Hi,
We've been happily using VDR with a video projector for almost a year now without any problems. The projector gets a s-video signal from VDR using one of those "add-on" cards that plug into the J2 connector. S-video gives an ok but kinda blurry image. The card seems to be able to provide some kind of component signal too as it has an additional SCART-connector as a separate module, but we've never figured out how to get that one connected and as we've already burnt the J2 of one DVB card, we're a bit careful with anything connected to the J2 connector.
So, as s-video seems to be the best we can get out from the DVB cards we've looked at getting the image out using the normal video output of the system. The projector has an input for DVI/HDMI/VGA and the VDR system has DVI/VGA out, so it seems like that would be the best way to go to get a good image out, preferably avoiding any analog conversions along the way.
There seems to be two solutions that are currently used:
Softdevice http://www.vdr-wiki.de/wiki/index.php/Softdevice-plugin
Xine-plugin http://www.vdr-wiki.de/wiki/index.php/Xine-plugin
Those seem to be the best homepages for the plugins, but as I'm no native German speaker (I can order a beer, but that's about it) I have a fairly hard time grokking the finer details of pros and cons.
What do you others use when you want to get a good image from VDR? Which of those plugins is the preferable solution in this case? The system in use should cope just fine with some decoding done by the CPU. The system has a Sempron-something 2400+ and an ATI fanless 9250 card (I think it was).
Comments?
On Mon, Sep 26, 2005 at 11:53:09AM +0300, Jan Ekholm wrote:
There seems to be two solutions that are currently used:
Softdevice http://www.vdr-wiki.de/wiki/index.php/Softdevice-plugin
I think that http://softdevice.berlios.de/ is the official home page.
Xine-plugin http://www.vdr-wiki.de/wiki/index.php/Xine-plugin
Those seem to be the best homepages for the plugins, but as I'm no native German speaker (I can order a beer, but that's about it) I have a fairly hard time grokking the finer details of pros and cons.
What do you others use when you want to get a good image from VDR? Which of those plugins is the preferable solution in this case? The system in use should cope just fine with some decoding done by the CPU. The system has a Sempron-something 2400+ and an ATI fanless 9250 card (I think it was).
I'm using softdevice with a budget DVB-T card (Hauppauge Nova-T PCI 90002) on a Matrox G450 on DirectFB. For some reason, the Matrox TV-out connection eats more CPU than VGA. The 900 MHz Intel Celeron has enough power to deinterlace the picture. I think that something like 500 or 600 MHz would be the minimum requirement on a Matrox card. Other cards may require more processing power.
I don't have any experience with the Xine-plugin, but softdevice has gotten much more stable during this year. Crashes occur very rarely nowadays, even if the signal is bad.
Marko
Jan Ekholm a écrit :
Hi,
Softdevice http://www.vdr-wiki.de/wiki/index.php/Softdevice-plugin Xine-plugin http://www.vdr-wiki.de/wiki/index.php/Xine-plugin
Comments?
Basically, you'd better use softdevice when your PC is dedicated to VDR (no other output than the DVB video, no keyboard/mouse). Xine is better suited if you don't mind use a keyboard, or the PC is not dedicated. Reason is : vdr + xine is a client + server setup, and you have to start both programs independently, or have a reliable starter to launch/restart both.
Xine can make use of XvMC accelerator, but you don't have one with ATI (ie. no implementation), and your CPU is good enough.
softdevice will bring you quite the same environment as the FF-card you leave (single process, a few more threads, etc.) It is in really good shape these days, and softplay is a great add-on to it. There may eventually be hardware acceleration things in the future.
There is another plugin : xineliboutput. This one is in the middle of the previous ones : uses xinelib for output, all linked in a single process.
I wrote a post on this list regarding the possibilities, and someone uploaded it to the english Wiki on linuxtv.org. Check there too.
Jan Ekholm wrote:
What do you others use when you want to get a good image from VDR? Which of those plugins is the preferable solution in this case?
