View unanswered posts    View active topics

All times are UTC - 6 hours





Post new topic Reply to topic  [ 20 posts ] 
Go to page 1, 2  Next

Print view Previous topic   Next topic  
Author Message
Search for:
PostPosted: Fri Nov 17, 2006 6:29 pm 
Offline
Joined: Wed Nov 16, 2005 8:55 pm
Posts: 1381
Location: Farmington, MI USA
Hi all,

Hoping someone can shed some light, point me in the right direction, etc. I have an R5D1 system with MythTV .20 applied as instructed in this thread http://mysettopbox.tv/phpBB2/viewtopic.php?t=11558&highlight=mythtv,

First my Hardware:
P4 2.4Ghz 533 FSB (non-HT)
Intel D845PEBT2 motherboard, using onboard sound
512Mb RAM
eVGA e-GeForce 6200 LE AGP 256Mb video adapter (currently using S-Video out)
1 - PVR-350
1 - PVR-150
...and trying to add a Fusion5 HDTV RT Lite
(I have attempted to use the Fusion alone in another test system I have (Celeron 2Ghz, ASRock mobo, 512Mb RAM, same video card) and experience the same issue as below)

OK, all works well until I try to add the Fusion Lite card. I configure it via mythtv-setup (I have deleted all cards and re-setup from scratch) without problems. I can scan the clear-QAM channels (BrightHouse analog cable) - no problem. I go to Watch TV and I get this insane "stuttering" - 2 seconds of video/audio (not sync'd), then 2 seconds of "stopped" (no audio, video not moving). If I record a program and play it back in Xine with the -V xvmc option, it plays perfectly!!

So, I say to myself "Self, others have surely overcome this problem. Off to Google!" But I don't know a modeline from a hole in my head, so I'm not sure which path to travel down.

The funny part (funny as in odd, not as in "Ha ha") is that if I install R5D1 from scratch and don't install .20, I can get fairly useable video out of it with Bob deint, Standard XvMC enabled, using the 8774 nVidia drivers. I say "fairly useable" because I do still notice some pausing/stutter, but it is for the most part watchable. I tried the SVN fixes using WattoToydarian's script from this thread http://mysettopbox.tv/phpBB2/viewtopic.php?t=11985&highlight=mythtv and still have the insane stuttering.

Sorry, didn't intend to write a book, just wanted to include as much information as possible. Anyone out there that would like to take pity on an old man and point out the errors in his ways? I would be most appreciative...

EDIT: After some playing with "Enable OpenGl vertical sync for timing" and a reboot, I now have a "rapid-fire" stutter - Not sure how to explain, more like a machine-gun sound - still not useable.


Last edited by slowtolearn on Fri Dec 15, 2006 6:45 pm, edited 3 times in total.


Top
 Profile  
 
 Post subject:
PostPosted: Fri Nov 17, 2006 10:07 pm 
Offline
Joined: Fri May 21, 2004 11:55 pm
Posts: 1206
Location: Silicon Valley, CA
For HD playback you really need to check the "extra audio buffering" box on the Playback setup screen (same place as the OpenGL box). I also had issues with OpenGL when I had used the "nvidia-settings" program and had selected the OpenGL sync option there. I had to UNselect the nvidia-settings checkbox and select the one in MythTv. Other things to try -- try it with and without the deint filter. The deint may be pushing your processor over the threshold.

I have a 3.0GHz P4 with HT and it has plenty of headroom. But I don't use any deint filtering as I'm using a 1080i native display.

_________________
Do you code to live, or live to code?
Search LinHES forum through Google


Top
 Profile  
 
 Post subject:
PostPosted: Sat Nov 18, 2006 8:14 am 
Offline
Joined: Wed Nov 16, 2005 8:55 pm
Posts: 1381
Location: Farmington, MI USA
Hi Liv2Cod, thanks for the reply.

Liv2Cod wrote:
For HD playback you really need to check the "extra audio buffering" box on the Playback setup screen (same place as the OpenGL box).


I have tried many combinations of settings in the General Playback options. The extra audio buffering option you referred to was enabled out of the box - Tried without, pretty much the same issue (a bit worse perhaps)

Quote:
I also had issues with OpenGL when I had used the "nvidia-settings" program and had selected the OpenGL sync option there. I had to UNselect the nvidia-settings checkbox and select the one in MythTv.


