[x265] [PATCH] pools: FE threads start from pool 1, since lookahead is assigned to pool 0

Steve Borho steve at borho.org
Tue Aug 4 17:58:16 CEST 2015


On 08/04, deepthi at multicorewareinc.com wrote:
> # HG changeset patch
> # User Deepthi Nandakumar <deepthi at multicorewareinc.com>
> # Date 1438650477 -19800
> #      Tue Aug 04 06:37:57 2015 +0530
> # Node ID 38f5c6c98ebf2ae934bb512df39120dbfec14401
> # Parent  d5278c76d341b3bac405938dbfb64cb7e2d9bce5
> pools: FE threads start from pool 1, since lookahead is assigned to pool 0
> 
> diff -r d5278c76d341 -r 38f5c6c98ebf source/encoder/encoder.cpp
> --- a/source/encoder/encoder.cpp	Mon Aug 03 10:18:46 2015 -0500
> +++ b/source/encoder/encoder.cpp	Tue Aug 04 06:37:57 2015 +0530
> @@ -172,7 +172,7 @@
>      {
>          for (int i = 0; i < m_param->frameNumThreads; i++)
>          {
> -            int pool = i % m_numPools;
> +            int pool = (i + 1) % m_numPools;

I'm not sure if this is a net gain or not; since most encoder memory
ends up allocated on the first socket by default. Did you see any
performance difference from this?

>              m_frameEncoder[i]->m_pool = &m_threadPool[pool];
>              m_frameEncoder[i]->m_jpId = m_threadPool[pool].m_numProviders++;
>              m_threadPool[pool].m_jpTable[m_frameEncoder[i]->m_jpId] = m_frameEncoder[i];
> _______________________________________________
> x265-devel mailing list
> x265-devel at videolan.org
> https://mailman.videolan.org/listinfo/x265-devel

-- 
Steve Borho


More information about the x265-devel mailing list