Re: [mythtv-users] Re: HDTV, Myth, HD-2000, HD-3000, questions galore (and hopefully answers from the smart people... )
On Tue, Feb 08, 2005 at 12:31:13PM -0600, Paul Leppert wrote: > were dropped (if any)? > - What's a good way to determine cpu "idleness" during the test? I have been considering that mythcommflag might be a good benchmark to test systems. We would put out a couple of standard videos, a few minutes long, in 1080i and 720p that take lots of CPU. Start them playing and run mythcommflag on another standard video, and it prints out the frames/second it is getting in the spare CPU. That is a good measure of the spare CPU because it does it trying a task we actually want to use the spare CPU for. One could also try transcoding. Mythcommflag currently does a cumulative fps, it could be modified for an fps averaged over the last 5 seconds or something. There are two kinds of benchmarks. Synthetics are easy to do and run but may not tell you enough about real world results. Real benchmarks on code you want to run tell you a lot more. Unfortunately the machine idle time on hyperthreaded Pentiums is somewhat synthetic. The CPU says it has 60% idle CPU available but it could be lying. ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Re: HDTV, Myth, HD-2000, HD-3000, questions galore (and hopefully answers from the smart people... )
I believe the way this should work would be to playback the video as fast as possible. As soon as one frame is finished decoding, decode the next one. Take a time for each frame and then make sure that no frame takes longer than (1 / FPS) seconds to decode. If no frame takes longer than that to decode, you should have skip-free playback. Really all you'd need would be the time for the longest frame. Other factors would be: time to deinterlace a frame, time to decode the audio for one frame, etc. Add them all the times that you require and make sure you are under (1 / FPS) seconds. Brian ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Re: HDTV, Myth, HD-2000, HD-3000, questions galore (and hopefully answers from the smart people... )
On Sat, 29 Jan 2005 12:02:21 -0800, Brad Templeton <[EMAIL PROTECTED]> wrote: > > What we care about is the following though: > > a) Is there enough spare capacity so that when minor system tasks >take place, they don't make video skip. > > b) Is there enough capacity to do that if you also throw in a niced >thread to do commercial elimination or transcode? > > c) Do we have enough capacity to handle "peak" decoding load, since >that varies with scene complexity. > > And also > > d) What's the cheapest processor we can get away with and still play >skip-free 1080i and 720p? > Is it possible for someone who understands the issues around "peak" decoding load (scene complexity / changes) to either create a "benchmark" video(s) or at least define what it should contain (think "test clip" like in Avia or Video Essentials). We could then use that as a standard for playback during benchmarking. What would the requirements be? - Would we need one for 1080i and one for 720p? - How long should it be to get a good benchmark (duration=1 minute?)? - Can we then use that (those) videos as a standard for people to test with in mythtv? - Is it possible to play a video and get output on how many frames were dropped (if any)? - What's a good way to determine cpu "idleness" during the test? - Ideally we would want to standardize the testing and be able to provide simple output results (e.g., for such and such a configuration, running the standard benchmark resulted in x% frames dropped and average cpu utilization of y%) - Ideally, we would define a small number of different tests: 1. Play the video through mythtv while mythbackend is idle 2. Play the video through mythtv while mythbackend is recording (1 stream) 3. Play the video through mythtv while mythbackend is recording and transcoding Thoughts? Thanks, phlepper -- I hear and I forget. I see and I remember. I do and I understand. -- Confucius ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Re: HDTV, Myth, HD-2000, HD-3000, questions galore (and hopefully answers from the smart people... )
