Cython并行循环prange太慢了

Question

我试图了解如何使Cython运行并行循环。下面的代码确实以大约相同的速度运行，系统监视器在任何时间点只显示一个核心负载。 Wham我做错了吗？

我正在使用Ipython Notebook来运行它。

from cython.parallel import prange, parallel, threadid

cdef long i,j
cdef long total,subtotal = 0
cdef long n = 10**10

for i in range(n):
    total += i

CPU times: user 9.02 s, sys: 0 ns, total: 9.02 s
Wall time: 9.03 s

with nogil, parallel():
    for i in prange(8):
        subtotal = 0
        for j in range(i,n,8):
            subtotal += i
        total += subtotal

CPU times: user 13.1 s, sys: 28 ms, total: 13.1 s
Wall time: 13.4 s