[M3devel] another change in hand.c

[Jay K]

Also I changed set_eq and set_ne to just use memcmp.

 

 

That certainly appears correct.

I believe I added some tests to m3-sys/m3tests.

I also changed the integrated backend to call memcmp directly.

Given the input is known to be ulongs not just bytes, this

might be a deoptimization. However hand.c isn't necessarily

compiled with optimizations.

 

 

You know, memcmp generally handles any alignment.

First it often has to handle a leading unaligned part before

proceeding to compare data in larger chunks.

Whereas if you know the alignment you can be faster.

 

 

I believe m3front inlines for small sets anyway.

 

 

 - Jay

 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://m3lists.elegosoft.com/pipermail/m3devel/attachments/20100118/71288892/attachment-0001.html>