Media Player Classic - BE Win32/x64

[Socket774]

Hello.
This is a latest mplayerc.ja.rc.txt.
It's based on [r1910].
https://www.zeta-uploader.com/1221628463

[CruNcher]

@Alex, Aleksoid & Volt

Shouldn't it be doable to get the after Start (idle) CPU Utilization into the <0.01 range with endthreadex ?

https://msdn.microsoft.com/en-us/en-en/library/hw264s73.aspx

Currently 0.02 looks ok but not optimal.

Calling it shouldn't really cause as much as overhead as leaving it like it is now.

[Aleksoid1978]

@Alex, Aleksoid & Volt

Shouldn't it be doable to get the after Start (idle) CPU Utilization into the <0.01 range with endthreadex ?

https://msdn.microsoft.com/en-us/en-en/library/hw264s73.aspx

Currently 0.02 looks ok but not optimal.

Calling it shouldn't really cause as much as overhead as leaving it like it is now.

I don't understand what about you are talk :)

[CruNcher]

Im currently wondering what could cause this idle overhead on the CPU (Cycles leak)

http://i1.sendpic.org/t/uO/uOQR7hgEhBSl5lPI2j4fKiGlENG.jpg (http://sendpic.org/view/1/i/owN7cqd1tjcEsGcihad0FCzAD1w.png)

[Aleksoid1978]

0.02% is bad - you are joke ?? :)

[CruNcher]

Eh please read again i didn't say it's bad my wording was not optimal ;)

[Aleksoid1978]

Who say you that is not optimal ??

[CruNcher]

Nevermind i see now it's the Gui Rendering in the DWM context causing this slight overhead in the DWMCore through the SwapChain setup because of the D3D9 Gui Rendering so indeed it's pretty optimal in that Desktop Window Manager context.

http://www.facepalm.de/images/facepalm.jpg


But wait what for a sense makes this ?

http://i1.sendpic.org/t/u3/u35fCGhgDGnBhIT0tTGQ7jHDYLO.jpg (http://sendpic.org/view/1/i/k1XberPpVbEivsNZlwoSfAr6dr8.png)

http://i1.sendpic.org/t/65/653VhPsTNhEhvAjZ9DUrVgu9ZFW.jpg (http://sendpic.org/view/1/i/gMXCFinlsQ13gj5ef5WiO0rCOpz.png)

[aufkrawall]

(Good thing there's already a facepalm posted, but why is it not in 4k size as well?)

Would it be possible to extend MPC BE's YouTube capabilities also for livestreams?
It can't open livestreams with its internal filters. It works with external LAV Filters, but resolution is just 144p. Would be nice if livestreams could work with full resolution (with both internal and external filters).

[Aleksoid1978]

(Good thing there's already a facepalm posted, but why is it not in 4k size as well?)

Would it be possible to extend MPC BE's YouTube capabilities also for livestreams?
It can't open livestreams with its internal filters. It works with external LAV Filters, but resolution is just 144p. Would be nice if livestreams could work with full resolution (with both internal and external filters).

1 - Internal filters don't support http live streams.
2 - check latest SVN build - it's always open high resolution when opened Youtube live streams. Also - you can select different resolution/url via playlist or menu "Navigate->Jump to ...".

[v0lt]

@Socket774
Thanks. Updated in r1921.

[aufkrawall]

2 - check latest SVN build - it's always open high resolution when opened Youtube live streams. Also - you can select different resolution/url via playlist or menu "Navigate->Jump to ...".
Thank you, I was on an older build.
Seems to work fine, really valuable feature.

[P.J]

Possible to use DXVA scaler in EVR-CP mode?