Checked that, thanks. Disabled in nvidia-settings, and either enabled/disabled in Myth gives me the same.

Quote:
Other things to try -- try it with and without the deint filter. The deint may be pushing your processor over the threshold.


Tried each of the deint options (including not using it). Didn't notice that either made a difference with the stuttering, but the picture sure did change for the worse! :)

Quote:
I have a 3.0GHz P4 with HT and it has plenty of headroom. But I don't use any deint filtering as I'm using a 1080i native display.


I've been wondering about this. Perhaps my setup is underpowered? I have seen posts where folks were displaying HD with far lower specs.
And since it works fairly well in 0.19 I'm guessing that's not it, but not sure...

And how do you determine the "modeline" (and any other pertinent) settings for the XF86Config? I will eventually hook this up to a Sony HD-ready wide screen, and assume I will need to make some adjustments there. Any good references you could point me to?

Again, thanks for your help!


Top
 Profile  
 
 Post subject:
PostPosted: Sat Nov 18, 2006 1:47 pm 
Offline
Joined: Mon Aug 01, 2005 10:18 am
Posts: 84
I've heard advice from some people that you have to disable the XVideo sync-to-vblank options in nvidia-settings. Have you tried this yet?

There was a fix in SVN to reduce some insanely high CPU usage with XvMC (http://svn.mythtv.org/trac/ticket/2412, http://svn.mythtv.org/trac/changeset/11327), but this hasn't found its way into 0.20-fixes yet, so you should probably keep trying with Watto's svn script.

As far as modelines go, this site has a pretty comprehensive list of standard modelines for HDTV. You'll probably want ATSC 720p or 1080i at 59.94 hz (or both).[/url]


Top
 Profile  
 
 Post subject:
PostPosted: Sun Nov 19, 2006 9:49 am 
Offline
Joined: Tue Jan 18, 2005 2:07 am
Posts: 1532
Location: California
Here is some info you may find useful:

1. A 2.4ghz P4 w/out HT and w/out XVMC is not enough processing power to deal with hidef if you also need to deinterlace, which it appears you do need to do from your original post.

2. A few of the posts I've read about 0.20 in this forum suggest that XVMC under 0.20 either doesn't work or uses too much CPU. Given your situation, 0.20 is not a good choice for you until this is fixed.

3. Of the various deinterlace methods available, bob2x deinterlacing will give the smoothest video playback for interlaced video. It is also the most processsor intensive method when used without XVMC. Finally, it is the only method that can be used in conjunction with XVMC, and when used with XVMC, requires very little processing power because the video card does a bunch of the work.

I suspect that if you remote logg'ed into your myth box while playing back that hidef video and ran "top -d 1", you would see that your CPU is pegged at full utilization.

In terms of xine being able to play back the video -- there are several possible explanations. If the only option you are using is "-V xvmc", then you are not deinterlacing, which saves a ton of CPU time.

So let me ask a few questions to see if I can help some more:

1. Under R5D1 / 0.19 (not 0.20), what is your CPU utilization while playing back hidef from the myth front end.

2. Same question as # 1 when you are playing back video with xine.

3. Are you deinterlacing when using xine? If not, and the results are acceptable, you may want to try video playback in the myth frontend w/ out deinterlacing.

4. What is you display device, what video resolution(s) does it support and how are u connecting to it (VGA, HDMI, etc)?

5. What is the resolution of your hidef recordings? Are they 720p, 1080i or some of each? If you playback the 720p recordings w/ out deinterlacing, do they play well from the myth front end?

Marc


Top
 Profile  
 
 Post subject:
PostPosted: Sun Nov 19, 2006 9:56 am 
Offline
Joined: Tue Jan 18, 2005 2:07 am
Posts: 1532
Location: California
A quick follow-up. Rather than running "top" to assess CPU utilization, run "vmstat 3". This will print out 1 line every 3 seconds with a general overview of how system resources are being used. Let it run for 30 seconds or so and then copy / paste the info into your reply. This will provide a bit more info for diagnostic purposes.

marc


Top
 Profile  
 
 Post subject:
PostPosted: Mon Nov 20, 2006 5:58 pm 
Offline
Joined: Wed Nov 16, 2005 8:55 pm
Posts: 1381
Location: Farmington, MI USA
Hi marc and misterflibble, thanks for your thoughts.

misterflibble wrote:
I've heard advice from some people that you have to disable the XVideo sync-to-vblank options in nvidia-settings. Have you tried this yet?


