[x264-devel] Yet another updated threaded slicetype patch (v14)

[Jason Garrett-Glaser]

On Wed, Jul 1, 2009 at 8:49 AM, Guillaume POIRIER<poirierg at gmail.com> wrote:
> Hello,
>
> On Wed, Jul 1, 2009 at 4:56 PM, Mike Kazmier <DaKaZ at zenbe.com> wrote:
>>
>> On Fri, Jun 5, 2009 at 10:17 PM "Steven Walters" <kemuri9 at gmail.com>
>> wrote:
>> >yeah it's no problem to help out, would like to see the patch get
>> > finalized and put to use for everyone...
>>
>> All - with a big HOORAY to Steven for his effort in fixing the v14 patch
>> for proper functionality - we are proud to submit this patch which has been
>> tested on linux and windows and appears to be functioning properly.  We
>> would love to get this committed.
>>
>> Thanks
>
> I had a quick look at it, and these lines hurt my eyes:
>
> +#ifndef USE_USLEEP
> +            x264_pthread_cond_wait( &h->lookahead->ofbuf.cv_empty,
> &h->lookahead->ofbuf.mutex );
>
> +#else
> +            x264_pthread_mutex_unlock( &h->lookahead->ofbuf.mutex );
> +            usleep( 100 );
> +            x264_pthread_mutex_lock( &h->lookahead->ofbuf.mutex );
> +#endif
>
>
> You shouldn't duplicate these lines over and over again. You can probably
> fix that problem by putting these changes directly in x264_pthread_cond_wait
> (but I'm not certain).

And furthermore, if we don't know why usleep is faster, we can't
commit anything using it, because we don't subscribe to cargo cult
programming.

Dark Shikari