On Sat, Jan 29, 2005 at 02:02:57PM -0500, James Armstrong wrote: > Brad Templeton wrote: > > I am getting the following while watching HD 1080i on CBS: 38% User, > 2.2% System and 56.1% Idle > >.3% cpu usage for mythfrontend >mythbackend on another computer >12% for X >.3% for lirc_dev ><1% for top and others > > This is on a 3gig Pentium4 HT with 8meg cache. 1gig memory, NVidia 5200. Yes, as noted, the HT P4s reprot very good numbers, though there's been some recent debate off list about whether they are true or not, I'm still researching it. Hyperthreading, for those who don't know, means the P4 pretends to be a dual-processor system by duplicating the registers and various elements of the ALU, but not the entire CPU by any stretch. This means that for certain algorithms, ones that are ALU bound rather than memory bound, and which are multi-threaded, you can see a real improvement. There is speculation that mpeg encode and decode are in this category. But it makes it hard to figure out how "idle" the processor is because sometimes one of the units may be "idle" but in fact it could not really do anything. So we need to do some tests as follows: a) Try mpeg decode on a P4 with the non-SMP kernel. b) Get a benchmark program which can be run niced, to try to figure out how much spare capacity each different test system has while playing 1080i. Note that this just tells us how much spare capacity there is to run the benchmark program. What we care about is the following though: a) Is there enough spare capacity so that when minor system tasks take place, they don't make video skip. b) Is there enough capacity to do that if you also throw in a niced thread to do commercial elimination or transcode? c) Do we have enough capacity to handle "peak" decoding load, since that varies with scene complexity. And also d) What's the cheapest processor we can get away with and still play skip-free 1080i and 720p? xvmc of course can also change this equation a lot. On my P4, for example, playing 1080i in mplayer with xvmc, I have seen idle times over 80% which sounds good and implies one could get away with something much less in this environment. ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Re: HDTV, Myth, HD-2000, HD-3000, questions galore (and hopefully answers from the smart people... )
Your problem is this: > ... nVidia FX 6600, ... Evidently the latest released driver (66.93) doesn't properly support XVideo on the 6600's. Many complaints about this over at the nvnews.net forum. Kyle ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Re: HDTV, Myth, HD-2000, HD-3000, questions galore (and hopefully answers from the smart people... )
Brad Templeton wrote: On Thu, Jan 27, 2005 at 11:46:08PM -0600, Paul Leppert wrote: When watching livetv on ABC (HD 720p), I end up with 0% idle using top 72% cpu usage for mythfrontend 18% cpu for mythbackend 6% for X 1% for lirc_dev <1% for top and others When watching livetv on NBC (HD 1080i), I end up with 0% idle using top 82% cpu usage for mythfrontend 9% cpu for mythbackend 7% for X 1% for lirc_dev <1% for the rest These numbers seem much worse than what I've read from other people online. What are others seeing with HDTV? Should I be getting better I am getting the following while watching HD 1080i on CBS: 38% User, 2.2% System and 56.1% Idle .3% cpu usage for mythfrontend mythbackend on another computer 12% for X .3% for lirc_dev <1% for top and others This is on a 3gig Pentium4 HT with 8meg cache. 1gig memory, NVidia 5200. - James ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Re: HDTV, Myth, HD-2000, HD-3000, questions galore (and hopefully answers from the smart people... )
Brad, My memory bandwidth is 779 MB/s (according to memtest). I have 512MB of RAM, but this is probably pretty old memory (PC2100, I think). Do you think this is the problem? Does anyone have any comparison they can provide in terms of their performance with HDTV and an Athlon XP 3200+ (or even less)? I don't want to keep throwing new hardware money at the problem if it looks like I have a bottleneck somewhere in the system. Thanks, phlepper On Fri, 28 Jan 2005 00:45:38 -0800, Brad Templeton <[EMAIL PROTECTED]> wrote: > On Thu, Jan 27, 2005 at 11:46:08PM -0600, Paul Leppert wrote: > > When watching livetv on ABC (HD 720p), I end up with 0% idle using top > > 72% cpu usage for mythfrontend > > 18% cpu for mythbackend > > 6% for X > > 1% for lirc_dev > > <1% for top and others > > > > When watching livetv