[aufkrawall]

It seems YT streaming for 4k 60fps is not aggressive enough, decoding queue often starts to get empty after a few dozen seconds of playing and playback stalls.
Increasing buffers in MPC BE's splitter, LAV or madVR doesn't really help, playback will stall eventually.
It's definitely not my connection, can play this video in Edge without drops:
https://www.youtube.com/watch?v=aqz-KE-bpKQ

[Volfield]

Hello.
Updated mplayerc.pl.rc.txt r1910
https://www.zeta-uploader.com/2003469510

[v0lt]

@Volfield
The format of your file is not plain text.
Please go to this URL (https://sourceforge.net/p/mpcbe/code/HEAD/tree/trunk/src/apps/mpcresources/text/mplayerc.pl.rc.txt) and click on the label "Download this file". Download the file and work with it.

[Volfield]

Got error 404 on this link.

[v0lt]

Sourceforge.net have any problems now.
Try this link.

[Volfield]

Thanks.

EDIT:
New link: https://www.zeta-uploader.com/501460620
based on r1926

[v0lt]

@Volfield
Updated in r1928.
You can send links to the updated language files via private messages in this forum.

[Volfield]

Ok.I do that next time.

[kopija]

Greetings,
I remember that it was possible to increase buffers in MPC-BE, but now I
cannot find that option.
Has it been renamed to "jump distances" in Misc?
I have 5400rpm HDD so when jumping forward it takes sometimes five seconds for playback to resume.
How much should I increase "jump distances" to resolve this?
Thanks.

[v0lt]

Has it been renamed to "jump distances" in Misc?
No.
I have 5400rpm HDD so when jumping forward it takes sometimes five seconds for playback to resume.
How much should I increase "jump distances" to resolve this?
No.
On the contrary, I would suggest to reduce the size of the buffer.
Large buffer gives a high reading load after the jump, but it can smooth out the jerks, if data reading is unstable.

View->Options->Internal Filters->Source Filters->Buffer lenght
This works for built-in source filters and splitters only.

[kopija]

View->Options->Internal Filters->Source Filters->Buffer lenght
This works for built-in source filters and splitters only.

Thanks,
but could you give me specific numbers? I have no idea.
Buffer length increase by how much for my particular case?
And then should I decrease jump buffers after increasing buffer length?

[v0lt]

Thanks,
but could you give me specific numbers? I have no idea.
Use the value that gives your case the best result. The optimum value for a typical use is 3 seconds.

And then should I decrease jump buffers after increasing buffer length?
Our player has no jump buffers.

[aufkrawall]

With build 1945, YouTube 4k 60fps streaming is still not stable, MPC BE doesn't buffer enough/aggressively enough -> queues get empty.

Could you please also add support for VP9 livestreams when using LAV splitter? Some livestreams offer VP9, I can watch them in Firefox. But MPC BE with LAV splitter only offers H.264.

[v0lt]

With build 1945, YouTube 4k 60fps streaming is still not stable, MPC BE doesn't buffer enough/aggressively enough -> queues get empty.
Which version works better?
You can show the statistics (Ctrl+J )of the version for build 1945 and "better" build?

[aufkrawall]

Can you suggest which build I should try instead?
But I suspect there is no build that works better.

[v0lt]

Can you suggest which build I should try instead?
But I suspect there is no build that works better.
1. Try set View->Options->Internal Filters->Source Filters->Buffer lenght = 10 seconds.
2. Try v1.5.0 (build 1783) beta.

[aufkrawall]

No success:
http://abload.de/thumb/drops4usxk.jpg (http://abload.de/image.php?img=drops4usxk.jpg)

[Aleksoid1978]

No success:
http://abload.de/thumb/drops4usxk.jpg (http://abload.de/image.php?img=drops4usxk.jpg)

Not enough internet bandwidth.

[P.J]

Possible to use DXVA scaler in EVR-CP mode?

:confused:

[wanezhiling]

Possible to use DXVA scaler in EVR-CP mode?

it exists for more than one year...

http://i.imgur.com/6BywMO3.png

[aufkrawall]

Not enough internet bandwidth.
Nope, I got 150Mbit internet connection (20MB/s), can watch the VP9 stream without problems in Edge with GPU decoding.
MPC BE does hardly use my available bandwidth, it mostly buffers with 2.6MB/s, which clearly isn't enough.

