On Fri, Nov 27, 2015 at 4:33 PM, Rostislav Pehlivanov
wrote:
> On Fri, 2015-11-27 at 15:59 -0500, Ganesh Ajjanagadde wrote:
>> I get build failures starting with commit
>> 3d62e7a30fa552be52d12b31e3e0f79153aff891 under
>> --enable-hardcoded-tables.
>> @Rostislav: can you check, reproduce, and reso
On Fri, 2015-11-27 at 15:59 -0500, Ganesh Ajjanagadde wrote:
> I get build failures starting with commit
> 3d62e7a30fa552be52d12b31e3e0f79153aff891 under
> --enable-hardcoded-tables.
> @Rostislav: can you check, reproduce, and resolve this?
Fixed, thanks for reporting it quickly. Problem was
that
On Fri, Nov 27, 2015 at 1:00 PM, Rostislav Pehlivanov
wrote:
> On Fri, 2015-11-27 at 18:50 +0100, Hendrik Leppkes wrote:
>> Note that the init is wrong for float as well. You use the same
>> condition for aacenc and aacdec, but they init different things.
>> aacdec inits much more, and if aacenc w
On Fri, 2015-11-27 at 18:50 +0100, Hendrik Leppkes wrote:
> Note that the init is wrong for float as well. You use the same
> condition for aacenc and aacdec, but they init different things.
> aacdec inits much more, and if aacenc would init first for some
> reason, it would leave aacdec without ha
On Fri, Nov 27, 2015 at 6:40 PM, Rostislav Pehlivanov
wrote:
> On Fri, 2015-11-27 at 09:08 -0800, Timothy Gu wrote:
>> On Fri, Nov 27, 2015 at 03:08:05PM +, Rostislav Pehlivanov wrote:
>> > I've just pushed a patch which makes the AAC encoder threadsafe, so
>> > now
>> > it should be safe to a
On Fri, 2015-11-27 at 09:08 -0800, Timothy Gu wrote:
> On Fri, Nov 27, 2015 at 03:08:05PM +, Rostislav Pehlivanov wrote:
> > I've just pushed a patch which makes the AAC encoder threadsafe, so
> > now
> > it should be safe to always generate that table at runtime.
>
> This commit seems to brea
On Fri, 27 Nov 2015 09:08:38 -0800
Timothy Gu wrote:
> On Fri, Nov 27, 2015 at 03:08:05PM +, Rostislav Pehlivanov wrote:
> > I've just pushed a patch which makes the AAC encoder threadsafe, so now
> > it should be safe to always generate that table at runtime.
>
> This commit seems to brea
On Fri, Nov 27, 2015 at 12:08 PM, Timothy Gu wrote:
> On Fri, Nov 27, 2015 at 03:08:05PM +, Rostislav Pehlivanov wrote:
>> I've just pushed a patch which makes the AAC encoder threadsafe, so now
>> it should be safe to always generate that table at runtime.
>
> This commit seems to break FATE
On Fri, Nov 27, 2015 at 03:08:05PM +, Rostislav Pehlivanov wrote:
> I've just pushed a patch which makes the AAC encoder threadsafe, so now
> it should be safe to always generate that table at runtime.
