Posted: February 03 2005 at 9:46am | IP Logged

I work with equipment that uses Osprey 230 boards on a daily basis.  Our goal is to capture video 24/7 but in a number of cases, we have run into different problems with different systems when video stops encoding.  Most of the time I can make it past several days before something dies.

I will list what I have come across and some of the things I have found that might help some people here.  Hopefully you guys can help me out too.  Browsing this forum, I've noticed there are similar threads (usually with 0 replies) so I'm going to bring up these again.

First off, the equipment encodes with WME having reasonable settings for PAL and NTSC:  20 sec between keyframes, Smoothness 10, Decoder complexity Simple, Buffer size is Use Default.  This keeps our CPU usage low while encoding otherwise WMEnc will immediately crash.

What we have noticed and learned: 

1) Low CPU usage is good while encoding.  At ~80% or higher, there's greater risk of failure.  I aim for ~40%

2) If the screen refresh rate is changed, wmenc can fail.

3) If you install UltraVNC on the system and try to connect it with a client machine, wmenc can fail.  This happens only when the Video Hook driver is enabled.

4) So far I noticed this on PAL but sometimes the encoded ASF video can get jittery and jumpy.  It speeds up a bit, frames are dropped, audio gets jumpy.  Then video slowly returns back to normal but audio stream is gone.  No more audio is encoded until the encoder process is restarted.  Seemed like there may not have been enough system resources.  I increasing the RAM on the system (originally 512MB with 2 Osprey encoders, now boosted to 1.5GB) and it might be fine now.

Now what I cannot figure out are the following problems that I have come across.  NOTE: I don't get this on all my encoding systems... just some of them and I can't seem to pinpoint what could be causing it.  Windows registry settings?  hardware?  Each of these systems are imaged from the same CD so they should be exactly the same in hardware and software.

1) WMEnc still fails unexpectedly regardless of satisfying the above points.

2) With a 2 card setup, sometimes just one wmenc will fail while the other keeps on going as though there's nothing wrong.  Wmenc process might hang... or fail peacefully by reporting errors in the Windows Event viewer

3) Windows Event viewer reports "Faulting application wmenc.exe, version 9.0.0.2980, faulting module wmadmoe.dll, version 9.0.0.2980, fault address 0x0001e4d6"

4) "Faulting application wmenc....." except this time in wmvdmoe2.dll

5) Sometimes encoded ASF files made by wmenc are partially corrupted.  Attempting to join these ASF files together or extracting portions of it will fail due to the corrupted portions of the ASF.

All encoding systems are running Windows 2000 with v2.2.2.0 of the driver.

Is there any possibility that there might be something in the video signal -- not standard-- that could cause the Osprey to output bad data to wmenc, and thus causing wmenc to die?  Could a bad Osprey card output bad data?

I'm just throwing some ideas around so maybe someone can help me out or give some opinions.

Thanks!

Dracore