EDIT:
misterflibble wrote:
As far as modelines go, this site has a pretty comprehensive list of standard modelines for HDTV. You'll probably want ATSC 720p or 1080i at 59.94 hz (or both).


mrflibble, thanks for that URL. I had missed it when I initially responded, lots of information there (that I can hopefully make heads-or-tails of!).
/EDIT

Yes, I have tried it with sync to vblank enabled and disabled, same issue. I found that nVidia released the 1.0-9629 version, so I installed those and tried them, same problems.

marc.aronson wrote:
Here is some info you may find useful:

1. A 2.4ghz P4 w/out HT and w/out XVMC is not enough processing power to deal with hidef if you also need to deinterlace, which it appears you do need to do from your original post.

2. A few of the posts I've read about 0.20 in this forum suggest that XVMC under 0.20 either doesn't work or uses too much CPU. Given your situation, 0.20 is not a good choice for you until this is fixed.

3. Of the various deinterlace methods available, bob2x deinterlacing will give the smoothest video playback for interlaced video. It is also the most processsor intensive method when used without XVMC. Finally, it is the only method that can be used in conjunction with XVMC, and when used with XVMC, requires very little processing power because the video card does a bunch of the work.


Well, sounds like I'm pretty much hosed here, then! :wink:
I do have a P4 3.0Ghz/1Gb DDR2 RAM system that I currently use for digital video processing and other digital media. Maybe I will have to turn that into my main TV's Myth system... Dang! It has PCIe, not AGP. Wife's not going to be happy about another video card purchase :lol:

marc.aronson wrote:
I suspect that if you remote logg'ed into your myth box while playing back that hidef video and ran "top -d 1", you would see that your CPU is pegged at full utilization.


While not a complete 100%, CPU utilization is way up there:
Code:
procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
 r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id wa
 2  0      8  10048   3552 294504    0    0     6   197  127    59 12  3 85  0
 2  0      8   5968   3564 298584    0    0  1367    15 1789  4194 91  9  0  0
 2  0      8   6980   3568 297656    0    0  1367     0 1789  4332 90 10  0  0
 2  0      8   7888   3572 296596    0    0  1452     9 1796  4390 93  7  0  0
 2  0      8   8616   3572 295804    0    0  1367     4 1796  4298 93  7  0  0
 2  0      8   9916   3584 294604    0    0  1281     4 1793  4257 92  8  0  0
 1  0      8   6076   3596 298420    0    0  1281     7 1799  4108 90 10  0  0
 2  0      8   7096   3600 297360    0    0  1367     0 1779  4324 92  8  0  0


marc.aronson wrote:
In terms of xine being able to play back the video -- there are several possible explanations. If the only option you are using is "-V xvmc", then you are not deinterlacing, which saves a ton of CPU time.

So let me ask a few questions to see if I can help some more:

1. Under R5D1 / 0.19 (not 0.20), what is your CPU utilization while playing back hidef from the myth front end.


I will have to load up another drive with R5D1 and get back to you on this. I do recall running "top" earlier on a straight R5D1 install, and I believe CPU usage was hovering around 65-70% using the same options (Bob deint, XvMC, nVidia 1.0-8774 driver)

marc.aronson wrote:
2. Same question as # 1 when you are playing back video with xine.


While playing via Xine (Bob deint, "-V xvmc" option, no reboot between here and the last sample)
Code:
procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
 r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id wa
 0  0      8   6696   2200 356576    0    0     9   196  131    68 12  3 84  0
 0  0      8  10336   2176 352772    0    0  2007     0 1775  3235 49  3 48  0
 0  0      8   9496   2188 353684    0    0  2008    12 1774  3399 47  7 46  0
 0  0      8   8416   2212 354716    0    0  2013    60 1760  3604 48  7 45  0
 0  0      8   7152   2204 355912    0    0  2051     0 1776  3481 46  6 48  0
 0  0      8   6140   2212 356960    0    0  2049    13 1782  3531 46  3 51  0
 0  0      8  10092   2196 353016    0    0  2008     0 1794  2934 38  6 56  0
 0  0      8   8832   2192 354208    0    0  2092     4 1778  3637 43  7 50  0


marc.aronson wrote:
3. Are you deinterlacing when using xine? If not, and the results are acceptable, you may want to try video playback in the myth frontend w/ out deinterlacing.


