[QE-users] Installing QE 670 on AMD EPYC

[Tobias Klöffel]

Hi Chandan,

if AMD does not set any magical compiler settings during compilation you 
should get the same performance. Personally, I just use the precompiled 
binaries.

Best,
Tobias

On 3/12/21 6:24 AM, Chandan Kumar Choudhury wrote:
> Thank you @Ye, @Pietro, @Tobias, @Paolo
>
> QE successfully installed. Yay!!!
>
> @Tobias,
> I have installed AMD libraries from source (amd-fftw, bliss and libFLAME).
> Will using precompiled fftw + blis, from AMD, called AOCL in QE 
> perform better when compared to libraries installed from scratch?
>
> Thank you!
>
>
> Chandan
> --
> Chandan Kumar Choudhury, PhD
> Senior Scientist (Computational Science)
> Prescience.in
>
>> On Mar 11, 2021, at 11:42 PM, Ye Luo <xw111luoye at gmail.com 
>> <mailto:xw111luoye at gmail.com>> wrote:
>>
>> since you turned on OpenMP. Add libfftw3_omp.a
>> Ye
>> ===================
>> Ye Luo, Ph.D.
>> Computational Science Division & Leadership Computing Facility
>> Argonne National Laboratory
>>
>>
>> On Thu, Mar 11, 2021 at 11:58 AM Chandan Kumar Choudhury 
>> <ckchoud at g.clemson.edu <mailto:ckchoud at g.clemson.edu>> wrote:
>>
>>     Hi Pietro,
>>
>>     This did help. I guess I am getting closer!!
>>      I could proceed further, but still got some error.
>>     Can you please suggest a workaround?
>>     I have compiled FFTW, openmpi, and QE with CC=clang,
>>     CPP=clang-cpp CXX=clang++ and F90=flang.
>>
>>
>>
>>        pwscf.o  libpw.a ../../Modules/libqemod.a
>>     ../../KS_Solvers/libks_solvers.a ../../upflib/libupf.a
>>     ../../FFTXlib/libqefft.a ../../LAXlib/libqela.a
>>     ../../UtilXlib/libutil.a ../../dft-d3/libdftd3qe.a
>>     /home/chandan_prescience_in/softwares/q-e-qe-6.7.0//clib/clib.a
>>     /home/chandan_prescience_in/softwares/q-e-qe-6.7.0//LIBBEEF/libbeef.a
>>     /home/chandan_prescience_in/apps/scalapack/libscalapack.a
>>     /home/chandan_prescience_in/apps/libFLAME/lib/libflame.a
>>     -L/home/chandan_prescience_in/softwares/q-e-qe-6.7.0//FoX/lib 
>>     -lFoX_dom -lFoX_sax -lFoX_wxml -lFoX_common -lFoX_utils
>>     -lFoX_fsys
>>     /home/chandan_prescience_in/apps/amd-fftw/double/lib/libfftw3.a
>>     /home/chandan_prescience_in/apps/blis/lib/libblis.a
>>     -L/home/chandan_prescience_in/apps/openmpi/410/lib
>>     ld.lld: error: undefined symbol: dfftw_cleanup_threads_
>>     >>> referenced by fft_scalar.FFTW3.f90:148
>>     >>>              fft_scalar.FFTW3.o:(fft_scalar_fftw3_cft_1z_) in archive
>>     ../../FFTXlib/libqefft.a
>>     >>> referenced by fft_scalar.FFTW3.f90:301
>>     >>>              fft_scalar.FFTW3.o:(fft_scalar_fftw3_cft_2xy_) in archive
>>     ../../FFTXlib/libqefft.a
>>     >>> referenced by fft_scalar.FFTW3.f90:447
>>     >>>              fft_scalar.FFTW3.o:(fft_scalar_fftw3_cfft3d_) in archive
>>     ../../FFTXlib/libqefft.a
>>     >>> referenced 1 more times
>>
>>     ld.lld: error: undefined symbol: fftw_init_threads
>>     >>> referenced by fft_scalar.FFTW3.f90:149
>>     >>>              fft_scalar.FFTW3.o:(fft_scalar_fftw3_cft_1z_) in archive
>>     ../../FFTXlib/libqefft.a
>>     >>> referenced by fft_scalar.FFTW3.f90:302
