You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 

115 lines
4.2 KiB

import os
import os.path
import subprocess
import re
import time
import math
import argparse
STORM_PATH= "/Users/mvolk/develop/storm/build/src/storm-dft"
EXAMPLE_DIR= "/Users/mvolk/develop/storm/examples/dft/"
benchmarks = [
("and", False, [3, 1]),
("and_param", True, ["(4*x^2+2*x+1)/((x) * (2*x+1))", "1"]),
("cm2", False, [0.256272, 1]),
#("cm4", False, [0, 1]),
("cps", False, ["inf", 0.333333]),
#("fdep", False, [0, 1]),
("mdcs", False, [2.85414, 1]),
("mdcs2", False, [2.85414, 1]),
("mp", False, [1.66667, 1]),
("or", False, [1, 1]),
("pand", False, ["inf", 0.666667]),
("pand_param", True, ["-1", "(x)/(y+x)"]),
("spare", False, [3.53846, 1]),
("spare2", False, [1.86957, 1]),
("spare3", False, [1.27273, 1]),
("spare4", False, [4.8459, 1]),
("spare5", False, [2.66667, 1]), # We discard the result 2.16667 from DFTCalc
("spare6", False, [1.4, 1]),
("spare7", False, [3.67333, 1]),
("tripple_and1", False, [4.16667, 1]),
("tripple_and2", False, [3.66667, 1]),
("tripple_and2_c", False, [3.6667, 1]),
("tripple_and_c", False, [4.16667, 1]),
("tripple_or", False, [0.5, 1]),
("tripple_or2", False, [0.666667, 1]),
("tripple_or2_c", False, [0.66667, 1]),
("tripple_or_c", False, [0.5, 1]),
("tripple_pand", False, ["inf", 0.0416667]),
("tripple_pand2", False, ["inf", 0.166667]),
("tripple_pand2_c", False, ["inf", 0.166667]),
("tripple_pand_c", False, ["inf", 0.0416667]),
("voting", False, [1.66667, 1]),
("voting2", False, [0.588235, 1])
]
def run_storm_dft(filename, prop, parametric, quiet):
# Run storm-dft on filename and return result
dft_file = os.path.join(EXAMPLE_DIR, filename + ".dft")
args = [STORM_PATH,
dft_file,
'--prop', prop]
if parametric:
args.append('--parametric')
output = run_tool(args, quiet)
# Get result
match = re.search(r'Result: \[(.*)\]', output)
if not match:
print("No valid result found in: " + output)
return
result = match.group(1)
return result
def run_tool(args, quiet=False):
"""
Executes a process,
:returns: the `stdout`
"""
pipe = subprocess.Popen(args, stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
result = "";
for line in iter(pipe.stdout.readline, ""):
if not line and pipe.poll() is not None:
break
output = line.decode(encoding='UTF-8').rstrip()
if output != "":
if not quiet:
print("\t * " + output)
result = output
return result
if __name__ == "__main__":
parser = argparse.ArgumentParser(description='Benchmarking DFTs via Storm')
parser.add_argument('--debuglevel', type=int, default=0, help='the debug level (0=silent, 1=print benchmarks, 2=print output from storm')
args = parser.parse_args()
count = 0
correct = 0
properties = ["ET=? [F \"failed\"]", "P=? [F \"failed\"]"]
start = time.time()
for index, prop in enumerate(properties):
for (benchmark, parametric, result_original) in benchmarks:
expected_result = result_original[index]
# Run benchmark and check result
count += 1;
if args.debuglevel > 0:
print("Running '{}' with property '{}'".format(benchmark, prop))
result = run_storm_dft(benchmark, prop, parametric, args.debuglevel<2)
if not parametric:
# Float
result = float(result)
if not math.isclose(result, float(expected_result), rel_tol=1e-05):
print("Wrong result on example '{}' with property '{}': result: {}, Expected: {}".format(benchmark, prop, result, expected_result))
else:
correct += 1
else:
# Parametric
if result != expected_result:
print("Wrong result on example '{}' with property '{}': result: {}, Expected: {}".format(benchmark, prop, result, expected_result))
else:
correct += 1
end = time.time()
print("Correct results for {} of {} DFT checks in {}s".format(correct, count, end-start))