[MPlayer-dev-eng] Re: Re: fast SPP

[Nikolaj]

On Sun, 24 Apr 2005 16:37:36 +0200, Jindrich Makovicka 
<makovick at kmlinux.fjfi.cvut.cz> wrote:

> Jindrich Makovicka wrote:
>> Nikolaj wrote:
>>
>>> Hello,
>>>
>>> Sorry for long delay. Included proposed patch to vf_spp. Please review.
>>> I know there is still some issues - maybe x86_64 compatibility, or smth
>>> else. I plan to correct this file later. And I need your opinions about
>>> what to do.
>>>
>>> PS. diff is quite strange looking - if you need, I can send just a full
>>> file.
>>
>>
>> I am very eager to test it. Unfortunately 7 of 10 hunks fail on current
>> CVS. Could you please provide a patch, which applies cleanly?
>
> are you sure you sent a right patch? this one applies when using -l, but
> has no chance to compile:
Very big sorry, but I was a bit tired of it today, will correct tomorrow 
or so.
I even hav'nt tried to compile it (but it should be correct 
_theoretically_ :) ). But I was thinking this form will be sufficent for 
review, so I decided to post it today.

>
> missing parameter for p:
>
> +static void mul_thrmat_mmx(int q)
> +{int adr=(int)(&p->threshold_mtx_noq[0]);
> +  asm volatile(
Yeah, will pass context.

> BLOCKSZ is undefined
#define BLOCKSZ 12
:)

> etc etc.

BTW, forgot to say (2 all):
webcvs at mplayerhq isn't functioning properly - I wasn't able to get 
current vf_spp in the plain text format.

filter has 3 arguments (level, constant quant, mode) - I changed last one 
into softness bias, >0 - blurrier.

-- 
Best regards,
    Nikolaj                          mailto:nialpof at pisem.net