This commit seems to break FATE on a couple of platforms (not sure about
Linux yet but proba
On Fri, 2015-11-27 at 09:16 -0500, Ganesh Ajjanagadde wrote:
> Ok, good. Let us proceed with this one step at a time. Here, I guess
> it really does not matter since the net size is ~ 3 kB. Nevertheless,
> it seems unnecessary now to do it at compile time, so unless AAC
> maintainers object, I will
On Fri, Nov 27, 2015 at 8:10 AM, wm4 wrote:
> On Fri, 27 Nov 2015 13:51:57 +0100
> Hendrik Leppkes wrote:
>
>> On Fri, Nov 27, 2015 at 1:48 PM, Ganesh Ajjanagadde wrote:
>> > On Fri, Nov 27, 2015 at 7:05 AM, wm4 wrote:
>> >> On Fri, 27 Nov 2015 06:42:21 -0500
>> >> Ganesh Ajjanagadde wrote:
>>
On Fri, Nov 27, 2015 at 2:10 PM, wm4 wrote:
> On Fri, 27 Nov 2015 13:51:57 +0100
> Hendrik Leppkes wrote:
>
>> On Fri, Nov 27, 2015 at 1:48 PM, Ganesh Ajjanagadde wrote:
>> > On Fri, Nov 27, 2015 at 7:05 AM, wm4 wrote:
>> >> On Fri, 27 Nov 2015 06:42:21 -0500
>> >> Ganesh Ajjanagadde wrote:
>>
On Fri, 27 Nov 2015 13:51:57 +0100
Hendrik Leppkes wrote:
> On Fri, Nov 27, 2015 at 1:48 PM, Ganesh Ajjanagadde wrote:
> > On Fri, Nov 27, 2015 at 7:05 AM, wm4 wrote:
> >> On Fri, 27 Nov 2015 06:42:21 -0500
> >> Ganesh Ajjanagadde wrote:
> >>
> >>> On Fri, Nov 27, 2015 at 5:35 AM, Rostisla
Le septidi 7 frimaire, an CCXXIV, Ganesh Ajjanagadde a écrit :
> However, for this to be valid, at least some basic tests need to be
> done, e.g what is the library size before and after
> --enable-hardcoded-tables?
That is not terribly difficult to do:
-rwxr-xr-x 1 cigaes cigaes 15732448 Nov 27
On Fri, Nov 27, 2015 at 1:48 PM, Ganesh Ajjanagadde wrote:
> On Fri, Nov 27, 2015 at 7:05 AM, wm4 wrote:
>> On Fri, 27 Nov 2015 06:42:21 -0500
>> Ganesh Ajjanagadde wrote:
>>
>>> On Fri, Nov 27, 2015 at 5:35 AM, Rostislav Pehlivanov
>>> wrote:
>>> > LGTM, but could you leave (just comment it ou
On Fri, Nov 27, 2015 at 7:05 AM, wm4 wrote:
> On Fri, 27 Nov 2015 06:42:21 -0500
> Ganesh Ajjanagadde wrote:
>
>> On Fri, Nov 27, 2015 at 5:35 AM, Rostislav Pehlivanov
>> wrote:
>> > LGTM, but could you leave (just comment it out) the old code in there
>> > so it's a little easier to follow?
>>
On Fri, 27 Nov 2015 06:42:21 -0500
Ganesh Ajjanagadde wrote:
> On Fri, Nov 27, 2015 at 5:35 AM, Rostislav Pehlivanov
> wrote:
> > LGTM, but could you leave (just comment it out) the old code in there
> > so it's a little easier to follow?
> >> //ff_aac_pow2sf_tab[i] = pow(2, (i - POW_S
On Fri, Nov 27, 2015 at 5:35 AM, Rostislav Pehlivanov
wrote:
> LGTM, but could you leave (just comment it out) the old code in there
> so it's a little easier to follow?
>> //ff_aac_pow2sf_tab[i] = pow(2, (i - POW_SF2_ZERO) / 4.0);
>> //ff_aac_pow34sf_tab[i] = pow(ff_aac_pow2sf_tab
LGTM, but could you leave (just comment it out) the old code in there
so it's a little easier to follow?
> //ff_aac_pow2sf_tab[i] = pow(2, (i - POW_SF2_ZERO) / 4.0);
> //ff_aac_pow34sf_tab[i] = pow(ff_aac_pow2sf_tab[i], 3.0/4.0);
The accuracy increase is always nice.
On Thu, 2015-
On Thu, Nov 26, 2015 at 4:31 PM, Ganesh Ajjanagadde
wrote:
> This speeds up aac_tablegen to a ludicruous degree (~97%), i.e to the point
> where it can be argued that runtime initialization can always be done instead
> of
> hard-coded tables. The only cost is essentially a trivial increase in
> t
20 matches
Mail list logo