09 May 2013

409.A.GAMESS US with GPU support on debian wheezy. This works (probably).

Update 27/6/2013:
Please note that Kirill Berezovsky has published a series of posts on GAMESS US, including how to compile it for both CPU and GPU use. See

Update 21 May 2013: See the comments below this post. This approach most likely works -- what has been confusing me is the lack of reports of GPU timings in the output, but this doesn't necessarily mean that the GPU isn't being used. The poster below this post, using nvidia-smi, observed GPU usage, although the speed-up was not major.

Update 10/05/2013: fixed libcchem compile.

Everything compiles fine and computations run fine and fast. To date there's only one other detailed step-by-step example of successful compilation of GAMESS with GPU support out there. At least based on google.

For various reasons I'm beginning to suspect that ATLAS isn't working out for me -- I've had issues getting things to converge with ATLAS, but which work fine with ACML (see post B).

I was in part following http://combichem.blogspot.com.au/2011/02/compiling-gamess-with-cuda-gpu-support.html and ./libcchem/aaa.readme.1st

This took a while to hammer out, so the write-up is a bit messy.

Set up
sudo apt-get install libboost-all-dev build-essential g++ gfortran automake nvidia-cuda-toolkit python-cheetah openmpi-bin libopenmpi-dev zlib1g-dev checkinstall
mkdir ~/tmp

Get gamess (see e.g. http://verahill.blogspot.com.au/2012/09/compiling-and-testing-gamess-us-on.html).

Put gamess-current.tar.gz in  ~/tmp

cd ~/tmp
tar xvf gamess-current.tar.gz
sudo mv gamess /opt/gamess_cuda
sudo chown $USER:$USER /opt/gamess_cuda -R

Preparing Boost
Edit /usr/include/boost/mpl/aux_/integral_wrapper.hpp
47 // other compilers (e.g. MSVC) are not particulary happy about it 48 #if BOOST_WORKAROUND(__EDG_VERSION__, <= 238) || defined(__CUDACC__) 49 typedef struct AUX_WRAPPER_NAME type;
Edit /usr/include/boost/mpl/size_t_fwd.hpp
20 21 BOOST_MPL_AUX_ADL_BARRIER_NAMESPACE_OPEN 22 #if defined(__CUDACC__) 23 typedef std::size_t std_size_t; 24 template< std_size_t N > struct size_t; 25 #else 26 template< std::size_t N > struct size_t; 27 #endif 28 29 BOOST_MPL_AUX_ADL_BARRIER_NAMESPACE_CLOSE
Edit /usr/include/boost/mpl/size_t.hpp
19 #if defined(__CUDACC__) 20 #define AUX_WRAPPER_VALUE_TYPE std_size_t 21 #define AUX_WRAPPER_NAME size_t 22 #define AUX_WRAPPER_PARAMS(N) std_size_t N 23 #else 24 #define AUX_WRAPPER_VALUE_TYPE std::size_t 25 #define AUX_WRAPPER_NAME size_t 26 #define AUX_WRAPPER_PARAMS(N) std::size_t N 27 #endif 28

You'll have to compile that yourself for now since H5Cpp.h missing in the debian packages.(i.e. cxx support)

mkdir ~/tmp
cd ~/tmp
wget http://www.hdfgroup.org/ftp/HDF5/current/src/hdf5-1.8.10-patch1.tar.gz
tar xvf hdf5-1.8.10-patch1.tar.gz
cd hdf5-1.8.10-patch1/
export CC=/usr/bin/gcc-4.6 && export CXX=/usr/bin/g++-4.6
./configure --prefix=/opt/gamess_cuda/hdf5 --with-pthread --enable-cxx --enable-threadsafe --enable-unsupported
mkdir /opt/gamess_cuda/hdf5/lib -p
mkdir /opt/gamess_cuda/hdf5/include -p
sudo checkinstall
This package will be built according to these values: 0 - Maintainer: [ root@neon ] 1 - Summary: [ hdf5-cxx] 2 - Name: [ hdf5-1.8.10 ] 3 - Version: [ 1.8.10-1 ] 4 - Release: [ 1 ] 5 - License: [ GPL ] 6 - Group: [ checkinstall ] 7 - Architecture: [ amd64 ] 8 - Source location: [ hdf5-1.8.10-patch1 ] 9 - Alternate source location: [ ] 10 - Requires: [ ] 11 - Provides: [ hdf5-1.8.10 ] 12 - Conflicts: [ ] 13 - Replaces: [ ]
Make sure to edit the Version field since Patch-1 leads to an error (must start with digit).
Openmpi 1.6 Can't remember why I ended up compiling it myself instead of using the stock debian version. From here.