on NBC (HD 1080i), I end up with 0% idle using top > > 82% cpu usage for mythfrontend > > 9% cpu for mythbackend > > 7% for X > > 1% for lirc_dev > > <1% for the rest > > > > When watching recorded Fox (no recording), I end up with 10% idle using top > > 56% cpu usage for mythfrontend > > 24% for mythbackend > > 5% for X > > 1% for lirc_dev > > <1% for the rest > > > > When watching recorded NBC (no recording), I end up with 0% idle using top > > 85% cpu usage for mythfrontend > > 10% for X > > 2% for mythbackend > > 1% for lirc_dev > > <1% for the rest > > > > These numbers seem much worse than what I've read from other people > > online. What are others seeing with HDTV? Should I be getting better > > Yes, they are worse. Though actually my Athlon-3K system is currently > doing 100% saturation because I am going through more nvidia driver > problems and booting it without AGP, so I get saturation but all the CPU > is in the X server. > > Surprised to see much CPU in the back end, that I don't normally see > when it's recording. All it has to do is demux the atsc stream and write > to disk. Have not tried the 1.6 drivers, they reported it doesn't work > well on 2.6.10 in the release notes for that driver. > > Is it possible you have some sort of memory bottleneck? How's your > memory bandwidth? Try bw_mem in lmbench or memtest86. > -- I hear and I forget. I see and I remember. I do and I understand. -- Confucius ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Re: HDTV, Myth, HD-2000, HD-3000, questions galore (and hopefully answers from the smart people... )
On Fri, 28 Jan 2005 14:20:00 -0500, Tom E. Craddock, Jr. <[EMAIL PROTECTED]> wrote: > Could someone let me know how to commet out the parameters, and where > exactly to do it? I just had to reinstall FC2 and the kernel Im using > is 2.6.10, so Ive not been using it. Thanks for any help. > There are two files (I believe one is only used for the HD-2000 and the other for the HD-3000, but you need to change both to get the make to work). cx88-video.c and bttv-driver.c Just remove the second argument, so this: pci_save_state(pci_dev, dev->state.pci_cfg); and this pci_restore_state(pci_dev, dev->state.pci_cfg); become pci_save_state(pci_dev); and this pci_restore_state(pci_dev); You will need to change one each of pci_save_state and pci_restore_state in the above two files. You should then be able to execute the "make" without any problems. phlepper -- I hear and I forget. I see and I remember. I do and I understand. -- Confucius ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Re: HDTV, Myth, HD-2000, HD-3000, questions galore (and hopefully answers from the smart people... )
Paul Leppert wrote: Brad, On Fri, 28 Jan 2005 00:45:38 -0800, Brad Templeton <[EMAIL PROTECTED]> wrote: -snipped- Surprised to see much CPU in the back end, that I don't normally see when it's recording. All it has to do is demux the atsc stream and write to disk. Have not tried the 1.6 drivers, they reported it doesn't work well on 2.6.10 in the release notes for that driver. The readme for 1.6 says that it doesn't compile with 2.6.10 due to PCI changes. Basically, the pci_save_state and pci_restore_state calls in 2.6.10 have had the second (of two) parameters obsoleted. So, you just have to edit the two files (one for bttv and one for cx8800, don't remember offhand which ones, but they'll show up as errors when you do the make ;) to remove the second parameter. Other than that, is seems to be working fine for me (except of course, the performance issue I working through). I'll check my memory bandwidth momentarily. Thanks, phlepper ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users Could someone let me know how to commet out the parameters, and where exactly to do it? I just had to reinstall FC2 and the kernel Im using is 2.6.10, so Ive not been using it. Thanks for any help. Tom ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Re: HDTV, Myth, HD-2000, HD-3000, questions galore (and hopefully answers from the smart people... )