I tried with a small VGA monitor with good results and with Nvidia s-video out for normal TV but found out soon that deinterlacing and again interlacing for s-video output is not a good idea.
Softdevice + I was able to make good looking 16:9 display settings (crop the nasty blue lines from left of 4:3 video) + Accompanied softplayer plugin uses familiar vdr look & feel for other video formats and music playback + Quite simple to setup for a set-top-box like setup - I have experienced some but fortunately quite rare hang-ups, watchdog restart may trash ongoing recordings - Not as many deinterlacers and post-processing settings as in Xine - Sometimes stuttering in video, especially in 1080i HDTV
Vdr-xine + Very stable, xine-ui or gxine may crash (very rare) but I have never seen vdr crashes or hangs + Smooth playback, normal TV & 1080i HDTV - I was not able to get rid of the blue vertical lines in 4:3 picture when rendered to 16:9 screen with correct aspect ratio. I couldn't find a suitable crop setting. Zoom in feature in xine doesn't help unless 4:3 picture is zoomed to fill entire 16:9 screen. - Requires quite a lot of scripting work for automatic running set-top-box like system, softdevice x11 requires too but here we have a bit more to care about +/- It is possible to use xine or vdr plugins to play other media such as mp3, avi, etc. I've preferred vdr plugins for music and xine for dvd playback.
Both of these are quite good choises in my opinion :^)
Cheers, Seppo
On Mon, 26 Sep 2005 11:53:09 +0300 (EEST) "Jan Ekholm" chakie@smultron.net wrote:
There seems to be two solutions that are currently used:
Softdevice http://www.vdr-wiki.de/wiki/index.php/Softdevice-plugin Xine-plugin http://www.vdr-wiki.de/wiki/index.php/Xine-plugin
Those seem to be the best homepages for the plugins, but as I'm no native German speaker (I can order a beer, but that's about it) I have a fairly hard time grokking the finer details of pros and cons.
There is also some information in Linuxtv.org VDR Wiki: http://www.linuxtv.org/vdrwiki/index.php
What do you others use when you want to get a good image from VDR? Which of those plugins is the preferable solution in this case? The system in use should cope just fine with some decoding done by the CPU. The system has a Sempron-something 2400+ and an ATI fanless 9250 card (I think it was).
Xine has superior deinterlacing filters such as TomsMoComp and Greedy2Frame, which is very important for video fluidity and quality. I have not followed Sofdevice development recently, but at least with Xine-plugin, the MPEG-2 decoding and video display is separated from VDR to the Xine process. This should prevent VDR from crashing at decoding bugs (if there are any) and you can close Xine and X without worrying about VDR.
ATI's X drivers may give you some gray hairs, but I have heard they've gotten better in the last year or so. Your system is powerful enough to handle HDTV software decoding and deinterlacing if you get XVideo working.
Regards, Niko Mikkilä
There is also some information in Linuxtv.org VDR Wiki: http://www.linuxtv.org/vdrwiki/index.php
Thanks, I didn't know about that one. Good stuff.
Xine has superior deinterlacing filters such as TomsMoComp and Greedy2Frame, which is very important for video fluidity and quality. I have not followed Sofdevice development recently, but at least with Xine-plugin, the MPEG-2 decoding and video display is separated from VDR to the Xine process. This should prevent VDR from crashing at decoding bugs (if there are any) and you can close Xine and X without worrying about VDR.
Well, as this is a dedicated VDR box without any keyboard, mouse or normal VGA screen restarting Xine is the same as restarting the whole system. I don't have a desktop where Xine will be running and where I can just restart it, it's all on the same headless system.
Good deinterlace is of course important, especially with a projector that natively wants a progressive signal.
ATI's X drivers may give you some gray hairs, but I have heard they've gotten better in the last year or so. Your system is powerful enough to handle HDTV software decoding and deinterlacing if you get XVideo working.
