compile AkaiKKR on GPU

Posted on : February 22, 2024 (Thu) 04:02:53

by Jerry Lee

Hello everyone,
I am trying to compile akaiKKR cpa2021v02gpu.tgz (February 01, 2024 [ 13:24:05 ]) on GPU NVIDIA Quadro M4000.
I got the following error:
"
ld: source/spmain.o: in function `spmain_':
spmain.f:(.text+0x15a89): undefined reference to `vcancy_'
make: *** [makefile:286: specx_GPU] Error 1
"

Cuda compilation tools, release 12.3, V12.3.107
Build cuda_12.3.r12.3/compiler.33567101_0

Your kind help is highly appreciated.

best
Jerry
~~~~~~~~~~~~Following is my makefile~~~~~~~~~~~~~~~~~~~~~~~

# ThecCommand invoking the fortran compiler of the
# system should be specified by "fort= ".
# The corresponding compiler option also should be given
# by "flag= ".
#
.PHONY: all
#all: $(program) gpd
#######################################
# specify the fortran compiler
fort = ifort
#######################################
ifeq ($(fort),ifort)
CC = icpc
flag = -O2 -mcmodel=medium
omp = -qopenmp
nomp = -qopenmp-stubs
# libs = -lstdc++ -L /octfs/apl/CUDA/cuda-9.0/lib64/ -lcudart -lcublas -L /octfs/apl/CUDA/cuda-9.0/lib64/stubs -lcuda
libs = -lstdc++ -L /usr/local/cuda-12.3/lib64/ -lcudart -lcublas -L /usr/local/cuda-12.3/lib64/stubs -lcuda
CFLAGS = -O3 -mcmodel=medium -shared-intel
else
flag = -O2
omp = -fopenmp -w
nomp =
endif
NVCC = nvcc
NVFLAGS = -arch=native --ptxas-options=-v -pg
#
#######################################
# Intel compiler used
# In the case of ifort 15.x.x or ealier
#######################################
#
program = specx_GPU
#all: $(program) gpd spc
objs = \
source/specx.o \
source/spmain.o \
source/alphac.o \