On Sat, Jul 19, 2014 at 09:26:48AM +0200, Christophe Gisquet wrote:
> Hi,
>
> 2014-07-10 19:35 GMT+02:00 Michael Niedermayer :
> > AV_COPY, AV_SWAP, AV_ZERO
> > AV_[RW]N[8-64]A need aligned memory
> >
> > AV_[RW][BLN][8-64]
> > AV_COPY*U doesnt need aligned memory, but might be faster if its
> > a
Hi,
2014-07-10 19:35 GMT+02:00 Michael Niedermayer :
> AV_COPY, AV_SWAP, AV_ZERO
> AV_[RW]N[8-64]A need aligned memory
>
> AV_[RW][BLN][8-64]
> AV_COPY*U doesnt need aligned memory, but might be faster if its
> aligned.
> These might be slower than the "aligned only" variants
Documentation sent i
On Thu, Jul 10, 2014 at 04:08:21PM +0200, Christophe Gisquet wrote:
> 2014-07-10 14:53 GMT+02:00 Michael Niedermayer :
> > On Thu, Jul 10, 2014 at 06:47:54AM +, Christophe Gisquet wrote:
> >> When dealing with MVs, both components may be processed at a time.
> >
> > mixing memory access sizes c
2014-07-10 14:53 GMT+02:00 Michael Niedermayer :
> On Thu, Jul 10, 2014 at 06:47:54AM +, Christophe Gisquet wrote:
>> When dealing with MVs, both components may be processed at a time.
>
> mixing memory access sizes can lead to speed penalties on some cpus
> when such mixed accesses are close t
Hi,
On Thu, Jul 10, 2014 at 2:47 AM, Christophe Gisquet <
christophe.gisq...@gmail.com> wrote:
> -mergecandlist[nb_merge_cand].mv[0].x = l0_cand.mv
> [0].x;
> -mergecandlist[nb_merge_cand].mv[0].y = l0_cand.mv
> [0].y;
> -mergecandlist[nb_
On Thu, Jul 10, 2014 at 06:47:54AM +, Christophe Gisquet wrote:
> When dealing with MVs, both components may be processed at a time.
mixing memory access sizes can lead to speed penalties on some cpus
when such mixed accesses are close together
did you benchmark this ?
[...]
--
Michael
When dealing with MVs, both components may be processed at a time.
---
libavcodec/hevc.c | 3 +--
libavcodec/hevc_mvs.c | 23 +--
2 files changed, 10 insertions(+), 16 deletions(-)
diff --git a/libavcodec/hevc.c b/libavcodec/hevc.c
index 4d3513a..f2bbe91 100644
--- a/liba