Yes sir, I am using Bob in Xine as well. The stats above are with Bob, don't notice too much difference in either the picture quality or the CPU usage with Bob on or off. Which is quite different from what I experience in Myth - If I don't use deinterlace, I see a lot of "comb tooth" (not sure of the correct terminology here).

marc.aronson wrote:
4. What is you display device, what video resolution(s) does it support and how are u connecting to it (VGA, HDMI, etc)?


Right now I am hooked up to a Dell 1707FP via DVI watching a recording of the Michigan/Ohio State game through Xine, and it looks great! I want to ultimately hook this up to my Sony KP-51WS500 (16:9 widescreen, HD-ready) via DVI, which I have read should work.

marc.aronson wrote:
5. What is the resolution of your hidef recordings? Are they 720p, 1080i or some of each? If you playback the 720p recordings w/ out deinterlacing, do they play well from the myth front end?


Now there's a darn good question, and I'll be showing even more of my HD-ignorance - How do I know what resolution this is in? Is there a utility I can run against one of them that will let me know that? The recording I'm watching right now was on ABC, which I believe I've read is 720p.

Sounds like for right now I would be best off going back to R5D1/0.19 and trying to tweak that for optimal playback. I will load R5D1 on another disk, play with that and the settings I have found here on the forum, and see what I can work out. I will get the vmstat output then as well, maybe I can get that going tonight.

Again, I thank you all for helping me out!


Top
 Profile  
 
 Post subject:
PostPosted: Mon Nov 20, 2006 8:39 pm 
Offline
Joined: Wed Nov 16, 2005 8:55 pm
Posts: 1381
Location: Farmington, MI USA
Just wanted to let you know, I have loaded R5D1 and the nvidia 1.0-9629 drivers, set the Playback settings to use XvMC (deinterlace, Linear Blend - Default setting).

I have one HD channel that has no stutter at all, and only consumes ~ 65% of the CPU resources. As soon as I change the channel CPU usage jumps to ~93% and the stuttering returns, although it isn't quite as bad as with 0.20/1.0-9629.

So, I have a couple of questions:
Am I perhaps running into a signal issue? I don't have anything to compare against (we only have basic cable service), so I can't tell. All channels show up on the OSD as Signal 100%, S/N 3.3 - Does this mean something?

Might these stations be producing an image at different resolutions, and therefore requiring different "horsepower"? Is there a way to tell which channel broadcasts at which resolution?

Maybe it's the Fusion5 RT Lite? Again, I have nothing here to test that theory against, and after everything I've read on the forums I would say that's not the case...

Gotta love a mystery! Thanks all for helping me out with this.

EDIT: Just after submitting the above the channel that was playing perfectly went to commercial, and I got some stuttering. Not quite as bad as everything on the other channels, but stuttering just the same. So, my feeble old brain is trying to tell me that this is going to be some type of transmission/interference issue in my home, or over the cable. I will try some testing over the next couple of days, if anyone has any ideas please do shout out!

Thanks again


Top
 Profile  
 
 Post subject:
PostPosted: Mon Nov 20, 2006 9:41 pm 
Offline
Joined: Tue Jan 18, 2005 2:07 am
Posts: 1532
Location: California
OK -- lots of good info. Let's take it one step at a time.

1. If you look back at your "vmstat" listing while playing back through the 0.20 mythfrontend you will see that the column labled "ID" is always at 0 (except the first time). This means 0 idle time, which means your CPU is maxed out.

2. You vmstat when using "xine" shows idle time in the 40% - 50% range, so you have plently of spare CPU cycles in this scenario. Looks like xine is using xvmc properly => hardware acceleration of playback => less CPU drain.

3. When you did your 0.19 experiment you choose linear blend deinterlacing w/ XVMC. Not a good choice. The only deinterlace option that works with XVMC is bob2x, so try that and see how it goes. Bob2x is fast when used with xvmc, because when in xvmc mode, the video card shoulders some (most) of the processing burden. Try running vmstat when playing wiht the 0.19 frontend, xvmc enabled, bob2x deinterlace. See what your idle ("ID") time is.

4. If the original recording is in 720p, you don't need to deinterlace to get good playback results. You only need to deinterlace 1080i and 480i. I don't recall off the top of my head how to find the resolution of a recording, but you might try to run "mplayer" on it, as I think mplayer will at some point spit out the resolution to standard out.

