[mythtv-users] Latest thoughts on small silent frontends

[Greg Thompson]

On Mar 10, 2014, at 1:57 PM, Gary Buhrmaster <gary.buhrmaster at gmail.com> wrote:

> On Sun, Mar 9, 2014 at 1:56 PM, Stephen Worthington
> <stephen_agent at jsw.gen.nz> wrote:
> ....
>> So that is good news - the cheaper Nvidia 610 chipset is OK for 1080i,
>> so it is easy to get a silent video card for any frontend box.
> 
> It may be good news, but it is not accurate news.  The GT610 is
> not capable (by objective, measurable values) capable of Advanced
> 2X de-interlacing of high bit-rate 1080i 60Hz content in all cases
> (it might, barely, be able to do so in 50Hz countries).
> 
> I am not saying that Greg is wrong.  He may very well observe
> a 610 is "buttery smooth".  But that is not (and cannot be)
> true for all configurations or requirements or individuals.
> 
> Here are some possible reasons of the disconnect:
> 
> * Sometimes (all too often) nVidia will use the same marketing
>  name for different underlying chips.  In this particular case,
>  both the GT610 and GT610M are currently GF119 chips,
>  so it should not apply, but at sometime in the future, a GT610
>  might be something else (see the GT630 comment below)
>  (should this post be read in the future).
> 
> * Some TV's have great de-interlacers and/or video processors
>  that hide "badness".  Ignoring the 960Hz video processing
>  marketing claims by some manufactures, some TVs are
>  still very good at hiding artifacts (and the default is usually
>  to smooth the picture and remove artifacts).
> 
> * Some TV's (especially smaller ones) down-sample everything
>  to a lower resolution (sometimes 1280x720) which can hide
>  artifacts as part of their post-processing.
> 
> * Some AV receivers have great video processors that
>  are as good (or better) than the TVs they are connected to.
> 
> * Some TVs (screens) are so slow, that they hide some
>  artifacts (i.e. slow refresh rate LCD screens).
> 
> * Some individuals visual processing (brain) simply do not see
>  artifacts.  Just like some people never see 60Hz refresh
>  rate artifacts (while for others, the headaches are terrible).
>  I am sure there is more recent research, but the old papers
>  I read showed no clear correlation with anything else about
>  the person and their ability to observe the visual artifacts.
>  Some people just observed what others did not.
> 
> * Some individuals just want to believe so hard that they
>  convince themselves there is no issue.  Selective hearing.
>  Selective seeing.  The mind is an impressive piece of work.
>  It matters not whether it is your new Bugatti, or your new
>  Pioneer Elite Kuro Pro.  One wants (needs) to believe it was
>  a good choice.
> 
> The "objective" qvdpautest results for a GT610 show a a value of
>    MIXER TEMPORAL_SPATIAL + IVTC (1920x1080): 51 fields/s
> which is less than the required 60 (plus some extra for
> headroom) for true minimization of visual artifacts due to
> de-interlacing for all cases.
> 
> However, that all said, subjective results may be what counts
> (to you).  Unless you have the exact same FE, TV, AV receiver,
> and visual processing, your experience *will* vary from any
> other individual.  It may be better, or worse.
> 
> Caveat emptor.
> 
> 
> 
> btw, at this time, the lowest end current generation nVidia card
> that has objective numbers that seem adequate for true Adv 2X
> is the gen 2 GT630 cards.  The gen 2 is important, with a GK208
> chipset.  Some manufactures have a passively cooled low profile
> capable version in the market.  [For the record, there was a OEM
> GT630 with a (sort of ) gen 1.5 chipset.  I do not recall ever seeing
> results from that version of the GT630.]
> _______________________________________________
> mythtv-users mailing list
> mythtv-users at mythtv.org
> http://www.mythtv.org/mailman/listinfo/mythtv-users
> http://wiki.mythtv.org/Mailing_List_etiquette
> MythTV Forums: https://forum.mythtv.org

Hi Gary,

I ran qvdpautest on my frontend just now, and I see this:
MIXER TEMPORAL_SPATIAL + IVTC (1920x1080): 83 fields/s

It appears to be fast enough now? 
Below is my full test.



qvdpautest 0.5.2
Intel(R) Atom(TM) CPU D2550   @ 1.86GHz
NVIDIA GPU GeForce 610M (GF119) at PCI:4:0:0 (GPU-0)

VDPAU API version : 1
VDPAU implementation : NVIDIA VDPAU Driver Shared Library  319.60  Wed Sep 25 14:27:11 PDT 2013

SURFACE GET BITS: 190.406 M/s
SURFACE PUT BITS: 160.32 M/s

MPEG DECODING (1920x1080): 184 frames/s
MPEG DECODING (1280x720): 406 frames/s
H264 DECODING (1920x1080): 128 frames/s
H264 DECODING (1280x720): 269 frames/s
VC1 DECODING (1440x1080): 83 frames/s
MPEG4 DECODING (1920x1080): 130 frames/s

MIXER WEAVE (1920x1080): 584 frames/s
MIXER BOB (1920x1080): 891 fields/s
MIXER TEMPORAL (1920x1080): 264 fields/s
MIXER TEMPORAL + IVTC (1920x1080): 167 fields/s
MIXER TEMPORAL + SKIP_CHROMA (1920x1080): 345 fields/s
MIXER TEMPORAL_SPATIAL (1920x1080): 104 fields/s
MIXER TEMPORAL_SPATIAL + IVTC (1920x1080): 83 fields/s
MIXER TEMPORAL_SPATIAL + SKIP_CHROMA (1920x1080): 116 fields/s
MIXER TEMPORAL_SPATIAL (720x576 video to 1920x1080 display): 354 fields/s
MIXER TEMPORAL_SPATIAL + HQSCALING (720x576 video to 1920x1080 display): 210 fields/s

MULTITHREADED MPEG DECODING (1920x1080): 178 frames/s
MULTITHREADED MIXER TEMPORAL (1920x1080): 209 fields/s

Greg