# Choose your word size
WSIZE = 64
OPTLEVEL = 1
# WSIZE = 32
CC = gcc
GFLAGS = -pg 
INC = ./include
MCFLAGS = -g -Wall -I. -I$(INC) -msse4
# From ics2:
# MCFLAGS = -g -Wall -O$(OPTLEVEL) -I. -I$(INC) -msse4 -fno-inline -fno-optimize-sibling-calls -lrt -mfpmath=sse
# Remove -m sse4
# MCFLAGS = -g -Wall -O$(OPTLEVEL) -I. -I$(INC) -fno-inline -fno-optimize-sibling-calls -lrt -mfpmath=sse
CFLAGS = $(MCFLAGS) -m$(WSIZE)
F386 = -march=i386
F686 = -march=i686
LIB = ./lib
LIBCSAPP = $(LIB)/libcsapp$(WSIZE).a -lrt
GETASM = ../getasmfun.pl
RUNPOS = 0.8
RUNCNT = 50
MULTIRUN = ./multirun.pl -n $(RUNCNT) -p $(RUNPOS)
BXTR = ./bxtr.pl
VXTR = ./vxtr.pl
TABULATE = ./tabulate.pl
# Choose your machine
# MACHINE = p4
# MACHINE = athlon
# MACHINE = haswell
MACHINE=machine
RESULTS=results

PROGS = ibench\
	lbench\
	fbench\
	dbench\

all: bench

bench: ibench lbench fbench dbench  

ibench:   vec.h poly.h poly.c pbenchmark.c vec.c $(LIBCSAPP)
	$(CC) $(CFLAGS) -O$(OPTLEVEL) -DINT -o ibench pbenchmark.c poly.c vec.c $(LIBCSAPP)
	$(MULTIRUN) -c ./ibench > $(RESULTS)/ib-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).txt
	$(BXTR) < $(RESULTS)/ib-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).txt > $(RESULTS)/ib-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).tab
	$(VXTR) < $(RESULTS)/ib-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).txt > $(RESULTS)/ib-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).vtab

lbench:   vec.h poly.h poly.c pbenchmark.c vec.c $(LIBCSAPP)
	$(CC) $(CFLAGS) -O$(OPTLEVEL) -DLONG -o lbench pbenchmark.c poly.c vec.c $(LIBCSAPP)
	$(MULTIRUN) -c ./lbench > $(RESULTS)/lb-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).txt
	$(BXTR) < $(RESULTS)/lb-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).txt > $(RESULTS)/lb-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).tab
	$(VXTR) < $(RESULTS)/lb-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).txt > $(RESULTS)/lb-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).vtab

fbench:   vec.h poly.h poly.c pbenchmark.c vec.c $(LIBCSAPP)
	$(CC) $(CFLAGS) -O$(OPTLEVEL) -DFLOAT -o fbench pbenchmark.c poly.c vec.c $(LIBCSAPP)
	$(MULTIRUN) -c ./fbench > $(RESULTS)/fb-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).txt
	$(BXTR) < $(RESULTS)/fb-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).txt > $(RESULTS)/fb-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).tab
	$(VXTR) < $(RESULTS)/fb-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).txt > $(RESULTS)/fb-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).vtab

dbench:   vec.h poly.h poly.c pbenchmark.c vec.c $(LIBCSAPP)
	$(CC) $(CFLAGS) -O$(OPTLEVEL) -DDOUBLE -o dbench pbenchmark.c poly.c vec.c $(LIBCSAPP)
	$(MULTIRUN) -c ./dbench > $(RESULTS)/db-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).txt
	$(BXTR) < $(RESULTS)/db-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).txt > $(RESULTS)/db-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).tab
	$(VXTR) < $(RESULTS)/db-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).txt > $(RESULTS)/db-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).vtab

tabulate: ibench lbench fbench dbench
	$(TABULATE) -w $(WSIZE) -d $(RESULTS) -m $(MACHINE)-O$(OPTLEVEL) > $(RESULTS)/summary-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).txt

bench-ncpe:	 $(LIBCSAPP)   vec.h poly.h poly.c benchmark-ncpe.c vec.c
	$(CC) $(CFLAGS) -O$(OPTLEVEL) -DSUM -DLONG -DUNROLL_ONLY -DCNT=31 -o bench-ncpe31 benchmark-ncpe.c poly.c vec.c $(LIBCSAPP) 
	$(CC) $(CFLAGS) -O$(OPTLEVEL) -DSUM -DLONG -DUNROLL_ONLY -DCNT=8192 -o bench-ncpe8192 benchmark-ncpe.c poly.c vec.c $(LIBCSAPP) 
	$(MULTIRUN) -c ./bench-ncpe31 > $(RESULTS)/ncpe-31-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).txt
	$(MULTIRUN) -c  ./bench-ncpe8192 > $(RESULTS)/ncpe-8192-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).txt
	$(TABULATE) -r ncpe-31:ncpe-8192 -w $(WSIZE) -d $(RESULTS) -m $(MACHINE)-O$(OPTLEVEL) > $(RESULTS)/summary-ncpe-$(MACHINE)-O$(OPTLEVEL)-$(WSIZE).txt

clean:
	rm -f $(PROGS) *.o *~ *.s *.exe