Heh, you expect us to see HDTV in Finland for a few years? Of course some HDTV content is available through satellite and could be fun to play with (still need to bite the gory tools and install a dish for that)... :)
Thanks for the feedback, it was much appreciated.
On Mon, Sep 26, 2005 at 11:53:09AM +0300, Jan Ekholm wrote:
I'm using softdevice with a budget DVB-T card (Hauppauge Nova-T PCI 90002) on a Matrox G450 on DirectFB. For some reason, the Matrox TV-out connection eats more CPU than VGA. The 900 MHz Intel Celeron has enough power to deinterlace the picture. I think that something like 500 or 600 MHz would be the minimum requirement on a Matrox card. Other cards may require more processing power.
Can one run DirectFB with a card that doesn't have a direct driver? AFAIK only Matrox cards and some Epias really work with DirectFB, others have no drivers. I fought quite a lot with DirectFB about a year and a half ago when it was investigated for another purpose, but in the end DFB didn't really work too well (and that was with a more or less supported CLE266 chip). Somehow DFB also felt a bit heavy and CPU intensive compared to using normal X11.
On Mon, 26 Sep 2005 15:06:07 +0300 (EEST) "Jan Ekholm" chakie@smultron.net wrote:
Well, as this is a dedicated VDR box without any keyboard, mouse or normal VGA screen restarting Xine is the same as restarting the whole system. I don't have a desktop where Xine will be running and where I can just restart it, it's all on the same headless system.
If the MPEG-2 decoder or some other display component in Xine crashes on a corrupted packet, VDR would still stay up. Also simultaneous recordings wont be interrupted. And even on a "headless" system (you still have the head which you use Xine on, don't you), with some scripting, you can shut Xine down when you are not actually watching anything, but there are recordings going on. This further improves stability.
Heh, you expect us to see HDTV in Finland for a few years? Of course some HDTV content is available through satellite and could be fun to play with (still need to bite the gory tools and install a dish for that)... :)
I meant that you don't have to worry about your system being fast enough for software MPEG-2 decoding. Our HDTV will be using MPEG-4 AVC format, which requires a couple times more horsepower though. I think we'll probably have to wait 3-5 years for terrestrial broadcasts.
-- Niko Mikkilä
Jan Ekholm wrote:
There seems to be two solutions that are currently used:
Softdevice http://www.vdr-wiki.de/wiki/index.php/Softdevice-plugin Xine-plugin http://www.vdr-wiki.de/wiki/index.php/Xine-plugin
Those seem to be the best homepages for the plugins, but as I'm no native German speaker (I can order a beer, but that's about it) I have a fairly hard time grokking the finer details of pros and cons.
vdr-xine plugin http://home.vr-web.de/~rnissl/ read the docs, they cover most issues regarding features and installation.
With a PJ I would suggest that the most critical thing to the setup after cabling is deinterlace quality. xine wins out on this. also best to use DVI where possible, ie any digital source.
I've been using vdr-xine since its release and the combination has been stunningly stable. I've had more trouble with hardware than this software. I started with the oxine frontend and later eboxy. oxine is looking quite good again, although Im now working on my own using xulrunner.
You can get away with not using a seperate frontend by adding more vdr plugins. I use the frontend to switch to a different xine setup to play dvds so I can get the best out of the material playing while using the same remote. The frontend is like a simple version of mythTV.
Simon
Le lundi 26 septembre 2005 à 15:33 +0300, Niko Mikkila a écrit :
If the MPEG-2 decoder or some other display component in Xine crashes on a corrupted packet, VDR would still stay up. Also simultaneous recordings wont be interrupted. And even on a "headless" system (you still have the head which you use Xine on, don't you), with some scripting, you can shut Xine down when you are not actually watching anything, but there are recordings going on. This further improves stability.
I have a button on the desktop which when clicked starts xine full screen and connected to the VDR live TV stream. From then on everything is controlled via the remote including switch off and playing back recordings. I have gnome configured to start xine full screen when I insert a DVD too and from then on everything is controlled via the remote.