5. What resolution are you driving your display at?

6. If xine is consistantly getting good results, there is a trick that can be done to use xine for playback from myth. It's a bit of a pain to set up, but once set up, works pretty well, although it does have some idosyncrocies. I used this technique when i was on R5A30.2 and the mythfrontend was unable to playback some hidef recordings. If we can't get the mythfrontend to work, I can provide you directons on how to configure this.


Marc


Top
 Profile  
 
 Post subject:
PostPosted: Mon Nov 20, 2006 9:49 pm 
Offline
Joined: Tue Jan 18, 2005 2:07 am
Posts: 1532
Location: California
One more comment. In re-reading your posts, I think you may be mis-interperting top. I suspect this because in one of your posts you stated:

Code:
While not a complete 100%, CPU utilization is way up there:
procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
 r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id wa
 2  0      8  10048   3552 294504    0    0     6   197  127    59 12  3 85  0
 2  0      8   5968   3564 298584    0    0  1367    15 1789  4194 91  9  0  0
 2  0      8   6980   3568 297656    0    0  1367     0 1789  4332 90 10  0  0
 2  0      8   7888   3572 296596    0    0  1452     9 1796  4390 93  7  0  0
 2  0      8   8616   3572 295804    0    0  1367     4 1796  4298 93  7  0  0
 2  0      8   9916   3584 294604    0    0  1281     4 1793  4257 92  8  0  0
 1  0      8   6076   3596 298420    0    0  1281     7 1799  4108 90 10  0  0
 2  0      8   7096   3600 297360    0    0  1367     0 1779  4324 92  8  0  0
  5968   3564 298584    0    0  1367    15 1789  4194 91  9  0  0
 


Actually, vmstat shows that your CPU utilization was at 100% because idle time (column "id") was at 0%. When looking at top's output, you need to look at the numbers on the third line that are to the left of "id" to understand how much free CPU time you have. This number is the percent of your cpu that is idle.

marc


Top
 Profile  
 
 Post subject:
PostPosted: Tue Nov 21, 2006 5:28 pm 
Offline
Joined: Wed Nov 16, 2005 8:55 pm
Posts: 1381
Location: Farmington, MI USA
Before we get into round 2 (or is it 3 now?) I would just like to thank you for all the input, Marc. Also, reading back through this I want to reiterate that I am now doing all this testing on a Celeron 2Ghz system with 512Mb RAM, eVGA nVidia 6200LE. I am not convinced the hardware is a problem, as playback in Xine with "-V xvmc" works perfectly, but just in case...

marc.aronson wrote:
OK -- lots of good info. Let's take it one step at a time.

1. If you look back at your "vmstat" listing while playing back through the 0.20 mythfrontend you will see that the column labled "ID" is always at 0 (except the first time). This means 0 idle time, which means your CPU is maxed out.


Yes, I was misreading this (as you pointed out in another post). Thanks for pointing that out.

marc.aronson wrote:
2. You vmstat when using "xine" shows idle time in the 40% - 50% range, so you have plently of spare CPU cycles in this scenario. Looks like xine is using xvmc properly => hardware acceleration of playback => less CPU drain.

3. When you did your 0.19 experiment you choose linear blend deinterlacing w/ XVMC. Not a good choice. The only deinterlace option that works with XVMC is bob2x, so try that and see how it goes. Bob2x is fast when used with xvmc, because when in xvmc mode, the video card shoulders some (most) of the processing burden. Try running vmstat when playing wiht the 0.19 frontend, xvmc enabled, bob2x deinterlace. See what your idle ("ID") time is.


Not any different from last night, I'm afraid. With 0.19, Bob deint, nVidia 9629 drivers:

Code:
procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
 r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id wa
 1  0  41480  96132   1040 187392    5    4    28   152  216   469  7  3 89  1
 2  0  41480  90560   1072 193036    0    0     0  2360 1897  5144 89 11  0  0
 1  0  41480  84992   1112 198540    0    0     1  1843 1870  5140 89 11  0  0
 2  0  41480  79412   1140 204056    0    0     0  2009 1884  5088 89 11  0  0
 1  0  41480  73708   1172 209700    0    0     0  1467 1886  5109 89 11  0  0
 1  0  41480  68004   1208 215340    0    0     0  1932 1879  5120 90 10  0  0
 1  0  41480  62176   1260 221096    0    0     0  2168 1870  5074 89 11  0  0
 1  0  41480  56724   1292 226608    0    0     0  1975 1876  5041 89 11  0  0
 1  0  41480  51020   1328 232248    0    0     0  2005 1883  5046 89 11  0  0


And still very choppy playback, although not nearly as bad as with myth 0.20.

