How much the added rules in NAL1-6 play into ONA's performance.

[Changxin Sun]

Are the new rules introduced in 0.9.0 critical?

[Patrick Hammer]

Hi!

v0.9.x is a bit slower than v0.8.x not due to added rules in NAL.h, but due to higher allowed term complexity and more recently, with v0.9.1, the removal of the FIFO structure.

But the reasoning abilities (NAL-5 support) are worth it, as it gives the system more inferential power, it can learn more general hypotheses than before, some of which in traditional FOPL-based reasoning systems had to be handcoded by humans.

With the coming releases v0.9.x will be as fast as v0.8.x again, or faster, especially when multithreading will be utilized properly.

Best regards,

Patrick

[Changxin Sun]

Hi!

I'm really interested in how to parallelize OpenNARS. ONA will choose multiple concepts to process concurrently or one task with many termlinks, or many tasklinks with many termlinks，or the other way around? 

In the case of concurrency, if both ^op1 and ^op2 can lead to goal G!,  as in Narsese: ^op1 =/> G.  ^op2 =/> G. which one should be choosed?  

Thanks,

Best regards,

Changxin Sun 😊

Patrick Hammer <pat...@gmail.com> 于2022年9月12日周一 23:59写道：

--
You received this message because you are subscribed to a topic in the Google Groups "open-nars" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/open-nars/wfcvXFGIAKo/unsubscribe.
To unsubscribe from this group and all its topics, send an email to open-nars+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/open-nars/33f7c0b6-39e2-4104-b825-1c04b5f25981n%40googlegroups.com.