21 Jul
2016
21 Jul
'16
2:58 p.m.
On 20.07.2016 11:52, Philipp Singer wrote:
While in general 2.16 is much faster on most operations, it is slower on some standard vertex access operations. For example:
def test(): g = gt.price_network(10000, 10, directed=False) for i in range(10000): rnd = g.vertex(randint(0, g.num_vertices())) %time test()
2.22: CPU times: user 189 ms, sys: 7.15 ms, total: 197 ms Wall time: 196 ms
2.16 CPU times: user 841 ms, sys: 776 µs, total: 841 ms Wall time: 838 ms
Do you have an idea why this might be the case?
This looks like a performance regression. Please just open an issue in the website, so I can take care of it. Best, Tiago -- Tiago de Paula Peixoto <tiago@skewed.de>