[CP2K-user] List of supported GPUs?

[Ryan Kingsbury]

Thank you, I didn't realize the other howto page had more detailed CUDA 
instructions. This is the arch file that generates the error above:

CC       = cc
CPP      =
FC       = gfortran
LD       = gfortran
AR       = ar -r
CPPFLAGS =
DFLAGS   = -D__FFTSG
FCFLAGS  = -O2 -fopenmp -funroll-loops -ftree-vectorize -ffree-form $(DFLAGS
)
LDFLAGS  = $(FCFLAGS)
LIBS     =  -llapack -lblas
NVCC    = "/cygdrive/c/Program Files/NVIDIA GPU Computing 
Toolkit/CUDA/v10.2/bin/nvcc"
DFLAGS += -D__ACC -D__DBCSR_ACC -D__PW_CUDA
LIBS   += -lcudart -lcublas -lcufft -lnvrtc

Based on the more detailed instructions, I revised the arch file to the 
following, but still get the 'Cannot open source file' error

CC       = cc
CPP      =
FC       = gfortran
LD       = gfortran
AR       = ar -r
CPPFLAGS =
DFLAGS   = -D__FFTSG
FCFLAGS  = -O2 -fopenmp -funroll-loops -ftree-vectorize -ffree-form $(DFLAGS
)
LDFLAGS  = $(FCFLAGS)
LIBS     =  -llapack -lblas
NVCC    = "/cygdrive/c/Program Files/NVIDIA GPU Computing 
Toolkit/CUDA/v10.2/bin/nvcc"
DFLAGS += -D__ACC -D__DBCSR_ACC -D__PW_CUDA -D_HAS_PATCHED_CUFFT_70
LIBS   += -lcudart -lcublas -lcufft -lnvrtc -lstdc++ -lcuda
NVFLAGS = --std=c++11 -arch=compute_50 -code=sm_50 -O3






On Thursday, March 19, 2020 at 12:32:52 AM UTC-7, Fabian Ducry wrote:
>
> Here are more detailed instructions
> cuda-optional-improved-performance-on-gpu-systems 
> <https://github.com/cp2k/cp2k/blob/master/INSTALL.md#2j-cuda-optional-improved-performance-on-gpu-systems>
> I don't know if that helps.
>
> Could you post the arch file?
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20200319/cc72759e/attachment.htm>