[Tremor] Memory Usage

[Ripinder Singh]

Well the host platform is not Arm but a 32Bit DSP like TI C62X series.
The compiler and tools are not as sophisticates as those available for
ARM, etc.

> 'alloca' does stack-based allocation. It's basically a way to (without
> changing the base language - though it requires compiler support)
> simulate variable size arrays (C99 has actual variable-size arrays,
> but alloca() is much older).

As said, I too feel that replacing alloca with malloc is not the right
way as it may lead to framentation of memory especially if repeaded
sequence of malloc and free are made use of. To rid of this problem
however requires the compiler support for dyanamic stack allocation
which is really in hands of compiler management.

> Yeah, it'd be good to have this sort of thing. We don't currently have
> any reference samples, or compliance criteria. Would you be interested
> in helping develop some of these?

Regarding defining compliance criterion, well we can study the
compliance criterion for other such codecs(mp3, wma). Maybe a little
bit is already known. MP3 decoders must obey the rms(root mean square)
criterion

Sample by sample difference < Particular Value &
standard deviation of Error < Another value

Similarly for WMA decoder there is just the Max difference criterion
only, ie, for each output sample

(Error Difference) <  0.1 % of signal at  that point.

A similar compliance criterionmay be defined but has to be studied for
the effects on the sound quality thru extensive testing.

Rip


On Wed, 2 Feb 2005 11:41:54 +0530, KumarBraj Bhushan
<kumar.brajbhushan at ittiam.com> wrote:
> Hi Rip,
> 
> >As the host platform doesn't support alloca, I tried to fit the same with malloc
> and (subsequent) free calls.
> 
> What do you mean by "host platform"? Is it ARM Simultor?
> 
> >And is there any compliance criterion defined for tremor ??
> 
> Yes, it is very suprising that till now no effort has been made in this direction. I have come across many mails requesting for a set of reference inputs and outputs with defined compliance tolerance. But there has been no reply from Monty till now on this subject, as far as I know.
> 
> regards,
> Kumar.
> 
> _______________________________________________
> Tremor mailing list
> Tremor at xiph.org
> http://lists.xiph.org/mailman/listinfo/tremor
> 


-- 
Regards,
Ripinder Singh