Hi everyone,

anyone know the speed of memmap on Mac, Linux and Win32?

Specifically, if I'm mapping say a file of around 256 or 512MB... but  
only reading small portions of it.

Say I might only read 10K of a 256MB file. Will memmap read the entire  
file off the disk? Or just the 10K that I read?

I need this, because I'm thinking of memmapping my ElfDataDictionary.  
And my ElfDataDictionary, can be read in a "Tree based" manner.  
Meaning there is a lot of "skipping" around, jumping past entire  
branches, jumping past entire sections.

If I can memmap my ElfDataDictionary... and the memmapping is no  
slower than if I manually read in the relevant things from hard  
disk... I can, at least in theory, load very big ElfDataDictionaries,  
dynamically in and out of RAM :)

This means I can process huge amounts of data, very fast.

So I'd like to know about memmap.

...

Also I have a plugin related question. My "ElfDataFuzzy" class is  
growing, and I have so many features in mind for it in the future... I  
am worried it will start to dominate my plugin. You see I'm trying to  
do this bio-informatics project, which is fascinating, to me.

But will it be fascinating to everyone else? Probably not! I can't see  
more than 3 or 4 people using my ElfData plugin for bio-informatics.  
But I can see hundreds using it for general string processing.

If my ElfData plugin now grows 100KB in size over what it was... is  
this a bad thing? Or should I shrug it off?

The thing is, my ElfDataFuzzy class relies HEAVILY on on the ElfData  
class, as well as the ElfDataNavigator class. So how do I move it into  
another plugin, if one plugin relies heavily on the internals of  
another plugin?

I really want fast functions... I suppose I could use RB to get a  
function pointer, right? I guess that'll be fast enough... Or would a  
proper function be faster? Remember, these functions are often pretty  
big and complex anyhow. It's not like I'm calling a minor inlinable  
function or a function which won't do loads of stuff with registers.  
It's definitely the kind of thing which needs to end up saving it's  
registers on the stack.

I guess function pointers aren't so bad.

But it's probably best I just grow the ElfDataFuzzy class for now as I  
am packed for time. I REALLY want to get this ElfDataFuzzy algorithm  
done quick!!

--
http://elfdata.com/plugin/
"String processing, done right"


_______________________________________________
Unsubscribe or switch delivery mode:
<http://www.realsoftware.com/support/listmanager/>

Search the archives:
<http://support.realsoftware.com/listarchives/lists.html>

Reply via email to