[FFmpeg-devel] [PATCH] HE-AAC v1 decoder try 4

[Vitor Sessak]

Alex Converse wrote:
> On Wed, Feb 17, 2010 at 1:45 PM, Vitor Sessak <vitor1001 at gmail.com> wrote:
>> Alex Converse wrote:
>>> On Tue, Feb 16, 2010 at 6:01 PM, Alex Converse
>>> <alex.converse at gmail.com>wrote:
>> My final comments:
>>
>>>> Notes:
>>>> *All the computation time is spent in ff_sbr_apply() and it's
>>>> children. If it isn't called from ff_sbr_apply() making it 100% faster
>>>> isn't going to buy us anything.
>>>> *Right now the synthesis filterbank is written on top on an MDCT. With
>>>> appropriate SIMD functions it may make sense to move it to an FFT.
>>>> Right now the MDCT version is much faster.
>> Since you have already done it, can you send it so we can have a look?
>>
>> Also if you still have the butterflies without the scratch buffer, I'd like
>> also to have a look. I'd like to understand why it is slower.
>>
> 
> I already chased down your cosine tables red herring and this one. If
> you want to make a very fast scalar filter bank be my guest. I can
> send you the relevant pages.

I wasn't asking you to retry it, I've already said I have no further 
comments and I know you lost quite some time already. I just thought you 
still had the code somewhere in your HD and it wouldn't take much time 
to copy and paste a few code lines here (could be useful the day someone 
decides to SIMDfy this transform). Same thing about the FFT, from your 
message you give the impression the code is already written and I think 
it would be useful to have it in the thread (and I'm curious to what it 
looks like).

-Vitor