The main advantage of running X windows is that the box can be used for other things than watching TV. And on LCD or DLP projectors the quality of small text is just fine. My dream is a big LCD and a wireless keyboard to replace the ugly desk in my SOHO.
I meant that you don't have to worry about your system being fast enough for software MPEG-2 decoding. Our HDTV will be using MPEG-4 AVC format, which requires a couple times more horsepower though. I think we'll probably have to wait 3-5 years for terrestrial broadcasts.
And then it will be so locked down by DRM that you will need a Nokia box to see it...
Cheers
Tony
hi,
but there are recordings going on. This further improves stability.
I have a dedicated VDR box running vdr-xine with P4 2.4GHz and Radeon 9200.
Stability is not currently a problem (1.3.27) - since spring the system has been easily up more than a day :) It has stayed up without problems for at least a week, but due to my usage of nvram-wakeup, I have no statistics of longer uptime.
The reason why a separate xine is nice is that currently the setup eats up all the CPU cycles. Killing xine for the time when I am doing something on the box like updating will speed things up considerably, and the VDR itself will continue recording if it has something to record etc. without disturbing.
Btw. if someone has any hints on how to reduce used cpu cycles, they are warmly welcome. And yes, I use the fglrx driver - without that the picture is jerky.
yours, Jouni
Jan Ekholm wrote:
Well, as this is a dedicated VDR box without any keyboard, mouse or normal VGA screen restarting Xine is the same as restarting the whole system. I don't have a desktop where Xine will be running and where I can just restart it, it's all on the same headless system.
you can use irexec and lircd to execute a script from your remote :-) Thus keeping a keyboard free zone.
Good deinterlace is of course important, especially with a projector that natively wants a progressive signal.
so right. The thing I really wish for in xine is a magic deblocking filter to remove artifacts in high motion scenes on dvb. This situation has improved a little over the last two years.
A good PJ can highlight even the most subtle artifacts in mpeg :-/ good video processing plugins are very important and xine provides the flexibility and choice here.
Simon
tony wrote:
I have a button on the desktop which when clicked starts xine full screen and connected to the VDR live TV stream. From then on everything is controlled via the remote including switch off and playing back recordings. I have gnome configured to start xine full screen when I insert a DVD too and from then on everything is controlled via the remote.
I use matchbox, I found the heavy weight desktops just got in my way.
http://projects.o-hand.com/matchbox/
The main advantage of running X windows is that the box can be used for other things than watching TV. And on LCD or DLP projectors the quality of small text is just fine. My dream is a big LCD and a wireless keyboard to replace the ugly desk in my SOHO.
check out the gyroscopic mouse. I bought one with a keyboard a few years ago, no desk required :-) dead handy when I want to do something from the sofa and cannot be bothered to walk into the study.
Simon
Jan Ekholm a écrit :
Can one run DirectFB with a card that doesn't have a direct driver? AFAIK
You can still use vesafb with the generic software renderer, but you will lack most hardware optimizations, specialy pixel format conversions.
only Matrox cards and some Epias really work with DirectFB, others have no drivers. I fought quite a lot with DirectFB about a year and a half ago when it was investigated for another purpose, but in the end DFB
There is an ATI driver since this time, I think. And DirectFB improves regularly.
didn't really work too well (and that was with a more or less supported CLE266 chip). Somehow DFB also felt a bit heavy and CPU intensive compared to using normal X11.
DFB must be way lighter than X. In my case, it is nearly 80MB RAM less (counting X + Xine - softdevice). CPU utilization depends on hardware acceleration, WTR MPEG2. Nothing should impact CPU in DFB per se.
Jouni Karvo a écrit :
hi,
but there are recordings going on. This further improves stability.
I have a dedicated VDR box running vdr-xine with P4 2.4GHz and Radeon 9200.