Brad, On Fri, 28 Jan 2005 00:45:38 -0800, Brad Templeton <[EMAIL PROTECTED]> wrote: -snipped- > Surprised to see much CPU in the back end, that I don't normally see > when it's recording. All it has to do is demux the atsc stream and write > to disk. Have not tried the 1.6 drivers, they reported it doesn't work > well on 2.6.10 in the release notes for that driver. The readme for 1.6 says that it doesn't compile with 2.6.10 due to PCI changes. Basically, the pci_save_state and pci_restore_state calls in 2.6.10 have had the second (of two) parameters obsoleted. So, you just have to edit the two files (one for bttv and one for cx8800, don't remember offhand which ones, but they'll show up as errors when you do the make ;) to remove the second parameter. Other than that, is seems to be working fine for me (except of course, the performance issue I working through). I'll check my memory bandwidth momentarily. Thanks, phlepper -- I hear and I forget. I see and I remember. I do and I understand. -- Confucius ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Re: HDTV, Myth, HD-2000, HD-3000, questions galore (and hopefully answers from the smart people... )
On Fri, 28 Jan 2005, Brad Templeton wrote: > Surprised to see much CPU in the back end, that I don't normally see > when it's recording. All it has to do is demux the atsc stream and write > to disk. Have not tried the 1.6 drivers, they reported it doesn't work > well on 2.6.10 in the release notes for that driver. If you want to try the 1.6 driver, I did the exercise to port it to 2.6.10-ac8. It works perfectly for me. If you want to try it, here's the pchdtv forum thread that has the instructions: http://pchdtv.com/forum/viewtopic.php?t=415 -Mike -- | Mike Isely | PGP fingerprint Spammers Die!! | | 03 54 43 4D 75 E5 CC 92 | isely @ pobox (dot) com | 71 16 01 E2 B5 F5 C1 E8 | | ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Re: HDTV, Myth, HD-2000, HD-3000, questions galore (and hopefully answers from the smart people... )
On Thu, Jan 27, 2005 at 11:46:08PM -0600, Paul Leppert wrote: > When watching livetv on ABC (HD 720p), I end up with 0% idle using top > 72% cpu usage for mythfrontend > 18% cpu for mythbackend > 6% for X > 1% for lirc_dev > <1% for top and others > > When watching livetv on NBC (HD 1080i), I end up with 0% idle using top > 82% cpu usage for mythfrontend > 9% cpu for mythbackend > 7% for X > 1% for lirc_dev > <1% for the rest > > When watching recorded Fox (no recording), I end up with 10% idle using top > 56% cpu usage for mythfrontend > 24% for mythbackend > 5% for X > 1% for lirc_dev > <1% for the rest > > When watching recorded NBC (no recording), I end up with 0% idle using top > 85% cpu usage for mythfrontend > 10% for X > 2% for mythbackend > 1% for lirc_dev > <1% for the rest > > These numbers seem much worse than what I've read from other people > online. What are others seeing with HDTV? Should I be getting better Yes, they are worse. Though actually my Athlon-3K system is currently doing 100% saturation because I am going through more nvidia driver problems and booting it without AGP, so I get saturation but all the CPU is in the X server. Surprised to see much CPU in the back end, that I don't normally see when it's recording. All it has to do is demux the atsc stream and write to disk. Have not tried the 1.6 drivers, they reported it doesn't work well on 2.6.10 in the release notes for that driver. Is it possible you have some sort of memory bottleneck? How's your memory bandwidth? Try bw_mem in lmbench or memtest86. ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Re: HDTV, Myth, HD-2000, HD-3000, questions galore (and hopefully answers from the smart people... )
Ok, I just updated my CPU from an Athlon XP 2000+ to an 3200+ and have some questions about performance with HDTV. First, my mythtv specs: Athlon XP 3200+ with 512Mb RAM, nVidia FX 6600, Nforce onboard LAN, SB Audigy 2 sound card, HD-3000 and PVR-250. /proc/cpuinfo confirms the 3200+ setting and a bogomips of 4341.76. I installed mythtv with Jarod's instructions using kernel 2.6.10 and pcHDTV v1.6. I'm using ALSA, /dev/dsp, KDE, and a resolution of 1024x768 (output to a VGA monitor). When watching livetv on ABC (HD 720p), I end up with 0% idle using top 72% cpu usage for mythfrontend 18% cpu for mythbackend 6% for X 1% for lirc_dev <1% for top and others When watching livetv on NBC (HD 1080i), I end up with 0% idle using top 82% cpu usage for mythfrontend 9% cpu for mythbackend 7% for X 1% for lirc_dev <1% for the rest When watching recorded Fox (no recording), I end up with 10% idle using top 