Windows Glulxe performance issues?

One of the testers on my WIP was experiencing serious delay on each turn, around one second, on Windows Glulx. I haven’t seen this on any Mac interpreter (minor slowness but nothing major), and it hasn’t happened on windows Gargoyle. Has anyone seen similar issue with Win Glulxe?

I’m planning to do some performance tuning anyway, but I was wondering if anyone had any experience with this terp in particular.

One thing my game has a substantial amount of is complicated grammar tokens like [any available quip] where available relies on a bunch of rules. I wonder if that would be particularly problematic for Windows Glulxe.

Nothing WinGlulxe specific for me, though I’ve seen general slowdowns. When I run a Linux VM, I see a bit more slowdown, because it has less RAM it can use.

So maybe the tester’s computer doesn’t have a lot of RAM? Or their computer is using it for something else? It’s quite possible if someone leaves firefox open, that gives WinGlulxe less resources.

Still it’s worthwhile to see what can be sped up, of course.

I don’t have a ton of data, but I can verify that running scripts on a big game with lots of rooms/things/commands takes more time per move than ones with fewer rooms and fewer commands. I can’t say for sure how much rooms, things and commands contribute to that.

I suppose we could create glulxe files with, say, 200 rooms only, 200 things only, etc., and take performance metrics. But it’s work!

I know regexes in i7, or even using “character number 1/word number 1,” can slow something down a lot. I was just working on my anagram games and switched them to i6, and I knocked 40% off the time for some of my tests.

Probably the big one, though, is if you refer to something as “visible.” Once I changed that to “touchable” I saw runtimes drop up to 80% in the anagram games.

This probably doesn’t answer your question, but I hope my experiences help eliminate a few possibilities that couldn’t be wrong, or maybe jog your memory for other things that might cause a slowdown.


Windows Git should be at least somewhat faster. They’ll have to try it to say whether it’s fast enough though.

1 Like

On the other hand, Lectrote may be somewhat slower.

1 Like

Aha! I found it. It’s some old code I wrote for sound propagation when I was first starting out with Inform 7. I haven’t analyzed it too closely, but I assume I have too many path finding operations in there.