Re: [mythtv-users] Request info from working M10K frontend.
Michael Carland wrote: On Apr 19, 2005, at 6:11 PM, Cecil Watson wrote: Hello, Michael Carland wrote: Howdy. ** Questions Could someone with a working M10K using built in video and ethernet, with a separate backend, provide me with the following info? What interrupts are your ethernet and video on? (cat /proc/interrupts) CPU0 15:1856705 XT-PIC uhci_hcd, eth0, [EMAIL PROTECTED]::01:00.0 CPU0 0: 185254 XT-PIC timer 1: 10 XT-PIC i8042 2: 0 XT-PIC cascade 5:120 XT-PIC uhci_hcd, uhci_hcd, yenta, VIA8233, eth0 6: 16802 XT-PIC uhci_hcd, ohci1394, yenta, ivtv0 7:285 XT-PIC parport0 8: 1 XT-PIC rtc 9: 0 XT-PIC acpi 12:110 XT-PIC i8042 14: 4790 XT-PIC ide0 15: 12 XT-PIC ide1 Interesting. I'm not able to get anything to use int 5, and you have eth0 there, which is what I would really like to try. I just tried turning on PNP and clearing the config in the BIOS, and trying ACPI. From the boot log, it looks like it will only route to interrupts 10, 11, and 15. Maybe I'll hop over the the viaarena forum, and see if they have any ideas. I notice you have yenta loaded, google tells me that's pcmcia related. Are you using an MII? Is your eth0 on pcmcia, or are you using built in? Yes, MII. Also, you have ivtv loaded, not too many ethernet interrupts, and quite a few ide0 interrupts. Is your system just a frontend, or is it both frontend and backend? I ask because when the nuv files are local, I don't see the problem. Both. I too am seeing some stuttering with 0.18 on the Epia. First it was due to DRM. Once I got that resolved, it would stutter when starting up. When I first included support for the Epia (Unichrome) in KnoppMyth, CPU usage was 50% when using the CD as a frontend, now it is 92%. Doing an install to hard drive uses the same amount. If I compile on the Epia or use my i686 debs, CPU usage is about 50%. Before, it was about 13%. So something is amiss, I just cannot put my hands on it... With XvMC and .18 on the frontend only Epia, I am seeing roughly 20% cpu usage, which makes this problem even stranger. About 3% cpu on hardware interrupts, 3% on software interrupts, and 0% wait. That's what's making me think network latency now, and checking out the shared video/ethernet interrupts situation. Now that I have a disk in there, and no longer need net boot, I'm going to try a nic in the pci, and see if I can get it on a different interrupt that way, just for giggles. Thanks for the info. You're welcome. -Michael Cecil ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Request info from working M10K frontend.
Title: Re: [mythtv-users] Request info from working M10K frontend. With XvMC and .18 on the frontend only Epia, I am seeing roughly 20% cpu usage, which makes this problem even stranger. About 3% cpu on hardware interrupts, 3% on software interrupts, and 0% wait. That's what's making me think network latency now, and checking out the shared video/ethernet interrupts situation. Now that I have a disk in there, and no longer need net boot, I'm going to try a nic in the pci, and see if I can get it on a different interrupt that way, just for giggles. I struggled for about 4 months to get this working. I created a diskless M10K based on FC2 and built the unichrome drivers, DRM and everything else. I built Myth 0.17 with XvMC and had pretty much the same problems. Lots of prebuffering and stuttering on playback. I also compiled xine with h/w decoding support and got that working quite well. I tried all sorts of stuff to get Myth working correctly including putting dual NICs in the B/E one dedicated to the F/E boot and Myth traffic. I rebuilt a minimal boot from scratch using the Linux from Scratch resources - same issues. I did network perf tests, bought a cable tester and found a couple of dodgy patch cables - still no joy. I put in a hard disk and found things got somewhat better (it would only stutter at startup) but then if I skipped forward or backwards in a stream I had to save the position and stop and restart the stream as it stuttered so badly afterward. sigh My M10K is now back in the back room on my bench. I've reverted to my physically larger Athlon XP 1800 + Geforce 4MX frontend in the family room which I rebuilt with FC3 and had up and running fine in about 4 hours. I'm not as happy with the quality of the video out as I was with the Epia's but it has never stuttered once (and that's without any h/w decoding on). I have tweaked it with