dir_guard = @mkdir -p $(@D)

INTEL_ROOT := /opt/intel/oneapi
MKL_ROOT = /opt/intel/oneapi/mkl/latest
TBB_ROOT = /opt/intel/oneapi/tbb/latest
USE_GPU = 1

FIND := find
CXX := icc
CXXFLAGS += -Wall -O3 -std=c++14
LDFLAGS += -lm  -lmkl_rt -ltbb
ifeq ($(USE_GPU), 1)
    CUDA_HOME := /usr/local/cuda-11.2
    NVCC := $(CUDA_HOME)/bin/nvcc
    NVCCFLAGS += --default-stream per-thread
    LDFLAGS += -L$(CUDA_HOME)/lib64 -lcudart -lcublas -lcurand -lcusparse
endif

CUDA_ARCH :=    -gencode arch=compute_80,code=sm_80 \
		-gencode arch=compute_35,code=sm_35 \
		-gencode arch=compute_50,code=sm_50 \
		-gencode arch=compute_50,code=compute_50