Re^5: Curses-based applications?

That was really good! It partly answers a nagging question I had after doing those speed comparisons between your Math::3Space vector calculations and doing it in PDL, which is still a bit slow in transitioning between Perl-land and C-land. One thing I know, thanks to using cachegrind, is that looking up a PDL object's pdl* spends a lot of time in hv_common, from sv_derived_from via isa_lookup, as called by the typemap-used pdl_SvPDLV.

In your opinion, would using this Perl magic scheme be quicker? And if so, does any hero feel like PR-ing (or just commenting some part-way code on https://github.com/PDLPorters/pdl/issues/451, the PDL performance issue) to make PDL-pointer-getting go quicker? (That issue has my previous research that found the hv_common stuff, and also links to the Math::3Space discussion issue)

Comment on Re^5: Curses-based applications? Select or Download Code

Replies are listed 'Best First'.
Re^6: Curses-based applications? by NERDVANA (Priest) on Jan 04, 2025 at 17:54 UTC
Yes I think there are some speedups to be had using Magic for this code, but it's dealing with an awful lot of special cases, so hard to say exactly what the best change would be. I'll leave comments on the issue since it's sort of offtopic for this thread.	[reply]