the nvidia-settings tool which has improved it somewhat. Last night I did a clean Fedora install on the M10K (to a local disk) but I only got as far as an apt-get dist-upgrade before hitting the sack. I was planning to start from scratch with DRM, Unichrome X11, etc testing at every install to try and track down the issue. I bought the Epia specifically for a diskless, small form factor front end and so far it's caused only grief and a total lack of Wife Appreciation whilst I was trying to make it work. :( BTW - has anyone gotten the new VIA open source Unichrome driver to work with recent releases ? I had a very quick look but the docs in the distribution seemed a bit poor and I havent looked at them in depth. I'll post any progess I make so who knows, between us perhaps we can crack it :) Cheers Paul Important. Confidentiality: This communication is intended for the above-named person and may be confidential and/or legally privileged. Any opinions expressed in this communication are not necessarily those of the company. If it has come to you in error you must take no action based on it, nor must you copy or show it to anyone; please delete/destroy and inform the sender immediately. Monitoring/Viruses. Orange may monitor all incoming and outgoing emails in line with current legislation. Although we have taken steps to ensure that this email and attachments are free from any virus, we advise that in keeping with good computing practice the recipient should ensure they are actually virus free. Orange PCS Limited is a subsidiary of Orange SA and is registered in England No 2178917, with its address at St James Court, Great Park Road, Almondsbury Park, Bradley Stoke, Bristol BS32 4QJ. ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Request info from working M10K frontend.
On 4/20/05, BARKER, Paul [EMAIL PROTECTED] wrote: BTW - has anyone gotten the new VIA open source Unichrome driver to work with recent releases ? Works fine for me. :) I had a very quick look but the docs in the distribution seemed a bit poor and I haven't looked at them in depth. Which docs in particular? Perhaps either pointing out which bits are bad and need improving or making the improvements would help? Regards, -- Ivor http://www.ivor.it ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Request info from working M10K frontend.
Thanks to everyone for the replies. I'm going to try installing knoppmyth, and upgrading it to .18, since that is what the backend is now, and I had the same problem with .17. Since Jon has it working, and his interrupts look the same as mine, I can stop going down that path. I did try a pci nic, and it had a different interrupt than the video, but that did not solve the problem. Bummer. I sat squinting at the -v all logs into the wee hours last night, and came up with this puzzling thought: I assume the capture card has to capture at the rate the video source is clocked at, and in my case, the audio is sampled at 48kHz. On the front end, the video is written in sync with vertical retrace, and I assume the audio is written in sync with the video according to timing info in the stream. My question is, what happens if the video source rate is slightly slower, or the because of my video modeline, the output is slightly faster, than the standard. Or forget the standard, what if the source is slightly slower than the sink. I know the standard sets the display rate, but no two clocks are identical. In the case of live tv, would the ringbuffer slowly be drained, causing the pauses when it runs out? Or is myth able to compensate somehow? How about the audio, what happens if the frontends clock is slightly faster than the backend, is myth able to adjust the audio output rate, or does the audio buffer drain, causing pauses? What I see in the logs is a REQUEST_BLOCK _RI immediately after every Read(). At first the time of the Read() after the request is very short, but then bumps up to a slightly larger number. At this point, the audio buffer is slowing falling behind, until it runs out and pauses. When it restarts, there is a brief period of fast responses from the backend, but it then falls back to the old rate, the audio buffer again falling behind. I'm going to try and graph the response times, the audio buffer size, and the pauses, and see if I still see that pattern, now that I've had some sleep. Thanks, -Michael ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Request info from working M10K frontend.
