Hi Jason,
I added the libraries you suggested in the config.h file.
Then pmemd.cuda was installed OK. Thank you!
However the pmemd.cuda does not pass the tests.
Here is the error message I get after typing 'make test.cuda'
Using default PREC_MODEL = SPDP
make[2]: Entering directory `/usr/local/amber12/test'
cd cuda && make -k test.pmemd.cuda PREC_MODEL=SPDP
make[3]: Entering directory `/usr/local/amber12/test/cuda'
------------------------------------
Running CUDA Implicit solvent tests.
Precision Model = SPDP
------------------------------------
cd gb_ala3/ && ./Run.igb1_ntc1_min SPDP /usr/local/amber12/include/netcdf.mod
Currently only GPU SM Revisions 1.3 and 2.X are supported.
./Run.igb1_ntc1_min: Program error
make[3]: *** [test.pmemd.cuda.gb.serial] Error 1
My Graphics Card is NVIDIAGTX680 which I think supports
GPU SM Revision 3.0.
Any idea how I can cope with this?
Regards,
Cenk Andac, M.S., Ph.D.
Asst. Professor of Medical Pharmacology
Department of Medical Pharmacology
School of Medicine, Mevlana University, Konya
Turkiye 42003
Cell: +90-(536)-4813012
E-Mail:cenk_andac.yahoo.com
Web : http://med.mevlana.edu.tr/personel.asp?ID=1
________________________________
From: Jason Swails <jason.swails.gmail.com>
To: Cenk (Jenk) Andac <cenk_andac.yahoo.com>; AMBER Mailing List <amber.ambermd.org>
Sent: Wednesday, May 15, 2013 8:03 PM
Subject: Re: [AMBER] pmemd.CUDA installation problem
Try adding the following to the end of the PMEMD_CU_LIBS variable in your
config.h file:
-L/usr/lib64 -lstdc++
Then try re-compiling.
The problem is that the standard C++ functions could not be found by the
linker...
HTH,
Jason
On Wed, May 15, 2013 at 12:38 PM, Cenk (Jenk) Andac <cenk_andac.yahoo.com>wrote:
> Dear all,
>
> I have recently purchased an NVIDIA GTX680 Graphiics card and installed it
> on my desktop
> OS is OpenSuse 12.1. I have installed the NVIDIA ToolsKit 5 as well as the
> driver for the new graphics card.
> I set the environment to point to /usr/local/cuda for CUDA_HOME,
> $CUDA_HOME/bin for the binaries and $CUDA_HOME/lib & lib64 for CUDA
> libraries
>
> I use Intel fortran and cc compilers v13
>
> Basically AMBER12 serial and parallel installations went OK and passed the
> tests..
>
>
> I configured AMBER12 CUDA as follows
>
> ./configure -cuda intel
>
> everything seems to be OK in the config.h file.
>
> NVCC=$(CUDA_HOME)/bin/nvcc -use_fast_math -O3 -gencode
> arch=compute_13,code=sm_13 -gencode arch=compute_20,code=sm_20
> PMEMD_CU_INCLUDES=-I$(CUDA_HOME)/include -IB40C -IB40C/KernelCommon
> PMEMD_CU_LIBS=-L$(CUDA_HOME)/lib64 -L$(CUDA_HOME)/lib -lcurand -lcufft
> -lcudart ./cuda/cuda.a
> PMEMD_CU_DEFINES=-DCUDA
>
> #PMEMD Specific build flags
> PMEMD_F90=ifort -DMKL -DBINTRAJ -DDIRFRC_EFS -DDIRFRC_COMTRANS
> -DDIRFRC_NOVEC -DFFTLOADBAL_2PROC -DPUBFFT
> PMEMD_FOPTFLAGS=-ip -O3 -no-prec-div -xHost
> PMEMD_CC=icc
> PMEMD_COPTFLAGS=-ip -O3 -no-prec-div -xHost -D_FILE_OFFSET_BITS=64
> -D_LARGEFILE_SOURCE -DBINTRAJ
> PMEMD_FLIBSF=-Wl,--start-group
> /opt/intel/composer_xe_2013.3.163/mkl/lib/intel64/libmkl_intel_lp64.a
> /opt/intel/composer_xe_2013.3.163/mkl/lib/intel64/libmkl_sequential.a
> /opt/intel/composer_xe_2013.3.163/mkl/lib/intel64/libmkl_core.a
> -Wl,--end-group -lpthread
> PMEMD_LD= ifort
> LDOUT= -o
>
>
> However, I seem to have a problem with pmemd.CUDA installation.
> I get the following error message after I type make install
>
>
> -L/usr/local/cuda-5.0/lib64 -L/usr/local/cuda-5.0/lib -lcurand -lcufft
> -lcudart ./cuda/cuda.a -L/usr/local/amber12/lib -L/usr/local/amber12/lib
> -lnetcdf -shared-intel -Wl,--start-group
> /opt/intel/composer_xe_2013.3.163/mkl/lib/intel64/libmkl_intel_lp64.a
> /opt/intel/composer_xe_2013.3.163/mkl/lib/intel64/libmkl_sequential.a
> /opt/intel/composer_xe_2013.3.163/mkl/lib/intel64/libmkl_core.a
> -Wl,--end-group -lpthread
> ld: ./cuda/cuda.a(kNLRadixSort.o): undefined reference to symbol
> '__cxa_pure_virtual..CXXABI_1.3'
> ld: note: '__cxa_pure_virtual..CXXABI_1.3' is defined in DSO
> /usr/lib64/libstdc++.so.6 so try adding it to the linker command line
> /usr/lib64/libstdc++.so.6: could not read symbols: Invalid operation
> make[3]: *** [pmemd.cuda] Error 1
> make[3]: Leaving directory `/usr/local/amber12/src/pmemd/src'
> make[2]: *** [cuda] Error 2
> make[2]: Leaving directory `/usr/local/amber12/src/pmemd'
> make[1]: *** [cuda] Error 2
> make[1]: Leaving directory `/usr/local/amber12/src'
> make: *** [install] Error 2
>
>
> I guess the error is related to "/usr/lib64/libstdc++.so.6: could not read
> symbols: Invalid operation"
>
> Does anyone have a clue what might be going wrong with the installation?
>
>
> Best regards,
>
> Cenk Andac, M.S., Ph.D.
>
> Asst. Professor of Medical Pharmacology
> Department of Medical Pharmacology
> School of Medicine, Mevlana University, Konya
> Turkiye 42003
>
>
> Cell: +90-(536)-4813012
> E-Mail:cenk_andac.yahoo.com
> Web : http://med.mevlana.edu.tr/personel.asp?ID=1
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed May 15 2013 - 23:30:02 PDT