Here's a new one: As I was writing this the screen froze during playback, then kicked me out of playback with "Error was encountered while displaying video", and my only option is "Return to Menu" at the bottom. I had not seen that before. Reading http://www.mythtv.org/wiki/index.php/Tr ... ying_video I may have changed channels too quickly, although I don't think so.

marc.aronson wrote:
4. If the original recording is in 720p, you don't need to deinterlace to get good playback results. You only need to deinterlace 1080i and 480i. I don't recall off the top of my head how to find the resolution of a recording, but you might try to run "mplayer" on it, as I think mplayer will at some point spit out the resolution to standard out.


OK, I think I see what you are after. I have the following after running mplayer against one of the "Live TV" mpegs:

VIDEO: MPEG2 1280x720 (aspect 3) 59.940 fps 19000.0 kbps (2375.0 kbyte/s)
(Mplayer (without options) also plays this without stuttering)

So, from what I've read I think that means it's 720 (?)

EDIT: Here's the vmstat while mplayer plays the mpeg:

Code:
procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
 r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id wa
 1  0  41476  78200   5448 320564    5    4    32   163  263   499  7  3 89  1
 1  0  41476  72560   5460 326228    0    0  1879     4 1872   715 75  8 17  0
 1  0  41476  66920   5484 331880    0    0  1880    16 1889   707 81  7 12  0
 1  0  41476  61280   5496 337544    0    0  1879     7 1881   710 77  8 15  0
 1  0  41476  55576   5508 343076    0    0  1879     4 1882   718 75  8 16  0
 1  0  41476  49936   5524 348736    0    0  1880     7 1876   702 80  7 13  0
 1  0  41476  44296   5536 354400    0    0  1879     4 1870   715 72  9 19  0
 0  0  41476  38656   5548 360064    0    0  1879     7 1868   718 74  7 19  0


Quite a bit higher CPU usage than Xine, but no stutter.
/EDIT

marc.aronson wrote:
5. What resolution are you driving your display at?


I'm not entirely sure. I am using the default XF86Config-4.nvidia-tvout.sample that came with R5D1, with the exception being "Option "ConnectedMonitor" "DFP"" in my Screen section for DVI output. I only see modes of 800x600 and 640x480 in the Screen section.

marc.aronson wrote:
6. If xine is consistantly getting good results, there is a trick that can be done to use xine for playback from myth. It's a bit of a pain to set up, but once set up, works pretty well, although it does have some idosyncrocies. I used this technique when i was on R5A30.2 and the mythfrontend was unable to playback some hidef recordings. If we can't get the mythfrontend to work, I can provide you directons on how to configure this.

Marc


If it turns out to be my only legitimate option, sure. I have seen some posts on the mythtv forum regarding high CPU usage as well, so I'm hoping some workaround or fix will come up.

I will continue hunting, and report any progress. Thanks again

Scott


Top
 Profile  
 
 Post subject:
PostPosted: Tue Nov 21, 2006 7:39 pm 
Offline
Joined: Wed Nov 16, 2005 8:55 pm
Posts: 1381
Location: Farmington, MI USA
I am attempting to put the hard drive I have been using for testing into my P4 - 3.0Ghz/1Gb RAM system to see how HDTV playback works. It has onboard Intel video (i945G chipset) with no AGP slot, so before I can test I have to install XFree86 4.6.0 to get the i810 drivers to install. Wish this old man luck!

EDIT: Not the way to go. Got mythfrontend running eventually, but any playback resulted in a blue box with no video but very good audio. At least there was no stuttering! :lol: Am now re-installing on test hard drive in P4 3.0Ghz system...