BTW - has anyone gotten the new VIA open source Unichrome driver to work with recent releases ? Oops, I've just realised you mean VIA's new open source Unichrome driver release as opposed to the new Unichrome open source driver release. Doh. Must pay attention, must pay attention. This is going to get confusing, VIA Open Source Unichrome driver vs Open Source Unichrome VIA driver Ho hum. Anyway, in summary, no there's no point trying to use the VIA driver with myth since the VIA driver only provides their proprietary ddmpeg API. -- Ivor http://www.ivor.it ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Request info from working M10K frontend.
Title: Re: [mythtv-users] Request info from working M10K frontend. Anyway, in summary, no there's no point trying to use the VIA driver with myth since the VIA driver only provides their proprietary ddmpeg API. Ah. Thanks - that saved me some (pointless) work ! I'll concentrate on the other one. Cheers Paul Important. Confidentiality: This communication is intended for the above-named person and may be confidential and/or legally privileged. Any opinions expressed in this communication are not necessarily those of the company. If it has come to you in error you must take no action based on it, nor must you copy or show it to anyone; please delete/destroy and inform the sender immediately. Monitoring/Viruses. Orange may monitor all incoming and outgoing emails in line with current legislation. Although we have taken steps to ensure that this email and attachments are free from any virus, we advise that in keeping with good computing practice the recipient should ensure they are actually virus free. Orange PCS Limited is a subsidiary of Orange SA and is registered in England No 2178917, with its address at St James Court, Great Park Road, Almondsbury Park, Bradley Stoke, Bristol BS32 4QJ. ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Request info from working M10K frontend.
On Apr 19, 2005, at 4:50 PM, Michael Carland wrote: ** Questions Could someone with a working M10K using built in video and ethernet, with a separate backend, provide me with the following info? One more question I forgot. What nfs options are you using to mount the shared drive? From the documentation I am using: stern:/myth /myth nfs rsize=8192,wsize=8192,hard,intr,nfsvers=3 0 0 -Michael ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Request info from working M10K frontend.
Hello, Michael Carland wrote: Howdy. ** Questions Could someone with a working M10K using built in video and ethernet, with a separate backend, provide me with the following info? What interrupts are your ethernet and video on? (cat /proc/interrupts) CPU0 15:1856705 XT-PIC uhci_hcd, eth0, [EMAIL PROTECTED]::01:00.0 CPU0 0: 185254 XT-PIC timer 1: 10 XT-PIC i8042 2: 0 XT-PIC cascade 5:120 XT-PIC uhci_hcd, uhci_hcd, yenta, VIA8233, eth0 6: 16802 XT-PIC uhci_hcd, ohci1394, yenta, ivtv0 7:285 XT-PIC parport0 8: 1 XT-PIC rtc 9: 0 XT-PIC acpi 12:110 XT-PIC i8042 14: 4790 XT-PIC ide0 15: 12 XT-PIC ide1 NMI: 0 LOC: 0 ERR: 0 MIS: 0 What kernel version are you using? (I'm at 2.6.10) 2.6.7.11.7-chw-2 Are you using the VIA Rhine ethernet driver? The Use MMIO instead of PIO option? (I'm going to try turning off MMIO now) Yes, via-rhine. Any other suggestions? Thank, -Michael I too am seeing some stuttering with 0.18 on the Epia. First it was due to DRM. Once I got that resolved, it would stutter when starting up. When I first included support for the Epia (Unichrome) in KnoppMyth, CPU usage was 50% when using the CD as a frontend, now it is 92%. Doing an install to hard drive uses the same amount. If I compile on the Epia or use my i686 debs, CPU usage is about 50%. Before, it was about 13%. So something is amiss, I just cannot put my hands on it... Regards, Cecil ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Request info from working M10K frontend.
