[prev in list] [next in list] [prev in thread] [next in thread] 

List:       bochs-dev
Subject:    Re: will add bochs updates and patches tomorrow
From:       Roger Fujii <rmf () unixbox ! lookhere ! com>
Date:       1999-02-16 20:36:53
[Download RAW message or body]

Kevin P Lawton wrote:

> Also, I could use some feedback data on the following.
> It certainly is not a good idea to
> translate every piece of new code as it is encountered.  If
> the code path is just "passing thru the neighborhood" once,
> might as well just execute it in the interpreter.  After
> visiting the code N times, it's a better candidate for translation.
> So, I need to figure out what 'N' is?  Or at least a reasonable
> N that satisfies most cases.  Of course this will be a tweakable
> setting in bochs, but I need a default. 

I would presume that a ratio between compilation + execution time
to interpreted time would determine N.  If your compilation is really
slow and interpreted speed is pretty fast, then N should be large.

I'm also not too sure if 'N' should be a fixed #.  One would
think a small tight loop (ala memset) should be done quickly,
as opposed to a large loop...

-r
-- 
Roger Fujii <rmf@lookhere.com>     Phone: (703)280-1243
Underemployed, and trying to keep it that way....


-- 
         To unsubscribe: mail bochs-developers-request@redhat.com with 
                       "unsubscribe" as the Subject.

[prev in list] [next in list] [prev in thread] [next in thread]