Stability is not currently a problem (1.3.27) - since spring the system has been easily up more than a day :) It has stayed up without problems for at least a week, but due to my usage of nvram-wakeup, I have no statistics of longer uptime.
On my side, I never had a crash with softdevice on a 24/7 machine since 05/2005 I sometimes need to reload DVB driver since Skystar looses IRQ handling (apparently solved in 2.6.13), so VDR is not really 24/7, but keeps working many days in a run.
Btw. if someone has any hints on how to reduce used cpu cycles, they are warmly welcome. And yes, I use the fglrx driver - without that the picture is jerky.
The softdevice developers are really productive : there is a recent "suspend-patch" for VDR, that allows to stop MPEG routing to the display when not in use. This should apply to any output method (FF-card, Xine, softdevice...) This is linked to a key on the remote, to allow easy suspend/resume.
Le lundi 26 septembre 2005 à 17:07 +0200, Nicolas Huillard a écrit :
I sometimes need to reload DVB driver since Skystar looses IRQ handling (apparently solved in 2.6.13), so VDR is not really 24/7, but keeps working many days in a run.
Hmmm... I think this may be the issue that I am having with 2.6.11.10 Guess I'll try newer modules
Tony
On Mon, 26 Sep 2005 17:01:42 +0200 Nicolas Huillard nhuillard@e-dition.fr wrote:
DFB must be way lighter than X. In my case, it is nearly 80MB RAM less (counting X + Xine - softdevice). CPU utilization depends on hardware acceleration, WTR MPEG2. Nothing should impact CPU in DFB per se.
I agree that DFB is lighter on memory usage, but to be fair, X is not such a memory hog either. Plain X.org reserves about 10 MB physical memory and that can be cut down by disabling unnecessary extensions. It may be a bit difficult to measure actual X.org/XFree86 memory usage because the numbers generally include the display framebuffer that actually resides on the graphics card (don't know how DFB manages display memory). X may also store some client pixmaps on the server.
So, for me X uses 10 MB without a window manager and Xine-ui uses 32 MB (of which 12 MB shared) when playing a 640x480 MPEG-2 video. Add a light WM and you'll get 40 to 50 MB combined. Hardly significant for a stand-alone box without other apps than VDR running.
-- Niko Mikkilä
On Mon, Sep 26, 2005 at 05:07:35PM +0200, Nicolas Huillard wrote:
The softdevice developers are really productive : there is a recent "suspend-patch" for VDR, that allows to stop MPEG routing to the display when not in use. This should apply to any output method (FF-card, Xine, softdevice...) This is linked to a key on the remote, to allow easy suspend/resume.
I wouldn't consider myself a softdevice developer. I've merely submitted some bug reports and suggestions, and Stefan Lucke and Martin Wache have been doing the hard work.
My suspend patch and the relay control plugin (available from http://www.funet.fi/~msmakela/software/vdr/) work on vdr 1.3.30 and above. I've tested the most recent versions on vdr 1.3.32 and 1.3.33.
I cannot recommend the suspend patch for users of hardware-based MPEG decoders. The proper way to suspend the video output would be to send an encoded black picture to the output device, like the suspendoutput plugin does. Some day, I might try to improve that plugin, but don't hold your breath.
Maybe vdr 1.5 will get a suspend feature out of the box, given that software-based decoding solutions can decode HDTV channels on powerful processors with advanced power saving features. On my Celeron system, suspending the video output will reduce power intake by about 15 watts - perhaps a little more if the VGA output were somehow switched off.
Marko
On Monday 26 September 2005 18:01, Nicolas Huillard wrote:
Jan Ekholm a écrit :
Can one run DirectFB with a card that doesn't have a direct driver? AFAIK
You can still use vesafb with the generic software renderer, but you will lack most hardware optimizations, specialy pixel format conversions.
Meaning that it's fast and good if there is a driver for the used card and a bit useless if there isn't?
only Matrox cards and some Epias really work with DirectFB, others have no drivers. I fought quite a lot with DirectFB about a year and a half ago when it was investigated for another purpose, but in the end DFB
There is an ATI driver since this time, I think. And DirectFB improves regularly.
