matrixSize = int(arg)
elif opt in ("-b", "--bandwidth"):
bandwidth = int(arg)
elif opt in ("-p", "--partitions"):
partitionNumber = int(arg)
elif opt in ("-r", "--runs"):
runs = int(arg)
config = {
'matrixSize': matrixSize,
'bandwidth': bandwidth,
'partitionNumber': partitionNumber,
}
# create Matrices
A = sparse_creator.create_banded_matrix(matrixSize, bandwidth / 2, bandwidth / 2)
#x = numpy.ones(matrixSize)
#x = numpy.random.rand(matrixSize)
#b = scipy.sparse.vstack(sparse_creator.create_rhs(A, x))
x_hat = np.ones(matrixSize, dtype=np.float32)
b = sp.sparse.vstack(sparse_creator.create_rhs(A, x_hat))
x_primes = []
bench = []
for i in xrange(runs):
x_prime, t = spike.spike(A, b, config, False)
bench.append(t)
x_primes.append(norm(x_hat - x_prime))
sys.stdout.write('.')
sys.stdout.flush()
res = fun(bench)
from python_project.utils import sparse_creator
from python_project.solver import LapackBenchmark
import numpy as np
import scipy as sp
from time import time
from numpy.linalg import norm
# set basic values
runs = 20
fun = min
# create Matrices
for bw in [2,16,32,128]:
for n in [2048,4096,8192]:
A = sparse_creator.create_banded_matrix(n, bw/2, bw/2)
x_hat = np.ones(n, dtype=np.float32)
b = sp.sparse.vstack(sparse_creator.create_rhs(A, x_hat))
for p in [2,4,8,16,32,64,128,256]:
if n/p <= bw:
continue
config = {
'matrixSize': n,
'bandwidth': bw,
'partitionNumber': p,
}
bench = []
