[FFmpeg-devel] [PATCH] libavutil: Make changes in softfloat needed for fixed point aac decoder.

Nedeljko Babic Nedeljko.Babic at imgtec.com
Wed Apr 29 14:13:53 CEST 2015


Ok, I'll split this patch some more and since this takes already too long I
will eliminate any change that is not absolutely necessary for implementation
of fixed point aac decoder (I am thinking here on new recip and div functions).

Since there will be two or three new patches from this patch I'll resubmit them
as a new patch set (as oppose resending them as replay to this thread).

Regarding comments about changes to exponent, it is change that is needed for
this soft float to be usable in implementation of fixed point aac decoder with
the new functions that we are adding.

I could also change the value of ONE_BITS and adjust the code accordingly, but
this comes down to the same.

The alternative would be to change the code that is already written (and
probably to recalculate some of the tables) but that is much more work than just
changing exponent here.
Since softfloat is not used anywhere currently, I don't see how this can be
justified.

Maybe the comment in commit message should specify that exponent is changed for
softfloat to be usable in fixed aac decoder, or something similar?

-Nedeljko
________________________________________
Od: ffmpeg-devel-bounces at ffmpeg.org [ffmpeg-devel-bounces at ffmpeg.org] u ime korisnika Michael Niedermayer [michaelni at gmx.at]
Poslato: 27. april 2015 18:31
Za: FFmpeg development discussions and patches
Tema: Re: [FFmpeg-devel] [PATCH] libavutil: Make changes in softfloat   needed for fixed point aac decoder.

On Mon, Apr 27, 2015 at 03:58:44PM +0200, Nedeljko Babic wrote:
> From: Djordje Pesut <djordje.pesut at imgtec.com>
>

> Functions for sqrt and sincos are added.

these could be in a seperate patch


>
> Div function is improved.

how?
if its about adding av_recip_sf/av_div_sf_approx() then that commit
message is outdated and should be changed to match the code


>
> Some changes are made in order for code in softfloat to be usable in fixed aac
> decoder code.

which and why?


>
> This doesn't create any impact on current ffmpeg code since softfloat is
> currently not in use and this way we don't need to make much changes in
> implementation of aac fixed point decoder that uses this code.
>

> Softfloat tests are adjusted.

no, this just changed the header files, the tests where in the c
file

also the commit message should mention how exactly the Softfloat
exp/mantisa are changed, is this a bugfix, is there some offset
added, is some sign changed, ...
if you flip the sign or add a offset to some field that should be a
seperate patch ideally so it can easily be understood. Otherwise
if its mixed in with other changes the whole commit looks
confusing

[...]
--
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

The real ebay dictionary, page 1
"Used only once"    - "Some unspecified defect prevented a second use"
"In good condition" - "Can be repaird by experienced expert"
"As is" - "You wouldnt want it even if you were payed for it, if you knew ..."


More information about the ffmpeg-devel mailing list