56% cpu usage for mythfrontend 24% for mythbackend 5% for X 1% for lirc_dev <1% for the rest When watching recorded NBC (no recording), I end up with 0% idle using top 85% cpu usage for mythfrontend 10% for X 2% for mythbackend 1% for lirc_dev <1% for the rest These numbers seem much worse than what I've read from other people online. What are others seeing with HDTV? Should I be getting better performance from a 3200+? Is there something else I should be looking at to improve? Thanks, phlepper > However, right now, all stations use the same format 24/7. So if > you are watching ABC or Fox, it's 720p. If you are watching CBS or > NBC or UPN or WB it's 1080i. > -- I hear and I forget. I see and I remember. I do and I understand. -- Confucius ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Re: HDTV, Myth, HD-2000, HD-3000, questions galore (and hopefully answers from the smart people... )
On Mon, Jan 24, 2005 at 03:04:24PM -0600, Paul Leppert wrote: > Blammo, > - How are HDTV channels set up? Using zap2it and an HDTV lineup > resulted in a mix of NTSC and ATSC channels with incorrect freqid for > the digital channels (e.g., in Chicago, Channel 7 (ABC) is actually > broadcasting on 52). How is this set up via mythtv? The easiest way is mythweb. Select your digital TV lineup as a subset of the broadcast lineup on zap2it. Mythsetup allows slower setup of the frequencies. If you want to get fancy, write a script that takes the output of dtvscan (which came with your pchdtv) and builds the database. Subchannels are entered with a dash into myth though many sources write them with a dot. > > - How are HDTV subchannels set up? Again, based on above, ABC is > multicasting on 52. In mythtv, are the channels specified as freqid > 52-1, 52-2? or some other format? How should these multi-channels be > set up for names (mythtv channel input won't allow '.' (7.1, 7.2, > etc.) or '-' (7-1, 7-2). Do these channels need to be renumbered > (e.g., 900, 901)? For me, I cannot seem to get any other subchannel > other than the first regardless of what I put in the freqid field in Myth 0.17 will offer the ability to specify subchannels. Today, you must allocate everything to an integer channel. Thus I just use spare iteger channels In mythweb/settings_channels.php, I have, for example, Channel 4-2 (which broadcasts as 57-2), show up as "4" (channum in mythweb.) This is the "real" channel 4, that you get if you tune them analog. I have channel 4-1, which broadcasts on 57-1 appear on 6, which is a spare channel in the area. > - Based on http://thegeorges.us/mythtv instructions, the mythconverg > database needs to be tweaked to set channel format to ATSC (assuming a > default of NTSC). Is this necessary? What are the correct commands > for setting this up? Don't recall having to do anything but set atsc in mythtvsetup. > - If your mythtv setup receives OTA NTSC and ATSC, is there any reason > to keep the NTSC channels within mythtv? If so, how should these be > set up (similar issues as above since you now have two (or more) > channel 7's)? Is there a priority scheme within mythtv to make sure > it records a show off the ATSC stream before choosing the same show on > the NTSC stream? You can put priorities on channels but generally since all analog channels are now broadcsting in digital -- or is that not true where you live? -- why would you want to tune the analog. Anyway, I think that myth only recently got patches to support the analog tuning on the pchdtv anyway. I don't use it though. > - Does transcoding need to be set up in mythtv for HD channels (my > understanding of the HD stream is that it contains all subchannels. > Does mythtv filter out the "extra" subchannels or do you need to use > the transcoding mechanism to reduce the file sizes?) I have no idea > how any of this works in mythtv. Yes, myth extracts just the subchannel when you record. Transcoding could be used to make it smaller. Mythtv 0.16 does not handle anything but one mp2->mp4 transcode profile, so it is not really able to deal with the fact that some shows are 1080 lines, some 720, some 480, it would transcode them all to the same size. Manual transcode with mencoder is your best bet. > > And questions for performance: > > - What components of the mythtv / computer setup impact performance of > HDTV (e.g., at