Project

General

Profile

Forums » General »

Problem transcoding from same MUX (Different channels)

Added by Poul-Henning Hemberg almost 8 years ago

Hi All,

I'm experiencing something very odd. I have six DVB-C tuners (2xUSB 1xHDhomerun Expand) hooked up to an Ubuntu 14.04 LTS machine (8 Gb Ram, AMD FX-8300 Black CPU - 3.3 GHz - AMD AM3+ - 8 Core + SSD).

When transcoding to several XMBC's and/or VLC everything works fine as long as they are on separate PID (Tuners). BUT when two devices is on same PID (and two separate channels)the picture on the receiving devices starts to get really unstable (Starts, stops flickering, sometimes stall). It isn't a problem if they are set for the exact same channel only if they are on same PID/ different channel (The TVH software will use the same tuner). The obvious solution is to set a limit to one user pr. tuner. But I can't figure out how to do that.

I have tried to set a lot of different transcode settings but it didn't help. Then I thought that it may be something to do with the bandwidth of the tuners. So (just for testing) I ran the devices on strict PASS & HTS. And guess what. Everything ran smooth also different channels on same tuner (PID). So it has to be something in transcoding that goes wrong.

I believe it isn't a lack of CPU power as the transcoding runs fine as long as it isn't on same tuner with different channels.

I really need some help/input here folks.

Update: I have attached a debug log file with lots of warnings. (This is what happens when transcoding two different channels from same PID/Tuner/Frequency). The ones saying:
2017-02-08 13:24:56.664 [WARNING]:mpegts: too much queued table input data (over 2MB), discarding new
must be the problem. I also got lots of these:
2017-02-08 13:24:54.101 [WARNING]:libav: forced frame type (5) at 1287 was changed to frame type (3)

Please feel free to take a look a the debug file to find out what exactly is going on.

(THV version: HTS Tvheadend 4.1-2415~ge5f5a42)

debug.txt (19.2 KB) debug.txt

Replies (1)

RE: Problem transcoding from same PID/Tuner (Different channels) - Added by Poul-Henning Hemberg almost 8 years ago

Issue reported as a bug:

https://tvheadend.org/issues/4219

Poul-Henning Hemberg wrote:

Hi All,

I'm experiencing something very odd. I have six DVB-C tuners (2xUSB 1xHDhomerun Expand) hooked up to an Ubuntu 14.04 LTS machine (8 Gb Ram, AMD FX-8300 Black CPU - 3.3 GHz - AMD AM3+ - 8 Core + SSD).

When transcoding to several XMBC's and/or VLC everything works fine as long as they are on separate PID (Tuners). BUT when two devices is on same PID (and two separate channels)the picture on the receiving devices starts to get really unstable (Starts, stops flickering, sometimes stall). It isn't a problem if they are set for the exact same channel only if they are on same PID/ different channel (The TVH software will use the same tuner). The obvious solution is to set a limit to one user pr. tuner. But I can't figure out how to do that.

I have tried to set a lot of different transcode settings but it didn't help. Then I thought that it may be something to do with the bandwidth of the tuners. So (just for testing) I ran the devices on strict PASS & HTS. And guess what. Everything ran smooth also different channels on same tuner (PID). So it has to be something in transcoding that goes wrong.

I believe it isn't a lack of CPU power as the transcoding runs fine as long as it isn't on same tuner with different channels.

I really need some help/input here folks.

Update: I have attached a debug log file with lots of warnings. (This is what happens when transcoding two different channels from same PID/Tuner/Frequency). The ones saying:
2017-02-08 13:24:56.664 [WARNING]:mpegts: too much queued table input data (over 2MB), discarding new
must be the problem. I also got lots of these:
2017-02-08 13:24:54.101 [WARNING]:libav: forced frame type (5) at 1287 was changed to frame type (3)

Please feel free to take a look a the debug file to find out what exactly is going on.

(THV version: HTS Tvheadend 4.1-2415~ge5f5a42)

    (1-1/1)