Ah, ok. I'll have a look. Back then DFB didn't really move at all, I think it was stuck on the same release for half a year or something, and the mailing-list I was on was dead for months.
didn't really work too well (and that was with a more or less supported CLE266 chip). Somehow DFB also felt a bit heavy and CPU intensive compared to using normal X11.
DFB must be way lighter than X. In my case, it is nearly 80MB RAM less (counting X + Xine - softdevice). CPU utilization depends on hardware acceleration, WTR MPEG2. Nothing should impact CPU in DFB per se.
I meant that it felt heavy and slow. Sure it is lighter when looking at raw software, there is just so many layers less and so much less memory used, but still X felt snappier at drawing stuff.
But, I'll believe someone who actually uses in and give it try. Thanks for the encouragement. :)
I demand that Nicolas Huillard may or may not have written...
[snip]
DFB must be way lighter than X. In my case, it is nearly 80MB RAM less (counting X + Xine - softdevice). CPU utilization depends on hardware acceleration, WTR MPEG2. Nothing should impact CPU in DFB per se.
Something else to try: GTK+ (fb port) with gxine. I'd be interested in knowing how well this works and in any patches for gxine which are needed to make this work (if it doesn't already).
I demand that Jouni Karvo may or may not have written...
[snip]
I have a dedicated VDR box running vdr-xine with P4 2.4GHz and Radeon 9200.
Stability is not currently a problem (1.3.27) - since spring the system has been easily up more than a day :) It has stayed up without problems for at least a week, but due to my usage of nvram-wakeup, I have no statistics of longer uptime.
The reason why a separate xine is nice is that currently the setup eats up all the CPU cycles. [...]
Hmm? I'm seeing around 30% CPU usage with gxine (2GHz Athlon, Radeon 9200, 266MHz bus, AGP 4x).
Btw. if someone has any hints on how to reduce used cpu cycles, they are warmly welcome. And yes, I use the fglrx driver - without that the picture is jerky.
Something's evidently not right with your setup. Are you using Xv? Have you tried the native R200 support in a recent 2.6 kernel and X?
A patch which recently went into xine-lib CVS HEAD helps somewhat (it optimises the video buffer allocation somewhat, which affects things like software deinterlacing).
Jan Ekholm skrev:
Hi, snip
What do you others use when you want to get a good image from VDR? Which of those plugins is the preferable solution in this case? The system in use should cope just fine with some decoding done by the CPU. The system has a Sempron-something 2400+ and an ATI fanless 9250 card (I think it was).
What about subtitling? Will ttxtsubs work with Softdevice?
/t
On Tue, Sep 27, 2005 at 09:06:09AM +0200, Tomas Prybil wrote:
What about subtitling? Will ttxtsubs work with Softdevice?
For what it is worth, DVBsubs (the vdr-subtitles plugin) does work.
Some months ago (before release 0.1.2 if I remember correctly), there was a bug in softdevice DirectFB output that displayed the OSD overlay at the wrong position, making vdr-subtitles output invisible.
Marko
On Mon, 2005-09-26 at 21:26 +0100, Darren Salt wrote:
I demand that Jouni Karvo may or may not have written...
Btw. if someone has any hints on how to reduce used cpu cycles, they are warmly welcome. And yes, I use the fglrx driver - without that the picture is jerky.
Something's evidently not right with your setup. Are you using Xv? Have you tried the native R200 support in a recent 2.6 kernel and X?
A patch which recently went into xine-lib CVS HEAD helps somewhat (it optimises the video buffer allocation somewhat, which affects things like software deinterlacing).
There's also a new R200 driver in DirectFB CVS. http://www.directfb.org/index.php/viewcvs.cgi/DirectFB/gfxdrivers/r200/
I haven't tested it myself nor heard from anyone who would have, though. If someone has tried it, reports would be appreciated.