a given hardware level, what is the CPU usage for > recording? what is it for playback, what is it for both)? No cpu usage for recording that you would notice. TONS and TONS of cpu usage for playback, you want a 3ghz processor or even better. Also lots for transcode, commercial eliminate etc. Playback with xvmc takes less CPU but mythtv has issues with seeking in such recordings. You can playback in mplayer with xvmc, but you would go out of myth to do that. It's not out of the question to build an HDTV mythbox, for example, using a lower powered machine with an xvmc video card (nvidia for now) and be unable to play your videos in myth, but instead go to the shell and use the mythname.pl program to give the files nice names, and play them in mplayer. > - Related to above, what is the impact of multiple HDTV captures > (e.g., two HD-3000 cards) on the processing requirements of the PC? > Anybody here yet? Anybody have an idea of what it takes > processing-wise for HDTV PIP with two cards? No impact for recording. Why would you want PIP on a PVR, I have always wondered? It makes no sense to me. PIP is for live programming, to watch two things at once at least one of them live. You don't watch live on a PVR after using it for a while. I have not tried it but I would imagine dual-HDTV playback would take a lot of CPU. > - What can be done to improve performance (how is resolution reduced? > Is there any impact (or even capability) of recording at lower > resolution than 720p or
Re: [mythtv-users] Re: HDTV, Myth, HD-2000, HD-3000, questions galore (and hopefully answers from the smart people... )
Great list of questions. Hopefully we can get some answers to these, and I"ll happily collate them into the master. -blam On Mon, 24 Jan 2005 15:04:24 -0600, Paul Leppert <[EMAIL PROTECTED]> wrote: > Blammo, > > How about questions related to HDTV setup in mythtv? > > - How are HDTV channels set up? Using zap2it and an HDTV lineup > resulted in a mix of NTSC and ATSC channels with incorrect freqid for > the digital channels (e.g., in Chicago, Channel 7 (ABC) is actually > broadcasting on 52). How is this set up via mythtv? > > - How are HDTV subchannels set up? Again, based on above, ABC is > multicasting on 52. In mythtv, are the channels specified as freqid > 52-1, 52-2? or some other format? How should these multi-channels be > set up for names (mythtv channel input won't allow '.' (7.1, 7.2, > etc.) or '-' (7-1, 7-2). Do these channels need to be renumbered > (e.g., 900, 901)? For me, I cannot seem to get any other subchannel > other than the first regardless of what I put in the freqid field in > mythweb (using an HD-3000). > - Based on http://thegeorges.us/mythtv instructions, the mythconverg > database needs to be tweaked to set channel format to ATSC (assuming a > default of NTSC). Is this necessary? What are the correct commands > for setting this up? > - If your mythtv setup receives OTA NTSC and ATSC, is there any reason > to keep the NTSC channels within mythtv? If so, how should these be > set up (similar issues as above since you now have two (or more) > channel 7's)? Is there a priority scheme within mythtv to make sure > it records a show off the ATSC stream before choosing the same show on > the NTSC stream? > - Does transcoding need to be set up in mythtv for HD channels (my > understanding of the HD stream is that it contains all subchannels. > Does mythtv filter out the "extra" subchannels or do you need to use > the transcoding mechanism to reduce the file sizes?) I have no idea > how any of this works in mythtv. > > And questions for performance: > > - What components of the mythtv / computer setup impact performance of > HDTV (e.g., at a given hardware level, what is the CPU usage for > recording? what is it for playback, what is it for both)? > - Related to above, what is the impact of multiple HDTV captures > (e.g., two HD-3000 cards) on the processing requirements of the PC? > Anybody here yet? Anybody have an idea of what it takes > processing-wise for HDTV PIP with two cards? > - What can be done to improve performance (how is resolution reduced? > Is there any impact (or even capability) of recording at lower > resolution than 720p or 1080i/p? Is there an easy way to playback at > different resolutions (via mythtv rather than different