On Apr 19, 2005, at 6:11 PM, Cecil Watson wrote: Hello, Michael Carland wrote: Howdy. ** Questions Could someone with a working M10K using built in video and ethernet, with a separate backend, provide me with the following info? What interrupts are your ethernet and video on? (cat /proc/interrupts) CPU0 15:1856705 XT-PIC uhci_hcd, eth0, [EMAIL PROTECTED]::01:00.0 CPU0 0: 185254 XT-PIC timer 1: 10 XT-PIC i8042 2: 0 XT-PIC cascade 5:120 XT-PIC uhci_hcd, uhci_hcd, yenta, VIA8233, eth0 6: 16802 XT-PIC uhci_hcd, ohci1394, yenta, ivtv0 7:285 XT-PIC parport0 8: 1 XT-PIC rtc 9: 0 XT-PIC acpi 12:110 XT-PIC i8042 14: 4790 XT-PIC ide0 15: 12 XT-PIC ide1 Interesting. I'm not able to get anything to use int 5, and you have eth0 there, which is what I would really like to try. I just tried turning on PNP and clearing the config in the BIOS, and trying ACPI. From the boot log, it looks like it will only route to interrupts 10, 11, and 15. Maybe I'll hop over the the viaarena forum, and see if they have any ideas. I notice you have yenta loaded, google tells me that's pcmcia related. Are you using an MII? Is your eth0 on pcmcia, or are you using built in? Also, you have ivtv loaded, not too many ethernet interrupts, and quite a few ide0 interrupts. Is your system just a frontend, or is it both frontend and backend? I ask because when the nuv files are local, I don't see the problem. I too am seeing some stuttering with 0.18 on the Epia. First it was due to DRM. Once I got that resolved, it would stutter when starting up. When I first included support for the Epia (Unichrome) in KnoppMyth, CPU usage was 50% when using the CD as a frontend, now it is 92%. Doing an install to hard drive uses the same amount. If I compile on the Epia or use my i686 debs, CPU usage is about 50%. Before, it was about 13%. So something is amiss, I just cannot put my hands on it... With XvMC and .18 on the frontend only Epia, I am seeing roughly 20% cpu usage, which makes this problem even stranger. About 3% cpu on hardware interrupts, 3% on software interrupts, and 0% wait. That's what's making me think network latency now, and checking out the shared video/ethernet interrupts situation. Now that I have a disk in there, and no longer need net boot, I'm going to try a nic in the pci, and see if I can get it on a different interrupt that way, just for giggles. Thanks for the info. -Michael ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Request info from working M10K frontend.
On Apr 19, 2005, at 4:50 PM, Michael Carland wrote: What interrupts are your ethernet and video on? (cat /proc/interrupts) CPU0 0: 1301531405 XT-PIC timer 1: 9328 XT-PIC i8042 2: 0 XT-PIC cascade 5: 2 XT-PIC uhci_hcd, ohci1394 7: 0 XT-PIC parport0 8: 1 XT-PIC rtc 9: 0 XT-PIC acpi 10:2023409 XT-PIC uhci_hcd, VIA8233 11: 126010154 XT-PIC uhci_hcd, eth0, ehci_hcd, [EMAIL PROTECTED]:1:0:0 12: 66 XT-PIC i8042 14: 784337 XT-PIC ide0 NMI: 0 LOC: 0 ERR: 0 MIS: 0 What kernel version are you using? (I'm at 2.6.10) 2.6.9-chw-4 // this is from a hacked-up Knoppmyth install Are you using the VIA Rhine ethernet driver? The Use MMIO instead of PIO option? (I'm going to try turning off MMIO now) Yes, I'm using Via-Rhine, in the default config. I've been having some trouble with it on my backend (an Epox 8K9A7) but it seems to be fine on the frontend. Any other suggestions? Well, I'm booting from CF and mounting /usr and the MythVideo library over nfs, but streaming actual TV playback. Haven't had too much trouble, except for that pesky backend NIC issue and the occasional FC3 vs ivtv IRQ storm. The frontend has been pleasantly uneccentric. ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
