[FFmpeg-devel] [PATCH] up VOFW in swscale?

[Michael Niedermayer]

On Sat, Apr 04, 2009 at 10:01:54AM +0200, Andreas ?man wrote:
> M?ns Rullg?rd wrote:
>> Andreas ?man <andreas at lonelycoder.com> writes:
>>> Michael Niedermayer wrote:
>>>> On Wed, Apr 01, 2009 at 11:59:46PM +0200, Andreas ?man wrote:
>>>>> Michael Niedermayer wrote:
>>>>>> On Wed, Apr 01, 2009 at 09:28:46PM +0200, Andreas ?man wrote:
>>>>>>> Hi
>>>>>>>
>>>>>>> I use the scaler to rescale photos (>2048 pixels wide) in my
>>>>>>> application.
>>>>>>> With the old scaler this was not a problem but nowadays it
>>>>>>> no longer works since the maximum width is limited to 2048 pix.
>>>>>>>
>>>>>>> Therefore I propose we up this to 4096.
>>>>>> can post some benchmarks with a CPU with small cache and a low 
>>>>>> resolution
>>>>>> video?
>>>>> Interesting exercise. Seems that increasing it to 4096 causes more
>>>>> cache aliasing. However, increasing it to 5120 makes it faster than
>>>>> the original code.
>>>> looks  good so far but i still would like to see benchmarks from one
>>>> or 2
>>>> other cpus, at least one (non P4) intel
>>> This is on a:
>>> Intel(R) Core(TM)2 CPU          6600  @ 2.40GHz
>>>
>>> I wouldn't say it's much of a difference on that one.
>>> I don't have access to any non-x86 platforms of significance.
>> You can have a login on my ppc if you like.
>
> cpu             : 7455, altivec supported
> clock           : 866.666664MHz
> L2 cache        : 256K unified
>
> Slightly more than a 1% slowdown.
> I've tried various other offsets (also sub-pagesize) and it seems to
> yield about the same result.

it might be interresting to see the actual benchmark results for
various values of 2^x + 2^y and 2^x
it seems unlikely that 2048 would be the fastest ...

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

Let us carefully observe those good qualities wherein our enemies excel us
and endeavor to excel them, by avoiding what is faulty, and imitating what
is excellent in them. -- Plutarch
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20090505/78dd6ea5/attachment.pgp>