On 07/12/2012 09:26 PM, Måns Rullgård wrote:
> Daniel Kang <[email protected]> writes:
>
>> Testing gives 25-30% gain on HD clips with two threads and up to 50%
>> gain with eight threads.
>>
>> Sliced threading uses more memory than single or frame threading.
>
> Does this patch increase the memory usage for the single-threaded or
> frame threaded decoding as currently supported? If yes, I'd like to
> see/perform some benchmarks on ARM.
>
+ s->mb_layout = (avctx->active_thread_type == FF_THREAD_SLICE) &&
(FFMIN(s->num_coeff_partitions, avctx->thread_count) > 1);
+ if (s->mb_layout == 0) { // Frame threading and one thread
+ s->macroblocks_base =
av_mallocz((s->mb_width+s->mb_height*2+1)*sizeof(*s->macroblocks));
+ s->intra4x4_pred_mode_top = av_mallocz(s->mb_width*4);
+ }
The code says no, Daniel could you make it !s->mb_layout for consistency?
lu
--
Luca Barbato
Gentoo/linux
http://dev.gentoo.org/~lu_zero
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel
