好的算法和差的算法对同一需求的实现，性能相差很大。

随着计算机硬件性能不断提高，我感觉自己越来越忽视算法的重要性了。今天在codility网站做了一道很简单的Demo算法题，发现性能差的不是一点半点的。

由于是Demo题，可以做多次，以下是我做2次的URL：

1. http://codility.com/demo/results/demoTY95EX-RGY/ [一开始就想用最简单的方式实现功能需求，但没想到性能这么差，分数才50（100分制），时间复杂度：O(N*N)]

2. http://codility.com/demo/results/demoWK3NEZ-5AB/ [这次实现还行，至少时间复杂度：O(N)]

我也在本地机器使用以下代码做了以下测试，发现相同数据，花费时间差距非常之大。

def solution(A):
    # http://codility.com/demo/results/demoTY95EX-RGY/
    import math
    _min_diff = None
    for _inx in range(1, len(A)):
        _diff = int(math.fabs(sum(A[:_inx])-sum(A[_inx:])))
        if _min_diff is None:
            _min_diff = _diff
        elif _min_diff > _diff:
            _min_diff = _diff
    return _min_diff
         
def solution2(A):
    # http://codility.com/demo/results/demoWK3NEZ-5AB/
    _min_diff = None
    _sum = sum(A)
    _left_sum = 0
    for _inx in range(len(A)-1):
        _left_sum += A[_inx]
        _diff = _sum - _left_sum - _left_sum
        if _diff < 0:
            _diff = -_diff
        if _sum&0x01 == 0 and _diff == 0: #for even, minimal value is 0
            _min_diff = 0
            break
        elif _sum&0x01 == 1 and _diff == 1: #for odd, minimal value is 1
            _min_diff = 1
            break
        if _min_diff is None:
            _min_diff = _diff
        elif _min_diff > _diff:
            _min_diff = _diff
    return _min_diff
    

def test(A):
    import time
    _start = time.time()
    _value = solution(A)
    _end    = time.time()
    print "solution result: time=%f, value=%d" % (_end-_start, _value)
    _start = time.time()
    _value = solution2(A)
    _end    = time.time()
    print "solution2 result: time=%f, value=%d" % (_end-_start, _value)

输出结果：

>>> A = [random.randint(-1000,1000) for _ in xrange(200)]
>>> test(A)
solution result: time=0.001000, value=14
solution2 result: time=0.000000, value=14
>>> A = [random.randint(-1000,1000) for _ in xrange(2000)]
>>> test(A)
solution result: time=0.077000, value=3
solution2 result: time=0.001000, value=3
>>> A = [random.randint(-1000,1000) for _ in xrange(2000)]
>>> A = [random.randint(-1000,1000) for _ in xrange(20000)]
>>> test(A)
solution result: time=5.497000, value=24
solution2 result: time=0.007000, value=24
>>> A = [random.randint(-1000,1000) for _ in xrange(200000)]
>>> test(A)
solution result: time=667.812000, value=0
solution2 result: time=0.016000, value=0

A中有200,000元素的时候，算法1花费667秒，算法2才0.016秒。有点无语了，应该好好反省，今后在日常的开发中要更加重视算法（及数据结构）的选择。