[tremor] stack+heap requirements for tremor

[Owen G. Emry]

Michael Smith <msmith at labyrinth.net.au> said:

> At 03:00 PM 9/18/02 -0000, you wrote:
> >Michael Smith <msmith at labyrinth.net.au> said:
> >
> >> At 04:34 PM 9/17/02 -0700, you wrote:
> >> >1MB would be an astronomical stack size in the context of an embedded 
> >system.  My default is 4k, and it's rather difficult to tell if I'm 
blowing 
> >it or not without some seriously painful debugging, so I thought I'd get a 
> >ballpark stack requirement from people who might know...
> >> 
> >> I couldn't say with any precision - but it'll be MUCH MUCH more than 4k.
> >> I'd guess you'll be looking at an order of magnitude more for stack 
usage,
> >> but that might be a little low (or a little high).
> >
> >I guess you're sure you're talking about stack and not heap, right?
> >
> >So where is all that stack going?  I don't see anything obvious besides a 
> >reasonable-looking number of function arguments and local variables, and 
in 
> >my debugging, I've never caught it more than ~15 levels deep.
> 
> Having looked again, I may have somewhat overestimated (I think the
> _encoder_ has much higher stack requirements, and I hadn't looked at
> what Tremor needed in detail, since as far as memory goes it's more
> or less the same as libvorbis).
> 
> Anyway, you get a pretty significant increase from some of the 
> alloca() calls. 

Should have mentioned those -- I patched out all but the one in lsptocurve 
because it seemed like it was used heavily during decode.  The others I 
replaced with malloc/free pairs...

Anyway, _something's_ not working, and it sure seems like stack weirdness...  
Thanks for your help.

-Owen

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'tremor-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.