MVallevand wrote: > I've spent a lot of time analyzing this problem over the last year > with emulation mode and the problem isn't restricted the the H3's. I > think the problem is most likely the NIC itself. There isn't a lot of > choice in PCI-less ethernet and the Dreambox code from tuxbox.org > uses the same driver which I actually used to set the driver to full > duplex.
Ah, very interesting. I presume this Ethernet driver follows a typical model of responding to an interrupt. Have you looked over its interrupt handler to see if it is attempting to do anything that might be rather time consuming on the MVP? Have you checked to see if there are other interrupt handlers (that are masking interrupts) that might be sluggish under some circumstances? I haven't proven this with any tests yet, but I get the impression that the overruns are occurring not during video playback, but during MythTV protocol commands. Are there some diagnostics to we can add into the code to get a better idea of where the time is being spent? > The changes I made for the --rtwin and --fs-rtwin give me the most > control over the problem. I can cut the errors to zero... By making the buffer small, right? That's because the flow control is then doing its job. > ...but with this I can end up with stuttering so for video I like 8192... And yet that probably produces the overrun errors. But did they happen during video playback or at other times? Perhaps a buffer size in between might be the thing to try. It's hard to say whether the software stack is under performing by not being able to handle a sustained stream, as we don't have any definitive benchmarks for the hardware. (Or do we? Anyone hack the original Hauppauge code, run bandwidth tests and check for overrun errors?) But when it is accepting streamed video, it seems to play fine (overruns should appear as glitches in playback), suggesting the hardware and Ethernet driver are doing their job correctly. > With 4096 I have streamed 3 days of internet audio with no errors and > I don't see it growing. But that's "lowing the bar" and avoiding the underlying problem, if 4096 in an inadequately large buffer to handle video. So on a related note, in your testing, have you observed any of the sluggish UI behavior I described? I'm still trying to figure out if it is directly or indirectly related to the overruns. -Tom ------------------------------------------------------------------------- This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/ _______________________________________________ Mvpmc-users mailing list Mvpmc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mvpmc-users mvpmc wiki: http://mvpmc.wikispaces.com/
