Page 1 of 1


PostPosted: Thu Feb 25, 2016 5:52 pm
by kachme
Hello dear VASP team,

last week I compiled the GPU version of VASP with this Makefile:

Code: Select all
# Precompiler options
CPP_OPTIONS= -DMPI -DHOST=\"Lichteb-5.41-gpu-half\" -DIFC \
             -DNGXhalf -DCACHE_SIZE=4000 -DPGF90 -Davoidalloc \
             -DMPI_BLOCK=8000 -Duse_collective \
             -DnoAugXCmeta -Duse_bse_te \
             -Duse_shmem -Dkind8

CPP        = fpp -f_com=no -free -w0  $*$(FUFFIX) $*$(SUFFIX) $(CPP_OPTIONS)

FC         = mpiifort
FCL        = mpiifort -mkl -lstdc++

FREE       = -free -names lowercase

FFLAGS     = -assume byterecl
OFLAG      = -O2
DEBUG      = -O0

MKL_PATH   = $(MKLROOT)/lib/intel64
BLAS       =
LAPACK     =
BLACS      = -lmkl_blacs_intelmpi_lp64
SCALAPACK  = $(MKL_PATH)/libmkl_scalapack_lp64.a $(BLACS)

OBJECTS    = fftmpiw.o fftmpi_map.o fft3dlib.o fftw3d.o \

INCS       =-I$(MKLROOT)/include/fftw


#OBJECTS_O1 += fft3dfurth.o fftw3d.o fftmpi.o fftmpiw.o
OBJECTS_O1 += fftw3d.o fftmpi.o fftmpiw.o
OBJECTS_O2 += fft3dlib.o

# For what used to be vasp.5.lib
CPP_LIB    = $(CPP)
FC_LIB     = $(FC)
CC_LIB     = icc

OBJECTS_LIB= linpack_double.o getshmem.o

# Normally no need to change this
SRCDIR     = ../../src
BINDIR     = ../../bin

# GPU Stuff


OBJECTS_GPU = fftmpiw.o fftmpi_map.o fft3dlib.o fftw3d_gpu.o fftmpiw_gpu.o \

CUDA_ROOT  := /shared/apps/cuda/7.5
NVCC       := $(CUDA_ROOT)/bin/nvcc
CUDA_LIB   := -L$(CUDA_ROOT)/lib64 -L$(CUDA_ROOT)/lib64/stubs -lnvToolsExt -lcudart -lcuda -lcufft -lcublas

GENCODE_ARCH    := -gencode=arch=compute_30,code=\"sm_30,compute_30\" -gencode=arch=compute_35,code=\"sm_35,compute_35\"

MPI_INC    =/shared/apps/intel/2016u2/compilers_and_libraries_2016.2.181/linux/mpi/intel64/include

However, when I try to run a calculation on one node (16 cores 32 GByte main memory, 2x “NVIDIA® Tesla™ K20X”) I get the following error message for a system of 318 atoms.

Code: Select all
creating 32 CUFFT plans with grid size 126 x 126 x 126...

 Failed to create CUFFT plan!

which refers to some kind of problem with memory (although in the cpu version it runs without problems). And the memory usage is small (from the the LSF outfile)

Code: Select all
Exited with exit code 255.

Resource usage summary:

    CPU time :               35.24 sec.
    Max Memory :             195 MB
    Average Memory :         195.00 MB
    Total Requested Memory : 28016.00 MB
    Delta Memory :           27821.00 MB
    (Delta: the difference between total requested memory and actual max usage.)
    Max Processes :          8
    Max Threads :            9

If I reduce the system size (40 atoms, 2x2x2 KP) it runs without errors, but very slowly: ~10times slower than the cpu version. I even aborted the 4x4x4 KPOINTS run, because it was just too slow. Playing around with NSIM doesn*t seem to change much.

My guess is that is has something to do with the compilation. I would like to experiment with values given in the first part of the Makefile, the CPP_OPTIONS (-DCACHE_SIZE=4000, -DMPI_BLOCK=8000), but I have no idea which values to plug in.

Help would be very much appreciated, thank you,
Kai Meyer


PostPosted: Thu Dec 21, 2017 3:29 am
by zhouych
Hi Kai,

I encounter the same issue. Have you solved it? Thanks.

Yecheng Zhou


PostPosted: Wed Jun 13, 2018 4:19 pm
by jperaltac
Hello, I have the same issue with vasp 5.4.4 and cuda 9.1

There is any hint or solution to this issue?

Thanks in advance