modelines). > - If a modeline is set up for a particular monitor at a particular > resolution, does that automatically reduce the processing of an HDTV > stream (in other words, if I'm playing a 720p ATSC stream (1280x720) > back on a monitor at 480p (704x480), have I somehow reduced the > processing impact, or is the mythtv box still handling the full > processing impact and all that is happening is a reduction in quality > on the output)? > - For playback, what is the processing impact for different > resolutions (playing back a 720p stream at 1280 x 720, 1024 x 576, 800 > x 450)? > - For those trying to get the best performance (at the tradeoff of > quality), what is the best way to play back HDTV? Reducing playback > resolution? Transcoding to reduced resolution? etc. etc. > - What about audio? Are there general rules about digital audio and > performance or is this generally "not an issue" based on using a 5.1 > sound card (e.g., SB Audigy)? > - What is XvMC and how does / will that help with the processing > requirements of HDTV playback? Are there particular video cards this > will / will not work with? Anyone have real world examples of > processing improvements (e.g., CPU usage) with XvMC and without? > > And a general question: > > - Is there a way to tell within mythtv what format the ATSC is being > broadcast (720p, 1081i, etc.)? > > Wish I had more answers than questions, but as I get up the learning > curve (and I receive my new, faster CPU), I should be able to > contribute some real-world experience with my HD-3000 (and PVR-250 if > I can ever get them working together). > > Thanks, > Paul > -- > I hear and I forget. I see and I remember. I do and I understand. -- > Confucius > ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Re: HDTV, Myth, HD-2000, HD-3000, questions galore (and hopefully answers from the smart people... )
Blammo, How about questions related to HDTV setup in mythtv? - How are HDTV channels set up? Using zap2it and an HDTV lineup resulted in a mix of NTSC and ATSC channels with incorrect freqid for the digital channels (e.g., in Chicago, Channel 7 (ABC) is actually broadcasting on 52). How is this set up via mythtv? - How are HDTV subchannels set up? Again, based on above, ABC is multicasting on 52. In mythtv, are the channels specified as freqid 52-1, 52-2? or some other format? How should these multi-channels be set up for names (mythtv channel input won't allow '.' (7.1, 7.2, etc.) or '-' (7-1, 7-2). Do these channels need to be renumbered (e.g., 900, 901)? For me, I cannot seem to get any other subchannel other than the first regardless of what I put in the freqid field in mythweb (using an HD-3000). - Based on http://thegeorges.us/mythtv instructions, the mythconverg database needs to be tweaked to set channel format to ATSC (assuming a default of NTSC). Is this necessary? What are the correct commands for setting this up? - If your mythtv setup receives OTA NTSC and ATSC, is there any reason to keep the NTSC channels within mythtv? If so, how should these be set up (similar issues as above since you now have two (or more) channel 7's)? Is there a priority scheme within mythtv to make sure it records a show off the ATSC stream before choosing the same show on the NTSC stream? - Does transcoding need to be set up in mythtv for HD channels (my understanding of the HD stream is that it contains all subchannels. Does mythtv filter out the "extra" subchannels or do you need to use the transcoding mechanism to reduce the file sizes?) I have no idea how any of this works in mythtv. And questions for performance: - What components of the mythtv / computer setup impact performance of HDTV (e.g., at a given hardware level, what is the CPU usage for recording? what is it for playback, what is it for both)? - Related to above, what is the impact of multiple HDTV captures (e.g., two HD-3000 cards) on the processing requirements of the PC? Anybody here yet? Anybody have an idea of what it takes processing-wise for HDTV PIP with two cards? - What can be done to improve performance (how is resolution reduced? Is there any impact (or even capability) of recording at lower resolution than 720p or 1080i/p? Is there an easy way to playback at different resolutions (via mythtv rather than different modelines). - If a modeline is set up for a