intfiction.org

The Interactive Fiction Community Forum
It is currently Wed Oct 24, 2018 1:16 am

All times are UTC - 6 hours [ DST ]


Search found 14 matches
Search these results:

Author Message

 Forum: Inform 6 and 7 Development   Topic: Inform 7's (lack of) Speed

Posted: Wed Dec 21, 2011 8:21 am 

Replies: 48
Views: 8366


So sorry I missed this. Afraid it came while I was on holiday. As of now, I've just wrapped the concealed possessions stuff into a larger (private) extension which helps me to move titles onto the Kindle. You're welcome to use my concealed possessions code in an extension of your own, if you like.

 Forum: Inform 6 and 7 Development   Topic: Inform 7's (lack of) Speed

Posted: Tue Oct 18, 2011 2:15 pm 

Replies: 48
Views: 8366


I'm interested, of course. I added the opcode calls partly for your benefit. Let me know how helpful they are, if any. I think for the typical turn in a typical game, the parser comprises most of the processing that a game needs to do, so speeding it up is fruitful for everyone. I've been experimen...

 Forum: Inform 6 and 7 Development   Topic: Inform 7's (lack of) Speed

Posted: Tue Oct 18, 2011 12:59 pm 

Replies: 48
Views: 8366


Quote:
I'd be interested in adding that code to the Scope Caching extension.


Really sorry to be so long in getting this to you. I got kind of pulled away for a while there. ("Squirreled," as my wife and I say in reference to our favorite character from Up.)

 Forum: Inform 6 and 7 Development   Topic: Inform 7's (lack of) Speed

Posted: Thu Oct 06, 2011 12:44 pm 

Replies: 48
Views: 8366


Give me a few days to clean it up a bit more, and I'll get it to you.

 Forum: Inform 6 and 7 Development   Topic: Inform 7's (lack of) Speed

Posted: Thu Oct 06, 2011 11:32 am 

Replies: 48
Views: 8366


You're right that it's once per scope test per object. However, scope tests are actually done for every noun in every grammar syntax line, over and over again, via ScopeWithin and DoScopeActionAndRecurse. Thus for a command like PUT in this game, which has 17 syntax lines, it's potentially done many...

 Forum: Inform 6 and 7 Development   Topic: Inform 7's (lack of) Speed

Posted: Thu Oct 06, 2011 9:36 am 

Replies: 48
Views: 8366


So, I feel like I've mostly solved these problems. I just thought I'd report back here in case there is any interest in what I found. Basically, getting into an acceptable window of performance involved four strategies. 1. A new accelerated opcode for the BlkSize function. 2. Eliminated calls to Whe...

 Forum: Inform 6 and 7 Development   Topic: Inform 7's (lack of) Speed

Posted: Fri Sep 30, 2011 9:59 am 

Replies: 48
Views: 8366


I do have some "After deciding the scope of" rules. I tried commenting them out, but this makes no real difference in parsing time. Obviously running those rules will lead to slight delays, but they appear to be very slight indeed. I don't believe I have any understand by relation rules. T...

 Forum: Inform 6 and 7 Development   Topic: Inform 7's (lack of) Speed

Posted: Fri Sep 30, 2011 3:15 am 

Replies: 48
Views: 8366


So, I've implemented BlkSize and WhetherProvides as accelerated functions, with some modest improvements. Along the way, though, I've learned something I find pretty interesting. The slowest command by far is "put." Something like "put hat in chest" could take upwards of 17 secon...

 Forum: Inform 6 and 7 Development   Topic: Inform 7's (lack of) Speed

Posted: Tue Sep 27, 2011 6:35 am 

Replies: 48
Views: 8366


Thanks for the suggestions! I'm afraid that the scope algorithms don't seem to be the source of the problem. Implementing Scope Caching yields no appreciable improvement. So, I took Zarf's advice and made a profiling build of Glulxe. It turns out the most expensive function is ProcessRulebook. No su...

 Forum: Inform 6 and 7 Development   Topic: Inform 7's (lack of) Speed

 Post subject: Inform 7's (lack of) Speed
Posted: Mon Sep 26, 2011 10:27 am 

Replies: 48
Views: 8366


So, I'm currently working with a partner to bring the second work of modern IF to the Kindle, and we're having a problem with performance. In itself this is nothing new; I had to hack the accelerated opcodes into King of Shreds and Patches and spend hours optimizing the Java code at the interpreter ...
Sort by:  
Page 1 of 2 [ Search found 14 matches ]


All times are UTC - 6 hours [ DST ]


Jump to:  
Powered by phpBB® Forum Software © phpBB Group