sudo apt-get install build-essential gfortran
wget http://www.open-mpi.org/software/ompi/v1.6/downloads/openmpi-1.6.tar.bz2
tar xvf openmpi-1.6.tar.bz2
cd openmpi-1.6/

sudo mkdir /opt/openmpi/
sudo chown ${USER} /opt/openmpi/
./configure --prefix=/opt/openmpi/1.6/ --with-sge

make install

compiling libcchem
cd /opt/gamess_cuda/libcchem
edit /opt/gamess_cuda/libcchem/rysq/src/externals/boost/cuda/device_ptr.hpp
  4 #include <cstdlib>
  5 #include <iterator>
  6 #include <stddef.h>
  8 namespace boost {
Edit /opt/gamess_cuda/libcchem/src/externals/boost/cuda/device_ptr.hpp
  4 #include <cstdlib>
  5 #include <iterator>
  6 #include <stddef.h>
  8 namespace boost {
  9 namespace cuda {
./configure --with-gamess --with-hdf5=/opt/gamess_cuda/hdf5 CPPFLAGS="-I/opt/gamess_cuda/hdf5/include" --with-cuda=/usr --disable-openmp --prefix=/opt/gamess_cuda/libcchem --with-gpu=fermi --with-integer8 --with-cublas
make install

Configure Gamess US Mainly follow this: http://verahill.blogspot.com.au/2012/09/compiling-and-testing-gamess-us-on.html
cd /opt/gamess_cuda
please enter your target machine name: linux64 GAMESS directory? [/opt/gamess_cuda] /opt/gamess_cuda Setting up GAMESS compile and link for GMS_TARGET=linux64 GAMESS software is located at GMS_PATH=/opt/gamess_cuda Please provide the name of the build locaation. This may be the same location as the GAMESS directory. GAMESS build directory? [/home/me/tmp/gamess] Please provide a version number for the GAMESS executable. This will be used as the middle part of the binary's name, for example: gamess.00.x Version? [00] 12r2 Please enter your choice of FORTRAN: gfortran gfortran is very robust, so this is a wise choice. Please type 'gfortran -dumpversion' or else 'gfortran -v' to detect the version number of your gfortran. This reply should be a string with at least two decimal points, such as 4.1.2 or 4.6.1, or maybe even 4.4.2-12. The reply may be labeled as a 'gcc' version, but it is really your gfortran version. Please enter only the first decimal place, such as 4.1 or 4.6: 4.6
Enter your choice of 'mkl' or 'atlas' or 'acml' or 'none': atlas Please enter the Atlas subdirectory on your system: /opt/ATLAS/lib Math library 'atlas' will be taken from /opt/ATLAS If you have an expensive but fast network like Infiniband (IB), and if you have an MPI library correctly installed, choose 'mpi'. communication library ('sockets' or 'mpi')? mpi Enter MPI library (impi, mvapich2, mpt, sockets): openmpi
Please enter your openmpi's location: /opt/openmpi/1.6

Build Gamess US
cd /opt/gamess_cuda/ddi/
cd ../

Edit comp
872 # see ~/gamess/libcchem/aaa.readme.1st for more information 873 set GPUCODE=true 874 if ($GPUCODE == true) then
1663 # -fno-whole-file suppresses argument's data type checking 1664 set OPT='-O0' 1665 if (".$GMS_DEBUG_FLAGS" != .) set OPT="$GMS_DEBUG_FLAGS"

Edit lked
69 # 70 set GPUCODE=true 71 # 72 # 5. optional MPQC interface
958 case openmpi: 959 set MPILIBS="-L$GMS_MPI_PATH/lib" 960 set MPILIBS="$MPILIBS -lmpi -lpthread" 961 breaksw
1214 if ($GPUCODE == true) then 1215 echo " Using 'libcchem' add-in C++ codes for Nvidia/CUDA GPUs." 1216 set GPU_LIBS="-L/opt/gamess_cuda/libcchem/lib -lcchem_gamess -lcchem -lrysq" 1217 set GPU_LIBS="$GPU_LIBS -lcudart -lcublas" 1218 ### GPU_LIBS="$GPU_LIBS -lcudart -lcublas" 1219 set GPU_LIBS="$GPU_LIBS /usr/lib/libboost_thread.a" 1220 set GPU_LIBS="$GPU_LIBS /opt/gamess_cuda/hdf5/lib/libhdf5.a" 1221 set GPU_LIBS="$GPU_LIBS /opt/gamess_cuda/hdf5/lib/libhdf5_cpp.a" 1222 set GPU_LIBS="$GPU_LIBS /opt/gamess_cuda/hdf5/lib/libhdf5_hl.a" 1223 set GPU_LIBS="$GPU_LIBS /opt/gamess_cuda/hdf5/lib/libhdf5.a" 1224 set GPU_LIBS="$GPU_LIBS /opt/ATLAS/lib/libcblas.a" 1225 set GPU_LIBS="$GPU_LIBS -lz" 1226 set GPU_LIBS="$GPU_LIBS -lstdc++" 1227 ### GPU_LIBS="$GPU_LIBS -lgomp" 1228 set GPU_LIBS="$GPU_LIBS -lpthread" 1229 echo " libcchem GPU code's libraries are" 1230 echo "$GPU_LIBS" 1231 else

