Mailing List Archive

On Sun, Mar 13, 2011 at 12:06 AM, <f-myth-users@media.mit.edu> wrote:
>    > Date: Sat, 12 Mar 2011 19:36:04 -0500
>    > From: Devin Heitmueller <dheitmueller@kernellabs.com>
>
>    > On Sat, Mar 12, 2011 at 7:05 PM, Andy Walls <awalls@md.metrocast.net> wrote:
>    > > I'm really busy for the next 2 weeks, so I won't be able to look into
>    > > any possible problem myself until then.  Even then I'll have to jump
>    > > through hoops to get a device that can receive S-Video near my PVR-350.
>
>    > Or just spend $20.00 and buy an Nvidia card with VDPAU and S-video
>    > output.  Given how cheap they are nowadays, it probably isn't worth
>    > the effort to debug.
>
> Are there any?
>
> S-video output on a VDPAU-capable card seems increasingly difficult
> to find, especially for newer feature sets.  Any recommendations for
> (a) PCI (not PCIe), (b) passively cooled or very quiet fan, (c) recent
> feature set, (d) S-video?  The combo of all of these seemed impossible
> the last time I looked.  Even dropping (b) didn't help much.  And, in
> my case, dropping (a) means a new frontend.  But to continue using my
> Sony CRT, I'd require S-Video, and that wasn't looking promising
> post-8400 or so, if 8400's are even still on the market.
>
> (Also, for someone who only does SD and has limited PCI slots, not
> using the 350's decoder means you need an extra slot for that new
> card, so abandoning the 350's output may present difficulties for
> the OP.  We don't know his configuration.)

Yeah, a quick search on newegg has some 6200s in PCI with S-video out,
but they don't do VDPAU.

I guess if his PC is 5+ years old and cannot handle SD MPEG decoding
in software, then his best bet is to not upgrade. Or track it down
himself and submit a patch. Or pay Andy a few thousand bucks to track
down the issue for him (let's not forget that Andy's time is
incredibly valuable and you could argue is much better spent on newer
devices).

Ultimately this is a matter of priorities: is Andy's time better
spent making current hardware work or tracing down regressions in six
year old devices that are largely obsolete? It's ultimately his
decision what he wants to waste^H^H^H^H^H spend his time on, but I
certainly couldn't blame him if this issue isn't it.

Devin

--
Devin J. Heitmueller - Kernel Labs
http://www.kernellabs.com

_______________________________________________
ivtv-users mailing list
ivtv-users@ivtvdriver.org
http://ivtvdriver.org/mailman/listinfo/ivtv-users

> looks bad on S-Video 1: grainy and desaturated color.
> but good on Composite 1.
>
> I'm using:
> * kernel 2.6.26-ARCH i686 (old because the Arch-based LinHES R6 uses
> only that kernel)
> * PVR-350
> * recommended firmware from ivtvdriver.org
> * S-video 1 output or Composite 1 output
>
> This identical hardware (including cables, tv, etc) gave me a nice
> picture recently, so I don't think it is hardware related.

>saa7127 2-0044: saa7129 found @ 0x88 (ivtv i2c driver #0)

There was for a short period a problem with S-Video and saa7129, which was
present in the first 2.6.28 kernel. Maybe your arch kernel has a v4l-dvb
snapshot from that time which includes this problem

Older driver versions showed a green screen when loading ivtv. This was
because the saa712x was activated before the MPEG decoder had been
initialized.
I fixed this with this patch:
http://linuxtv.org/hg/v4l-dvb/rev/e95d50985dde

Unfortunately it had a side effect for saa7129 chips which would explain your
problem. This was fixed with this patch:
http://linuxtv.org/hg/v4l-dvb/rev/4c1561cccc22

Have a look at the sources for your kernel. maybe it has the first patch, but
not the second.

_______________________________________________
ivtv-users mailing list
ivtv-users@ivtvdriver.org
http://ivtvdriver.org/mailman/listinfo/ivtv-users

