[FFmpeg-devel] [PATCH] IFF: New heavy optimization of decodeplane32

[Ronald S. Bultje]

Hi,

On Mon, May 10, 2010 at 1:08 PM, Sebastian Vater
<cdgs.basty at googlemail.com> wrote:
> Ronald S. Bultje a ?crit :
>> On Mon, May 10, 2010 at 9:24 AM, Sebastian Vater
>> <cdgs.basty at googlemail.com> wrote:
>>> + ? ? ? ?dst[0] |= lut[mask++];
>>> + ? ? ? ?dst[1] |= lut[mask++];
>>> + ? ? ? ?dst[2] |= lut[mask++];
>>> + ? ? ? ?dst[3] |= lut[mask];
>>
>> Why |=? All callers precede this by a memset(row, 0, width<<2), which
>> seems illogical. Remove the | and the memset and you've got much
>> faster code.
>>
>> Same applies to decodeplane8() btw.
>
> Ehhm, then you will end up only being having the last plane
> decoded...all planes have to be OR'ed together...the memset is not in
> the for (plane ...) loop, but before.

Hm, right, that makes sense.

OK, let me digest this some more and then I'll apply.

Ronald