Message70522
Timings on x86/Linux are similar: the lsum-based version is around
10% slower on average, 25% slower in the worst case, and significantly
faster for the msum worst cases.
There's probably still some snot left to optimize out, though. Some
tempting ideas are:
(1) to try using doubles instead of longs for the accumulator digits
(with 51 or 52 bits of precision), and
(2) to split each mantissa into (nearest_integer, fraction) instead
of (next_smallest_integer, fraction), using rint or lrint.
Anything else? |
|
Date |
User |
Action |
Args |
2008-07-31 15:55:23 | mark.dickinson | set | recipients:
+ mark.dickinson, tim.peters, rhettinger, MrJean1 |
2008-07-31 15:55:23 | mark.dickinson | set | messageid: <1217519723.22.0.97208848367.issue2819@psf.upfronthosting.co.za> |
2008-07-31 15:55:22 | mark.dickinson | link | issue2819 messages |
2008-07-31 15:55:22 | mark.dickinson | create | |
|