On Sun, 2011-03-13 at 12:18 +0100, Martin Dauskardt wrote:
> > looks bad on S-Video 1: grainy and desaturated color.
> > but good on Composite 1.
> >
> > I'm using:
> > * kernel 2.6.26-ARCH i686 (old because the Arch-based LinHES R6 uses
> > only that kernel)
> > * PVR-350
> > * recommended firmware from ivtvdriver.org
> > * S-video 1 output or Composite 1 output
> >
> > This identical hardware (including cables, tv, etc) gave me a nice
> > picture recently, so I don't think it is hardware related.
>
> >saa7127 2-0044: saa7129 found @ 0x88 (ivtv i2c driver #0)
>
> There was for a short period a problem with S-Video and saa7129, which was
> present in the first 2.6.28 kernel. Maybe your arch kernel has a v4l-dvb
> snapshot from that time which includes this problem
>
> Older driver versions showed a green screen when loading ivtv. This was
> because the saa712x was activated before the MPEG decoder had been
> initialized.
> I fixed this with this patch:
> http://linuxtv.org/hg/v4l-dvb/rev/e95d50985dde
>
> Unfortunately it had a side effect for saa7129 chips which would explain your
> problem. This was fixed with this patch:
> http://linuxtv.org/hg/v4l-dvb/rev/4c1561cccc22


Thanks Martin.

I didn't remember these.

Regards,
Andy


_______________________________________________
ivtv-users mailing list
ivtv-users@ivtvdriver.org
http://ivtvdriver.org/mailman/listinfo/ivtv-users

On Sun, 2011-03-13 at 00:26 -0500, Devin Heitmueller wrote:
> On Sun, Mar 13, 2011 at 12:06 AM, <f-myth-users@media.mit.edu> wrote:
> > > Date: Sat, 12 Mar 2011 19:36:04 -0500
> > > From: Devin Heitmueller <dheitmueller@kernellabs.com>
> >
> > > On Sat, Mar 12, 2011 at 7:05 PM, Andy Walls <awalls@md.metrocast.net> wrote:
> > > > I'm really busy for the next 2 weeks, so I won't be able to look into
> > > > any possible problem myself until then. Even then I'll have to jump
> > > > through hoops to get a device that can receive S-Video near my PVR-350.
> >
> > > Or just spend $20.00 and buy an Nvidia card with VDPAU and S-video
> > > output. Given how cheap they are nowadays, it probably isn't worth
> > > the effort to debug.
> >
> > Are there any?
> >
> > S-video output on a VDPAU-capable card seems increasingly difficult
> > to find, especially for newer feature sets. Any recommendations for
> > (a) PCI (not PCIe), (b) passively cooled or very quiet fan, (c) recent
> > feature set, (d) S-video? The combo of all of these seemed impossible
> > the last time I looked. Even dropping (b) didn't help much. And, in
> > my case, dropping (a) means a new frontend. But to continue using my
> > Sony CRT, I'd require S-Video, and that wasn't looking promising
> > post-8400 or so, if 8400's are even still on the market.
> >
> > (Also, for someone who only does SD and has limited PCI slots, not
> > using the 350's decoder means you need an extra slot for that new
> > card, so abandoning the 350's output may present difficulties for
> > the OP. We don't know his configuration.)
>
> Yeah, a quick search on newegg has some 6200s in PCI with S-video out,
> but they don't do VDPAU.
>
> I guess if his PC is 5+ years old and cannot handle SD MPEG decoding
> in software, then his best bet is to not upgrade. Or track it down
> himself and submit a patch.



> Or pay Andy a few thousand bucks to track
> down the issue for him (let's not forget that Andy's time is
> incredibly valuable and you could argue is much better spent on newer
> devices).

For the record, my time is non-existant ATM. Two different divisions of
$VERY_BIG_CUSTOMER need products from me in the immediate future.
$VERY_BIG_CUSTOMER pays the bills....


> Ultimately this is a matter of priorities: is Andy's time better
> spent making current hardware work or tracing down regressions in six
> year old devices that are largely obsolete?

To be fair, obsolete hardware (standard PCI and CX2341[56] chips) is
implied in the name of the ivtv mailing lists, so it is on-topic, and
valid to ask.

But yes, this comic strip captures the technology price trends nicely:

http://www.dilbert.com/fast/2002-06-08/


> It's ultimately his
> decision what he wants to waste^H^H^H^H^H spend his time on, but I
> certainly couldn't blame him if this issue isn't it.

1. Old kernel and driver versions
2. A problem I am unlikely to reproduce
3. A card that is obsolete and could be starting to fail due to age
4. No free time
5. Compsite out works
6. A user who has "an itch to scratch"
7. A publicly available datasheet
8. S-Video fixes already applied in more modern kernels

All those factors do drive this to a low-prioirty problem for me.



>From a system perspective, which includes human eyes and grey-matter,
the big difference between CVBS and S-Video is the high frequency notch
in the CVBS Luminance. A receiver's "sharpness" control can
perceptually make the image better. We're talking about interlaced
video, with digitally undersampled color, potentially sourced from
decompressed MPEG which is lossy, so human perception is already being
greatly exploited anyway.


