what is happening is that there is special code for m&-. (same timing as no binding), but not for -. &n.
You can tell by the different memory footprints. It kind of makes sense that -. needs to bind to m first, since that is what will be taken away from. So the code has to unpack n from -.&n before applying it to both. Of course it would be nice if it were faster. ----- Original Message ----- From: Brian Schott <[email protected]> To: "[email protected]" <[email protected]> Cc: Sent: Sunday, October 13, 2013 12:52:50 PM Subject: Re: [Jprogramming] Odd timings for dyadic not (-.) 10 ts 'm -. n' 0.0899539 5.24346e6 g=:m&-. 10 ts 'g n' 0.0908445 5.24365e6 Is the above relevant? Used iPad, pre OS7. --- (B=) On Oct 13, 2013, at 12:07 AM, Ganesh Rapolu <[email protected]> wrote: > m =. 1e6 ?@$ 1e6 > n =. 50000 ?@$ 1e6 > f =. -.&n > 10 ts 'm -. n' > 0.0261088 9.43834e6 > 10 ts 'f m' > 0.115288 8.39014e6 > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