>>     >>>              fft_scalar.FFTW3.o:(fft_scalar_fftw3_cft_2xy_) in archive
>>     ../../FFTXlib/libqefft.a
>>     >>> referenced by fft_scalar.FFTW3.f90:448
>>     >>>              fft_scalar.FFTW3.o:(fft_scalar_fftw3_cfft3d_) in archive
>>     ../../FFTXlib/libqefft.a
>>     >>> referenced 1 more times
>>
>>     ld.lld: error: undefined symbol: dfftw_plan_with_nthreads_
>>     >>> referenced by fft_scalar.FFTW3.f90:150
>>     >>>              fft_scalar.FFTW3.o:(fft_scalar_fftw3_cft_1z_) in archive
>>     ../../FFTXlib/libqefft.a
>>     >>> referenced by fft_scalar.FFTW3.f90:303
>>     >>>              fft_scalar.FFTW3.o:(fft_scalar_fftw3_cft_2xy_) in archive
>>     ../../FFTXlib/libqefft.a
>>     >>> referenced by fft_scalar.FFTW3.f90:449
>>     >>>              fft_scalar.FFTW3.o:(fft_scalar_fftw3_cfft3d_) in archive
>>     ../../FFTXlib/libqefft.a
>>     >>> referenced 1 more times
>>     clang-11: error: linker command failed with exit code 1 (use -v
>>     to see invocation)
>>     make[2]: *** [Makefile:262: pw.x] Error 1
>>     make[2]: Leaving directory
>>     '/home/chandan_prescience_in/softwares/q-e-qe-6.7.0/PW/src'
>>     make[1]: *** [Makefile:9: pw] Error 1
>>     make[1]: Leaving directory
>>     '/home/chandan_prescience_in/softwares/q-e-qe-6.7.0/PW'
>>     make: *** [Makefile:66: pw] Error 1
>>
>>
>>     Thank you!
>>
>>     Chandan
>>     --
>>     Chandan Kumar Choudhury, PhD
>>     Senior Scientist (Computational Science)
>>     Prescience.in
>>
>>>     On Mar 11, 2021, at 8:52 PM, Pietro Delugas <pdelugas at sissa.it
>>>     <mailto:pdelugas at sissa.it>> wrote:
>>>
>>>     Dear Chandan
>>>     Fox is probably compiling using a different compiler.
>>>     Check the make.inc file and see if F90 is set to the same
>>>     compiler used by mpif90
>>>     The command
>>>     mpif90  --show
>>>     will tell you  which compiler is used by mpif90
>>>     set F90 to the same compiler do make clean and then make pw again.
>>>     Removing mod files in never effective because tha make command
>>>     does not check them It check .o files only.
>>>     Do make clean.
>>>     Pietro
>>>     Sent fromMail
>>>     <https://go.microsoft.com/fwlink/?LinkId=550986>for Windows 10
>>>     *From:*Chandan Kumar Choudhury <mailto:ckchoud at g.clemson.edu>
>>>     *Sent:*Thursday, March 11, 2021 3:45 PM
>>>     *To:*Quantum ESPRESSO users Forum
>>>     <mailto:users at lists.quantum-espresso.org>
>>>     *Subject:*Re: [QE-users] Installing QE 670 on AMD EPYC
>>>     Thank you Paolo!
>>>
>>>
>>>     I did the following steps:
>>>     make pw
>>>     rm  FoX/finclude/m_common_io.mod
>>>     make clean
>>>     make pw
>>>
>>>
>>>     make clean removes the FoX directory and it recreates when we do
>>>     'make pw’
>>>     So, I still the same error
>>>     mpif90 -O3 -g -fopenmp -cpp -fopenmp -D__FFTW3 -D__MPI
>>>     -D__SCALAPACK -I/home/chandan_prescience_in/apps/blis/include
>>>     -I/home/chandan_prescience_in/apps/libFLAME/include
>>>     -I/home/chandan_prescience_in/apps/amd-fftw/double/include
>>>     -I/home/chandan_prescience_in/softwares/q-e-qe-6.7.0/FoX/finclude
>>>     -I/home/chandan_prescience_in/softwares/q-e-qe-6.7.0/include