Regards,
Andy


_______________________________________________
ivtv-users mailing list
ivtv-users@ivtvdriver.org
http://ivtvdriver.org/mailman/listinfo/ivtv-users

> Date: Sun, 13 Mar 2011 00:26:37 -0500
> From: Devin Heitmueller <dheitmueller@kernellabs.com>

> [ . . . ]

> Ultimately this is a matter of priorities: [ . . . ]

You've sorta half missed my point here, which was probably my fault.

My point was -not- to browbeat Andy into fixing this possible
reversion---though I'll note he's already sent some debugging
info to the OP and maybe if it's a true reversion we'll hear
from others as well. (Plus all the recent traffic that implies
this might already have been fixed.)

My point is that, like the OP, I have a 350 driving S-Video to a CRT.
(It's using very old software that I have no intention of upgrading
until I go to VDPAU, or I'd try to reproduce the OP's issue.) I would
like to upgrade someday, but it's very difficult for analog-input CRT
people like myself. Flip answers like "just use VDPAU to S-Video"
ignore the problem that very few cards even have it.

The best alternative I've seen is the 8400GS, which supposedly will do
VDPAU and S-Video---but it's an old card, without the more powerful
feature sets, it's not clear how long it'll be around, and its power
envelope (and general thermal design) could be better. I'm SD-only at
the moment, but I'd like to transcode a large amount of video to X.264
at some point and play that using VDPAU. When I play that, I'd like
to use good deinterlacers and/or GPU decode support, because even at
SD I might otherwise need a fairly beefy frontend just to play the
video even if I'm not deinterlacing (e.g., if it's still going to a
CRT). And I'd like to be relatively sure that recent firmware won't
accidentaly break interlaced output, which I've occasionally heard.

If I have to dump S-Video, that means dumping a $2K CRT TV that is
working perfectly and takes every analog standard around (composite,
component, and S-Video), and replacing it with something that takes
digital inputs, is probably HD, and probably (therefore) doesn't
render SD content as nicely--- not to mention that I just like CRTs
better for video than LCDs anyway. It also means replacing a whole
bunch of furniture, too. So I'm really not wild about switching away
from S-Video output until and unless I go to HD display.

[.Or, trying out the quality of HDMI-to-S-Video converters, which cost
between $90-$300 or so, which is pretty damned pricey.]

So that's why I asked---since you seemed so confident that this was a
viable alternative, I was hoping you had a rabbit in your hat that I
hadn't seen yet. I wish you had. It's a pity that supporting analog
that's been around for more than half a century isn't worth doing when
the gamer crowd all has LCDs and that's what everyone's pushing now,
but oh well.

So take this as it was intended: a semi-off-topic plea for people to
mention good S-Video alternatives for people who still need analog
outputs (S-Video, composite, or component---in my case, I can use any).

Thanks.

_______________________________________________
ivtv-users mailing list
ivtv-users@ivtvdriver.org
http://ivtvdriver.org/mailman/listinfo/ivtv-users

> S-video output on a VDPAU-capable card seems increasingly difficult
> to find, especially for newer feature sets.  Any recommendations for
<snip>
> (Also, for someone who only does SD and has limited PCI slots, not
> using the 350's decoder means you need an extra slot for that new
> card, so abandoning the 350's output may present difficulties for
> the OP.  We don't know his configuration.)
It's a pretty poor and noisy one :). I also had trouble finding a
fanless PCI Nvidia card with SVideo... My next step may be upgrading
my whole system, although I am very fond of the PVR-350.

Dirk

_______________________________________________
ivtv-users mailing list
ivtv-users@ivtvdriver.org
http://ivtvdriver.org/mailman/listinfo/ivtv-users

> There was for a short period a problem with S-Video and saa7129, which was
> present in the first 2.6.28 kernel. Maybe your arch kernel has a v4l-dvb
> snapshot from that time which includes this problem
<snip>
> Unfortunately it had a side effect for saa7129 chips which would explain your
> problem. This was fixed with this patch:
> http://linuxtv.org/hg/v4l-dvb/rev/4c1561cccc22
>
> Have a look at the sources for your kernel. maybe it has the first patch, but
> not the second.

Thank you Martin! Will definitely check that out.

Cheers,
Dirk

_______________________________________________
ivtv-users mailing list
ivtv-users@ivtvdriver.org
http://ivtvdriver.org/mailman/listinfo/ivtv-users