Makefile

PULP_APP = mrbci

PULP_APP_FC_SRCS = \
    src/fc/main.c\
    src/cl/cluster.c \
    src/cl/input.c \
	src/cl/func/convert.c \
	src/cl/func/copy_mat.c \
	src/cl/func/covmat.c \
	src/cl/func/matmul.c \
	src/cl/func/sos_filt.c \
	src/cl/func/swap_mat.c \
	src/cl/linalg/matop_f.c \
	src/cl/linalg/svd.c \
	src/cl/linalg/svd_parallel.c \
	src/cl/mrbci/covmat.c \
	src/cl/mrbci/feature_extraction.c \
	src/cl/mrbci/filter.c \
	src/cl/mrbci/half_diag.c \
	src/cl/mrbci/logm.c \
	src/cl/mrbci/mrbci.c \
	src/cl/mrbci/mrbci_params.c \
	src/cl/mrbci/svm.c \
	src/cl/mrbci/whitening.c \
    dsp/plp_mat_mult_i32.c \
    dsp/plp_mat_mult_i32_parallel.c \
    dsp/plp_mat_mult_i16.c \
    dsp/plp_mat_mult_i16_parallel.c \
    dsp/kernels/plp_mat_mult_i32s_rv32im.c \
    dsp/kernels/plp_mat_mult_i32s_xpulpv2.c \
    dsp/kernels/plp_mat_mult_i32p_xpulpv2.c \
    dsp/kernels/plp_mat_mult_i16p_xpulpv2.c \
    dsp/kernels/plp_mat_mult_i16s_rv32im.c \
    dsp/kernels/plp_mat_mult_i16s_xpulpv2.c


PULP_CFLAGS = -O3 -g

# the vega sdk is not stable yet, so I can't compile the pulp-dsp library and install in it, besides the library has to be readapted for vega. Hence, I copied the needed functions from pulp-dsp library to folder dsp
IDIR=$(CURDIR)/dsp
PULP_CFLAGS += -I$(IDIR)

#PULP_LDFLAGS += -lplpdsp -lm
PULP_LDFLAGS += -lm

# enable slow householder
# PULP_CFLAGS += -DHOUSEHOLDER_SLOW

# enable parallel computation
PULP_CFLAGS += -DPARALLEL

# enable FMA support
PULP_CFLAGS += -DUSE_FUSED_FPU

# disable square root unit
# PULP_CFLAGS += -DUSE_SOFT_SQRTDIV
# PULP_CFLAGS += -mno-fdiv

# do Power Measurement
# PULP_CFLAGS += "-DPOWER"

include $(GAP_SDK_HOME)/tools/rules/pulp_rules.mk