Top
 Profile  
 
 Post subject:
PostPosted: Tue Nov 21, 2006 11:42 pm 
Offline
Joined: Tue Jan 18, 2005 2:07 am
Posts: 1532
Location: California
Scott:

Sigh... It's clear that both xine & mplayer are capable of taking sufficient advantage of xvmc to play hidef, as both leave you with idle CPU. Not sure why the myth front end is using so much more processing power. A few things you might want to "poke" at in your quest to get this working:

1. Try looking at the OSD (On Screen Display) while playing back a video. When XVMC is active the OSD is in gray scale, so if it shows up in color, then at least you know that the problem is that somehow XVMC isn't really being activated.

2. Look in the file /var/log/XFree86.0.log. Look for lines towards the end that look something like "GetModeLine - hdsp: 1280 hbeg: 1392 hend: 1432 httl: 1648" followed by "vdsp: 720 vbeg: 725 vend: 730 vttl: 750 flags: 5". This tells me that my display is being driven at 1280x720 (hdsp x vdsp) resolution. What is your display resolution?

3. In setup->TV settings->playback make sure that you have: No custom filters; preferred MPEG2 decode of "Standard XVMC"; OpenGL vertical sync off; realtime priority off; use video as timebase off; extra audio buffering on; overscan settings all at 0;

That's all I can think of for now -- if I can think of any other things to look at I'll let you know. Best of luck and sorry I wasn't able to help you get your system running.

Marc


Top
 Profile  
 
 Post subject:
PostPosted: Wed Nov 22, 2006 5:33 pm 
Offline
Joined: Wed Nov 16, 2005 8:55 pm
Posts: 1381
Location: Farmington, MI USA
marc.aronson wrote:
1. Try looking at the OSD (On Screen Display) while playing back a video. When XVMC is active the OSD is in gray scale, so if it shows up in color, then at least you know that the problem is that somehow XVMC isn't really being activated.


Yep, made sure it was grey-scale after I had seen a number of posts referencing that.

marc.aronson wrote:
2. Look in the file /var/log/XFree86.0.log. Look for lines towards the end that look something like "GetModeLine - hdsp: 1280 hbeg: 1392 hend: 1432 httl: 1648" followed by "vdsp: 720 vbeg: 725 vend: 730 vttl: 750 flags: 5". This tells me that my display is being driven at 1280x720 (hdsp x vdsp) resolution. What is your display resolution?


I find this on the 0.19 install right now:
Code:
GetModeLine - hdsp: 800 hbeg: 816 hend: 896 httl: 1056
              vdsp: 600 vbeg: 601 vend: 604 vttl: 625 flags: 5


So it looks like 800x600, right? I am hooked up to the Dell 1707FP right now, which, when I run "nvidia-settings", tells me 800x600 as well. Maybe I should play around with different resolutions? I wasn't going to deal with that until I connected it to my TV, but what the heck...

marc.aronson wrote:
3. In setup->TV settings->playback make sure that you have: No custom filters; preferred MPEG2 decode of "Standard XVMC"; OpenGL vertical sync off; realtime priority off; use video as timebase off; extra audio buffering on; overscan settings all at 0;


Got 'em! If they would only work :wink:
I found this post http://www.mythtvtalk.com/forum/viewtopic.php?t=2509&start=0 that seems to indicate that it may be the video card itself (6200).

marc.aronson wrote:
That's all I can think of for now -- if I can think of any other things to look at I'll let you know. Best of luck and sorry I wasn't able to help you get your system running.

Marc


And again, your assistance is very much appreciated. If I happen to stumble on the solution(s) I will certainly post them. Off for more research, this thing hasn't seen the last of me!

Scott


Top
 Profile  
 
 Post subject:
PostPosted: Wed Nov 22, 2006 8:56 pm 
Offline
Joined: Tue Jan 18, 2005 2:07 am
Posts: 1532
Location: California
Scott, as an FYI I'm using the EVGA e-geforce 6200 card and it works fine. The only two differences I see on the video card front are:

1. Looks like you're using a "6200 LE" -- I'm using the "6200". I also owns an FX-5200 and I found that 6200 worked better for me than the FX5200.

2. You're using nvida driver version 9629 -- I'm using version 8774.

Not sure if any of this makes any difference, but thought I would provide it just in case. In terms of resolution -- I doubt that changing from 800x600 to another resolution will have an impact.

Marc


Top
 Profile  
 

Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 20 posts ] 
Go to page 1, 2  Next



All times are UTC - 6 hours




Who is online

Users browsing this forum: No registered users and 62 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Jump to:  
Powered by phpBB® Forum Software © phpBB Group

Theme Created By ceyhansuyu