>>>     -I/home/chandan_prescience_in/softwares/q-e-qe-6.7.0//upflib
>>>     -I/home/chandan_prescience_in/softwares/q-e-qe-6.7.0//Modules
>>>     -I/home/chandan_prescience_in/softwares/q-e-qe-6.7.0//FFTXlib
>>>     -I/home/chandan_prescience_in/softwares/q-e-qe-6.7.0//LAXlib
>>>     -I/home/chandan_prescience_in/softwares/q-e-qe-6.7.0//UtilXlib
>>>     -I/home/chandan_prescience_in/softwares/q-e-qe-6.7.0//FoX/finclude
>>>     -I../ELPA/src -c fox_init_module.f90
>>>     F90-F-0004-Corrupt or Old Module file
>>>     /home/chandan_prescience_in/softwares/q-e-qe-6.7.0/FoX/finclude/m_common_io.mod
>>>     (fox_init_module.f90: 4)
>>>     F90/x86-64 Linux Flang - 1.5 2017-05-01: compilation aborted
>>>     make[1]: *** [../make.inc:16: fox_init_module.o] Error 1
>>>     make[1]: Leaving directory
>>>     '/home/chandan_prescience_in/softwares/q-e-qe-6.7.0/Modules'
>>>     make: *** [Makefile:179: mods] Error 1
>>>
>>>
>>>     --
>>>     Chandan Kumar Choudhury, PhD
>>>     Senior Scientist (Computational Science)
>>>     Prescience.in
>>>
>>>         On Mar 11, 2021, at 7:40 PM, Paolo Giannozzi
>>>         <p.giannozzi at gmail.com <mailto:p.giannozzi at gmail.com>> wrote:
>>>         rm FoX/include/delete m_commom_io.mod
>>>         Not enough? "make clean", then "make [what-you-need]"
>>>
>>>     _______________________________________________
>>>     Quantum ESPRESSO is supported by MaX (www.max-centre.eu
>>>     <http://www.max-centre.eu/>)
>>>     users mailing listusers at lists.quantum-espresso.org
>>>     <mailto:users at lists.quantum-espresso.org>
>>>     https://lists.quantum-espresso.org/mailman/listinfo/users
>>>     <https://lists.quantum-espresso.org/mailman/listinfo/users>
>>
>>     _______________________________________________
>>     Quantum ESPRESSO is supported by MaX (www.max-centre.eu
>>     <http://www.max-centre.eu/>)
>>     users mailing list users at lists.quantum-espresso.org
>>     <mailto:users at lists.quantum-espresso.org>
>>     https://lists.quantum-espresso.org/mailman/listinfo/users
>>     <https://lists.quantum-espresso.org/mailman/listinfo/users>
>>
>> _______________________________________________
>> Quantum ESPRESSO is supported by MaX (www.max-centre.eu 
>> <http://www.max-centre.eu>)
>> users mailing list users at lists.quantum-espresso.org 
>> <mailto:users at lists.quantum-espresso.org>
>> https://lists.quantum-espresso.org/mailman/listinfo/users 
>> <https://lists.quantum-espresso.org/mailman/listinfo/users>
>
>
> _______________________________________________
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
> users mailing list users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users


-- 
M.Sc. Tobias Klöffel
=======================================================
HPC (High Performance Computing) group
Erlangen Regional Computing Center(RRZE)
Friedrich-Alexander-Universität Erlangen-Nürnberg
Martensstr. 1
91058 Erlangen

Room: 1.133
Phone: +49 (0) 9131 / 85 - 20101

=======================================================

E-mail: tobias.kloeffel at fau.de

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20210313/092ede9a/attachment.html>