[mythtv-users] Intel i965 video buffers errors on frontend - Live TV?

[evade.]

On 12/07/17 03:47, faginbagin wrote:
>> On 7/10/2017 6:06 AM, evade. wrote:
>>> On 07/07/17 18:22, faginbagin wrote:
>>>> On 7/6/2017 7:35 AM, evade at internode.on.net wrote:
>>>>> Hello,
>>>>> I'm having difficult problems with playback in the frontend version 
>>>>> 0.28.1-3 and would love some help please!
>>>>>
>>>>>
>>>  >> <snip>
>>>>>
>>>>> So I switched to VAAPI, despite it lacking deinterlacing.  I'm 
>>>>> still having playback problems with stuttering, especially with 
>>>>> live TV. My PC also has Intel HD audio (Realtek ALC892).
>>>>>
>>>>> Although playback works, when I enable debug logging I see multiple 
>>>>> occurrences of both:
>>>>>
>>>>> <date> <time> I  Player(0): Waiting for video buffers...
>>>>>
>>>>> and:
>>>>>
>>>>> <date> <time> I  Player(0): Video is 3.54437 frames ahead of audio,
>>>>>                          doubling video frame interval to slow down.
>>>>>
>>>>> Relevant devices:
>>>>> Intel H170M chipset, Intel i5-6500 CPU with HD Graphics 530
>>>>> Intel "HD Audio" (Realtek ALC892)
>>>  >
>>>>
>>>> I suggest trying an OpenGL playback profile. That's what I'm using 
>>>> on mythtv 0.27 with an i3-3220. My CPU with HD graphics 2500 should 
>>>> be far less capable than your i5-6500 with HD graphics 530. FWIW, I 
>>>> had to create a custom playback profile that I called "OpenGL 
>>>> Slimmer" because I needed to change the primary deinterlacer from 
>>>> Linear Blend (2X, HW) to Kernel and the fallback from Linear Blend 
>>>> (HW) to One field. Looks fine on a 32 inch TV in the bedroom. I 
>>>> would imagine your system can use a better deinterlacer.
>>>>
>>>> HTH,
>>>> Helen
>>>  >
>>>
>>> Thanks for the fast response Helen!
>>>
>>> Your answer was helpful as I didn't realise that 2D video playback 
>>> could be so taxing (I guess it's the deinterlacer?).
>>>
>>> You gave me the idea to increase the minimum video memory to the 
>>> graphics card in the BIOS.  It's now 128 MiB, although that doesn't 
>>> seem to have changed anything.
>>>
>>>
>>> I've just replied to another post and explained that I purchased the 
>>> i5 for my media PC because I hope to run two small VMs on it at the 
>>> same time. (They use different storage)
>>> I had hoped the GPU would easily be sufficient for 2D video playback 
>>> at 1080p and I'd really like to avoid software rendering.  Aren't the 
>>> OpenGL playback profiles using software?
>>>
>>>
>>> Hmm.  I just tried the high quality OpenGL playback profile and it 
>>> has the same problem with my buffer being 0% when I first change 
>>> channel. I could skimp lower, but I will be watching this on our main 
>>> 55 inch TV.
>>>
>>> Any other ideas, please?
>>>
>>> Thank you,
>>> evade.
>>
>> I didn't realize your problems were while watching Live TV. Could it 
>> be that Live TV is the problem and not which playback profile you're 
>> using? MythTV's Live TV functionality has always had a bad reputation. 
>> If I were you, I would first use recordings to test and configure the 
>> optimal playback profile, and only after you've got that working to 
>> your satisfaction, test it with Live TV. And you may just have to 
>> accept the fact you've got to pause for a few seconds to get Live TV 
>> to work reasonably well.
>>
>> To give you some points of comparison:
>> - My CPU is an i3-3220 (2 physical cores with hyper-threading)
>> - The MB has a B75 chipset
>> - I have changed the "Internal Graphics Memory Size" in the BIOS from 
>> the default 64M to the max 1024M.
>> - The machine has a single 8GB RAM stick.
>> - I have no SSDs, there's one HDD for the OS, database and 8GB swap 
>> (which I've never observed being used). One HDD for recordings and 2 
>> for "Watch Videos" content.
>> - I have often had up to 4 simultaneous recordings, with commercial 
>> flagging jobs going, along with transcoding via Handbrake, and still 
>> had smooth playback of 1080i mpeg2 recordings.
>> - I don't see an issue with software rendering, what I care about are 
>> the results.
>>
>> -  I don't have access to H.264 1080i content, but when I playback a 
>> 1080i mpeg2 recording (nothing else going on) my Playback Data screen 
>> reads numbers like:
>> Storage to Buffer:    >1Gbps
>> Buffer to Decoder:    3.1 Mbps
>> Available Buffer:    99% of 4Mb
>> Video:              1920x1080 at 29.97fps
>> Codec/Dec:            MPEG-2 ffmpeg
>> CPUs:               15% 13% 25% 15%
>> FPS:                30.01+-0.00
>> A/V Sync:           -0.05
>> Frames decoded/free:    29/1
>> Audio:              AC3
>> Channels:           6
>> Sample Rate:        48 GHz
>>
>> -When I playback a 1080p H.264 Bluray rip (nothing else going on), my 
>> Playback Data screen reads numbers like:
>> Storage to Buffer:    >1Gbps
>> Buffer to Decoder:    24.5 Mbps
>> Available Buffer:    99% of 4Mb
>> Video:              1920x1080 at 23.98fps
>> Codec/Dec:            H.264 ffmpeg
>> CPUs:               10% 21% 16% 12%
>> FPS:                23.80+-0.00
>> A/V Sync:           -0.08
>> Frames decoded/free:    25/1
>> Audio:              DTS
>> Channels:           6
>> Sample Rate:        48 GHz
>>
>> You've got a 55 inch TV, so you'll probably notice artifacts I 
>> wouldn't on my 32 inch TV with OpenGL. Maybe VAAPI will help, maybe 
>> not. MythTV's OpenGL support provides more deinterlacing options than 
>> VAAPI and some are supposedly hardware based, so I suspect you can get 
>> better results with OpenGL.
>>
>> HTH,
>> Helen
> 
(Resend to list as I accidentally replied directly to Helen)

Thanks Helen,

You're totally right that by trying to diagnose live TV playback and 
general playback problems at the same time that I have confused things!

Thankfully, as I've written in another reply to Stuart, increasing video 
memory to 512 MiB in the BIOS seems to have resolved the playback issue, 
even with bluray rips.

Thank you for those stats, it's very helpful for me to have a reference!

You're right about VAAPI.  If a previous post of Stuart's is still 
accurate (https://lists.gt.net/mythtv/users/603885), the lack of decent 
deinterlacer could be a real issue.  I'll keep exploring playback profiles.

Thanks,
evade.