[Issue 16517] topN performance very poor on certain data sets

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16517

--- Comment #3 from Jon Degenhardt <jrdemail2000-dlang at yahoo.com> ---
Apologies for the slow response. I'm not building my own LDC, just using
current release build, so I'm not quite set up to pull the patch and test it.
I'll take a look and what it takes, if I can do it easily I will.

To do a bit of comparison with your numbers - On my machine, the DMD numbers
are a bit faster than yours. Typical DMD times for the sort version on the 3
fields are 350/345/330ms, about 85% of your times. By the same token, on my
machine LDC runs are about 85% of the DMD times. Dangerous to assume the ratios
will hold up for topN, but if they do that be outstanding.

--