On Tuesday 27 September 2005 18:43, Ville Skyttä wrote:
There's also a new R200 driver in DirectFB CVS. http://www.directfb.org/index.php/viewcvs.cgi/DirectFB/gfxdrivers/r200/
I haven't tested it myself nor heard from anyone who would have, though. If someone has tried it, reports would be appreciated.
Any ideas wether the CVS version normally is useable? Some projects mostly check in stuff that could be a release, while some projects check in stuff that doesn't even compile. My playing time is very limited so I rather not test with something that doesn't even have a chance of working.
On Tue, 2005-09-27 at 20:38 +0300, Jan Ekholm wrote:
On Tuesday 27 September 2005 18:43, Ville Skyttä wrote:
There's also a new R200 driver in DirectFB CVS. http://www.directfb.org/index.php/viewcvs.cgi/DirectFB/gfxdrivers/r200/
I haven't tested it myself nor heard from anyone who would have, though. If someone has tried it, reports would be appreciated.
Any ideas wether the CVS version normally is useable?
No idea.
Some projects mostly check in stuff that could be a release, while some projects check in stuff that doesn't even compile. My playing time is very limited so I rather not test with something that doesn't even have a chance of working.
Yep, that's why I haven't bothered to test it either :)
On Tuesday 27 September 2005 20:40, Ville Skyttä wrote:
On Tue, 2005-09-27 at 20:38 +0300, Jan Ekholm wrote:
On Tuesday 27 September 2005 18:43, Ville Skyttä wrote:
There's also a new R200 driver in DirectFB CVS. http://www.directfb.org/index.php/viewcvs.cgi/DirectFB/gfxdrivers/r200/
I haven't tested it myself nor heard from anyone who would have, though. If someone has tried it, reports would be appreciated.
Any ideas wether the CVS version normally is useable?
No idea.
Some projects mostly check in stuff that could be a release, while some projects check in stuff that doesn't even compile. My playing time is very limited so I rather not test with something that doesn't even have a chance of working.
Yep, that's why I haven't bothered to test it either :)
Maybe I could give it a shot one of these days. The thing with DFB was at least a few years ago that it wasn't simple to test. You couldn't just compile it and then start, there was so much to configure before that and then you had to have some application that showed something meaningful, otherwise you only had garbage on the screen. But I remember that it had quite flexible "modes" so it shouldn't be impossible to tweak for our projectors 852x480 resolution. Just have to find that simple app that always works and can give some initial output before I dive into softdevice. :)
hi,
Darren Salt writes:
The reason why a separate xine is nice is that currently the setup eats up all the CPU cycles. [...]
Hmm? I'm seeing around 30% CPU usage with gxine (2GHz Athlon, Radeon 9200, 266MHz bus, AGP 4x).
I used earlier a 2.6GHz P4 HT processor, and the processor load was approx 20% or less. It was really a surprise that a 2.4GHz non HT P4 would be so much slower. So yes, I suspect too that there is something weird going on.
It might be some setting in the kernel compilation parameters (no so obvious as MTRR, though). Top reports quite low percentages for different processes. 60% is shown to be user and 40% sy time, which is weird since the highest %CPU is with XFree86 at approx 6%.
Perhaps something about copying data unoptimally or weird context switching frenzy? (DMA?) I checked the kernel timer settings already (now use HPET).
Btw. if someone has any hints on how to reduce used cpu cycles, they are warmly welcome. And yes, I use the fglrx driver - without that the picture is jerky.
Something's evidently not right with your setup. Are you using Xv? Have you tried the native R200 support in a recent 2.6 kernel and X?
Yes, I use xv (xshm simply is too slow). I have not tried kernel radeon driver - and the current kernel is 2.6.9 (I have not yet been able to get the ATI driver to work on newer kernel versions)
A patch which recently went into xine-lib CVS HEAD helps somewhat (it optimises the video buffer allocation somewhat, which affects things like software deinterlacing).
