadspec.  Does that file exist?  I've changed the 
names of some files in the new code version I'm preparing.

 

Cheers,

 

Kevin

 

 

On Mon, Dec 8, 2014 at 3:53 PM, Meng, Qingping  wrote:

Hi Kevin,

Thank you so much for your answer. I need the data after broadening 
calculation. I will plot figure myself. I do not know the name of output file 
after broadening. Could you tell me the file name. Thank you.

Best wishes,

Qingping

 

From:wien-boun...@zeus.theochem.tuwien.ac.at 
[mailto:wien-boun...@zeus.theochem.tuwien.ac.at] On Behalf Of Kevin Jorissen
Sent: Monday, December 08, 2014 9:44 AM


To: A Mailing list for WIEN2k users
Subject: Re: [Wien] about EELS

 

Hi Qingping,

 

the unbroadened eels is in the case.elnes file.  The broadening is done in a 
separate step, by the program "broadening".  You can call it by typing "x 
broadening".  It uses the file "case.inb".  This file has an ugly input format, 
but the parameters are explained in the manual.  The file is written 
automatically by the TELNES program.  You can change the parameters, e.g. the 
spectrometer broadening (Gaussian).  You can also change the lifetime (final 
state) broadening. Sometimes it is too aggressive and then it can be better to 
just turn it off.

If you are using w2web, then there should be a button for broadening in the 
TELNES workflow page.

 

The corehole lifetime broadening is taken from a table.  The spectrometer 
broadening is taken from input (case.innes).  The final state lifetime 
broadening is a guess.  But you should not hesitate to change the parameters so 
that you best match your experiment.  (After all, the true physics is to have 
energy-dependent broadening determined by the electron self-energy, and what we 
do in WIEN2k is only an approximation.)

 

Please let me know if you need further help.

 

Cheers,

 

Kevin

 

 

On Sun, Dec 7, 2014 at 6:29 PM, Meng, Qingping  wrote:

Dear Kevin,

Thank you so much for your answer. Now I have another question. When I finished 
TELNES calculation, I need the data of broadening. Where can I find it? I only 
can find unbroadening data in case.eln