Hi,
2014-07-21 11:11 GMT+02:00 Christophe Gisquet :
> The change makes sense by itself, but it doesn't seem worth it. I'd
> still like to see if Mickaƫl can provide more meaningful numbers.
No strong reason behind this, but let's drop that patch.
--
Christophe
__
Hi,
2014-07-20 18:58 GMT+02:00 Ronald S. Bultje :
> That's a fair explanation - and yes, 1% is totally worth it (imo).
For another sample with more IDRs on a 4yo linux system with 6 cores,
there is actually a decrease. The sequence or system also scales
poorly: 6 cores yields less than 3x speedup
Hi,
On Sun, Jul 20, 2014 at 12:31 PM, Christophe Gisquet <
christophe.gisq...@gmail.com> wrote:
> Hi,
>
> 2014-07-20 13:35 GMT+02:00 Ronald S. Bultje :
> > Do you get a speedup? I tinkered with some related problem in vp9 once
> and
> > I never got a significant speedup out of it...
>
> Whatever
Hi,
2014-07-20 13:35 GMT+02:00 Ronald S. Bultje :
> Do you get a speedup? I tinkered with some related problem in vp9 once and
> I never got a significant speedup out of it...
Whatever your definition of significant, I don't think the figure I have is.
It's around 1% on a dual core system on ave
Hi,
On Sun, Jul 20, 2014 at 4:21 AM, Christophe Gisquet <
christophe.gisq...@gmail.com> wrote:
> Hi,
>
> I don't really like the attached patch firstly because it's a hack,
> and secondly because it may break with other kinds of threading
> (slice/wpp).
>
> It depends on the previous patch aroun
Hi,
I don't really like the attached patch firstly because it's a hack,
and secondly because it may break with other kinds of threading
(slice/wpp).
It depends on the previous patch around tmvp.
--
Christophe
From 225ab2adceef611b4d6ff7825767a741d26395b6 Mon Sep 17 00:00:00 2001
From: Christoph