./lked gamess gpu.12

Create gpurun
#!/bin/csh set TARGET=mpi set SCR=$HOME/scratch set USERSCR=/scratch set GMSPATH=/opt/gamess_cuda set JOB=$1 set VERNO=$2 set NCPUS=$3 @ NUMGPU=1 if ($NUMGPU > 0) then @ NUMCPU = $NCPUS - 1 echo libcchem kernels will use $NUMCPU cores and $NUMGPU GPUs per node... set echo setenv CCHEM_PROFILE 1 setenv NUM_THREADS $NCPUS #--if ($NUMGPU == 0) setenv GPU_DEVICES -1 #--if ($NUMGPU == 2) setenv GPU_DEVICES 0,1 #--if ($NUMGPU == 4) setenv GPU_DEVICES 0,1,2,3 #setenv LD_LIBRARY_PATH /share/apps/cuda/lib64:$LD_LIBRARY_PATH ###### LD_LIBRARY_PATH /usr/local/cuda/lib64:$LD_LIBRARY_PATH unset echo else setenv GPU_DEVICES -1 endif if ( $JOB:r.inp == $JOB ) set JOB=$JOB:r echo "Copying input file $JOB.inp to your run's scratch directory..." cp $JOB.inp $SCR/$JOB.F05 setenv TRAJECT $USERSCR/$JOB.trj setenv RESTART $USERSCR/$JOB.rst setenv INPUT $SCR/$JOB.F05 setenv PUNCH $USERSCR/$JOB.dat if ( -e $TRAJECT ) rm $TRAJECT if ( -e $PUNCH ) rm $PUNCH if ( -e $RESTART ) rm $RESTART source $GMSPATH/gms-files.csh setenv LD_LIBRARY_PATH /opt/openmpi/lib:$LD_LIBRARY_PATH set path= ( /opt/openmpi/bin $path ) mpiexec -n $NCPUS $GMSPATH/gamess.gpu.$VERNO.x|tee $JOB.out cp $PUNCH .

echo 'export PATH=$PATH:/opt/gamess_cuda' >> ~/.bashrc
source ~/.bashrc
chmod +x gpurun
cd test/standard/
 gpurun exam44 12 2

The only evidence of GPU usage in the output is e.g. in exam44.out:
388           -----------------------
390           -----------------------
391           NACORE =        6  NBCORE =        6
392           LMOMP2 =        F  AOINTS = DUP
393           METHOD =        2  NWORD  =               0
394           MP2PRP =        F  OSPT   = NONE
395           CUTOFF = 1.00E-09  CPHFBS = BASISAO
396           CODE   = GPU
398           NUMBER OF CORE -A-  ORBITALS =     6
399           NUMBER OF CORE -B-  ORBITALS =     6

but in the summary only CPU utilisation is mentioned.

I modified rungms:

me@neon:/opt/gamess_cuda/tests/standard$ diff /opt/gamess_cuda/gpurungms /opt/gamess/rungms 
< set TARGET=mpi
< set SCR=$HOME/scratch
< set USERSCR=/scratch
< set GMSPATH=/opt/gamess_cuda
> set TARGET=sockets
> set SCR=/scr/$USER
> set USERSCR=~$USER/scr
> set GMSPATH=/u1/mike/gamess
< set NNODES=1
< set PPN=$3
>    set PPN=$4
<          @ PPN2 = $PPN
>          @ PPN2 = $PPN + $PPN
<    @ NUMGPU=1
>    @ NUMGPU=0
< #      setenv LD_LIBRARY_PATH /share/apps/cuda/lib64:$LD_LIBRARY_PATH
>       setenv LD_LIBRARY_PATH /share/apps/cuda/lib64:$LD_LIBRARY_PATH
<       /opt/openmpi/1.6/bin/mpiexec -n $NPROCS $GMSPATH/gamess.$VERNO.x < /dev/null
>       mpiexec.hydra -f $PROCFILE -n $NPROCS \
>             /home/mike/gamess/gamess.$VERNO.x < /dev/null

No comments:

Post a Comment