There is something else fishy, since changing software deinterlacing options does not change cpu usage considerably.
Thanks for your comments.
yours, Jouni
On Dienstag, 27. September 2005 20:37, Jan Ekholm wrote:
On Tuesday 27 September 2005 20:40, Ville Skyttä wrote:
On Tue, 2005-09-27 at 20:38 +0300, Jan Ekholm wrote:
On Tuesday 27 September 2005 18:43, Ville Skyttä wrote:
There's also a new R200 driver in DirectFB CVS. http://www.directfb.org/index.php/viewcvs.cgi/DirectFB/gfxdrivers/r200/
I haven't tested it myself nor heard from anyone who would have, though. If someone has tried it, reports would be appreciated.
Last time I tested my radeon 9200 with directfb it worked.
Any ideas wether the CVS version normally is useable?
I'm currently working with directfb cvs version.
No idea.
Some times it is hard to get compile at the first time. Needed autotools etc.
Hi,
Darren Salt wrote:
Something else to try: GTK+ (fb port) with gxine. I'd be interested in knowing how well this works and in any patches for gxine which are needed to make this work (if it doesn't already).
This is a bit off-topic but perhaps not totally:
I haven't tried fb but instead a minimal X running as 2nd session by enabling one more session and automatically logging in user vdr through gdm.conf settings. The problem with gxine is that command line parameter "-f" doesn't work trough my .xsession script. I think I got a full screen mode but no video (black screen, I don't remember what happened to audio). Gxine also seems to require a window manager so I started metacity as well. This same .xsession script works without "-f" for gxine. Similar script works with xine-ui and it doesn't require a window manager. Gxine has a very nice user interface, it would be good if this could be corrected.
The versions I tested were about two weeks old cvs versions of libxine and gxine.
Here is my .xsession
--
#!/bin/sh
## Start a window manager /usr/bin/metacity &
## TV-out setttings /usr/bin/nvidia-settings -l
## Run Gxine in fullscreen mode /usr/local/bin/gxine -f -S -V xv -A alsa vdr:/tmp/vdr-xine/stream#demux:mpeg_pes
## Run xine in fullscreen mode #/usr/local/bin/xine --no-splash --hide-gui --fullscreen -V xv -A alsa vdr:/tmp/vdr-xine/stream#demux:mpeg_pes
--
Best regards, Seppo
Jan Ekholm a écrit :
On Tuesday 27 September 2005 18:43, Ville Skyttä wrote:
There's also a new R200 driver in DirectFB CVS. http://www.directfb.org/index.php/viewcvs.cgi/DirectFB/gfxdrivers/r200/
I haven't tested it myself nor heard from anyone who would have, though. If someone has tried it, reports would be appreciated.
Any ideas wether the CVS version normally is useable? Some projects mostly check in stuff that could be a release, while some projects check in stuff that doesn't even compile. My playing time is very limited so I rather not test with something that doesn't even have a chance of working.
DirectFB is the kind of project where CVS check-ins work. I vaguely remember the author's post announcing this ATI support, stating that it works great for him and asking for testers. Check the directfb-users ML.
hi,
Darren Salt writes:
The reason why a separate xine is nice is that currently the setup eats up all the CPU cycles. [...]
Hmm? I'm seeing around 30% CPU usage with gxine (2GHz Athlon, Radeon 9200, 266MHz bus, AGP 4x).
<...>
Something's evidently not right with your setup. Are you using Xv? Have you tried the native R200 support in a recent 2.6 kernel and X?
Finally, after upgrading to 2.6.13.2, I found the problem! (what a relief) After upgrading, the CPU cycles were properly reported and top gave the crucial hint...
FYI:
If you start xine in background directly from init scripts with --stdctl it seems to hog all your spare CPU cycles. Removing this (unnecessary in this case) option did drop CPU usage from constant 100% to a total of about 20%.
yours, Jouni