26 Jan
2013
26 Jan
'13
1:02 p.m.
On 1 July 2012 00:41, Robert Smith quad@symbo1ics.com wrote:
With the code above, I can compute the same binomial coefficient in about 7.503 seconds, so about a 2.5 second speed up.
I'd like to submit a patch using the code above, but I could see how these speed ups are both implementation-dependent and platform-dependent. A particular unfortunate consequence is that we don't have those tweakable limits which (in theory; as the comments say) could be computed at compile time.
Any thoughts?
(Gears of Alexandria grind slow.)
Please do send in a patch.
If this turns out to be a performance regression for other compilers, we can worry about it then.
Cheers,
-- Nikodemus