particular monitor at a particular resolution, does that automatically reduce the processing of an HDTV stream (in other words, if I'm playing a 720p ATSC stream (1280x720) back on a monitor at 480p (704x480), have I somehow reduced the processing impact, or is the mythtv box still handling the full processing impact and all that is happening is a reduction in quality on the output)? - For playback, what is the processing impact for different resolutions (playing back a 720p stream at 1280 x 720, 1024 x 576, 800 x 450)? - For those trying to get the best performance (at the tradeoff of quality), what is the best way to play back HDTV? Reducing playback resolution? Transcoding to reduced resolution? etc. etc. - What about audio? Are there general rules about digital audio and performance or is this generally "not an issue" based on using a 5.1 sound card (e.g., SB Audigy)? - What is XvMC and how does / will that help with the processing requirements of HDTV playback? Are there particular video cards this will / will not work with? Anyone have real world examples of processing improvements (e.g., CPU usage) with XvMC and without? And a general question: - Is there a way to tell within mythtv what format the ATSC is being broadcast (720p, 1081i, etc.)? Wish I had more answers than questions, but as I get up the learning curve (and I receive my new, faster CPU), I should be able to contribute some real-world experience with my HD-3000 (and PVR-250 if I can ever get them working together). Thanks, Paul -- I hear and I forget. I see and I remember. I do and I understand. -- Confucius ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Re: [mythtv-users] Re: HDTV, Myth, HD-2000, HD-3000, questions galore (and hopefully answers from the smart people... )
On Tue, 2005-01-18 at 15:21 -0700, Blammo [doh] wrote: > (thanks all for both your posted and offline feedback. Per > suggestions, I'm going to try to turn this into a FAQ. I'm not much of > a writer, so sorry in advance. Please reply with anything to have to > fill in the gaps, add catagories, etc. I'll try to keep this update, > and maybe find a Web home for it somewhere. ) You might also want to add Firewire HD input to the list. At least insofar that isn't currently supported in myth... -I ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
[mythtv-users] Re: HDTV, Myth, HD-2000, HD-3000, questions galore (and hopefully answers from the smart people... )
(thanks all for both your posted and offline feedback. Per suggestions, I'm going to try to turn this into a FAQ. I'm not much of a writer, so sorry in advance. Please reply with anything to have to fill in the gaps, add catagories, etc. I'll try to keep this update, and maybe find a Web home for it somewhere. ) FAQ: MythTV, HDTV, and you 1. Hardware 2. Playback 3. Operating System. 4. Overall tips / hints 5. URL's 1. Hardware: 1.1 Do the HD-2000 work better, or are better supported than the HD-3000 cards? HD-2000 cards have more stable drivers 1.2 Are there any other cards which do OTA ATSC HD recordings? 1.3 Which cards only work for DVB? Which only for OTA? 1.4 Which cards work with encrypted sources? 2. Playback: 2.1 How much CPU power is required to playback 1080i or 720p streams? Playback of 1080i / 720p streams takes about 40-50% of a P4 3gig w/o XvMC 2.2 Are most converting down to 540p for playback? What steps are involved in doing that? 3. OS: 3.1 What Kernel, IVTV/V4L/BTTV are recommended for: 3.1.1 HD-2000 2.6.9 seems to be the most commonly used, with pcHTDV 1.6 driver 3.1.2 HD-3000 3.1.3 Air2PC-ATSC-HDTV 3.1.4 PVR-x50 coexistant 4. Tips / Hints 4.1 From what I can glean, best way to succeed is to install the HD card into a different backend (if you already have a backend with PVR-XXX's in them) run a 2.6 kernel, and patch patch patch. 4.2 Thanks to all in advance for any contributions to this information.. 5. URL's: 5.1 Hardware 5.1.1 Hauppauge http://www.hauppauge.com/ 5.1.2 Air2PC http://www.cyberestore.com/product_info.php?cPath=28&products_id=10 5.1.3 5.2 Software 5.2.1 Linux DVB http://www.linuxtv.org/dvb/ 5.2.2 IVTV 5.2.3 BTTV 5.2.4 MythTV http://mythtv.org/ 5.3 Receptioni 5.3.1 Local HDTV Info http://www.avsforum.com/avs-vb/forumdisplay.php?s=dfa53122cfc95344b2de8648e38c9379&forumid=45 ___ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