What about an option to buffer/download as fast as possible?
This should solve the problem for good.

According to YT stats, buffer with Edge holds for ~18s.

[Volfield]

Does internal video decoder support HEVC DXVA?
I have Intel atom x5-z8300 with Intel(R) HD Graphics (8086:22B0) (DXVA checker: hevc_vld_main dxva2/d3d11 up to 8k) and has cpu decoding on that file:

Format : HEVC
Format/Info : High Efficiency Video Coding
Codec ID : V_MPEGH/ISO/HEVC
Duration : 24 min
Width : 1 280 pixels
Height : 720 pixels
Display aspect ratio : 16:9
Frame rate mode : Constant
Frame rate : 23.976 (24000/1001) FPS


When I play this file using Win10 app Movies and TV it use DXVA.

[v0lt]

What about an option to buffer/download as fast as possible?
This should solve the problem for good.
What exactly is an option to do?

[aufkrawall]

It would be like an unlimited buffer. Doesn't have to be kept in RAM, could also be cached on disk.
The main problem is: Program doesn't download fast enough. It most likely wouldn't download fast enough with the fastest internet connection in the world.

[v0lt]

It would be like an unlimited buffer.
Put in priority File source (URL) filter in External Filters.

[aufkrawall]

Doesn't help.

[P.J]

it exists for more than one year...

http://i.imgur.com/6BywMO3.png

:thanks: Wonder why it's missed in MPC-HC :rolleyes:

[vBm]

:thanks: Wonder why it's missed in MPC-HC :rolleyes:
And it's not... You're just not looking at the right place.

http://i.imgur.com/N39WNcx.png

[huhn]

there is no DXVA >resizer< in lavfilter.

the normal EVR in mpc-hc is using DXVA as a resizer.

EVR-CP doesn't have this option in mpc-hc unlike mpc-be.

[Mate94]

there is no DXVA >resizer< in lavfilter.

the normal EVR in mpc-hc is using DXVA as a resizer.

EVR-CP doesn't have this option in mpc-hc unlike mpc-be.

Do you think we are we going to see this option (EVR-CP with DXVA2) in MPC-HC anytime soon?

[jkauff]

Do you think we are we going to see this option (EVR-CP with DXVA2) in MPC-HC anytime soon?
MPC-HC nightlies have not been updated in almost three months. I don't think you'll be seeing the option in MPC-HC anytime soon.

[v0lt]

DXVA2 resizer works correctly for EVR-CP only on the Intel video cards. On other cards it is equivalent to bilinear interpolation.

[P.J]

Anyone know what kind of resizer methods do Nvidia/AMD/Intel use?

DXVA2 resizer works correctly for EVR-CP only on the Intel video cards. On other cards it is equivalent to bilinear interpolation.

hmmm

[Mate94]

DXVA2 resizer works correctly for EVR-CP only on the Intel video cards. On other cards it is equivalent to bilinear interpolation.

This is just wrong. The majority of the population is using iGPUs from Intel, so by your logic since it is only benefical for Intel GPU users, and not for all nVidia and AMD users as well, then there is no need for it.:mad:

By the way, I made a feature request ticket. Let's hope it is going to be evaluated positively.:D

[Aleksoid1978]

This is just wrong. The majority of the population is using iGPUs from Intel, so by your logic since it is only benefical for Intel GPU users, and not for all nVidia and AMD users as well, then there is no need for it.:mad:

By the way, I made a feature request ticket. Let's hope it is going to be evaluated positively.:D

You don't understand - current implementation of DXVA2 interpolation normal working ONLY on Intel IGPU. That's all - there is no discrimination.

[Mate94]

You don't understand - current implementation of DXVA2 interpolation normal working ONLY on Intel IGPU. That's all - there is no discrimination.

Ok, now I understand. Erm, no, I don't. But, it still doesn't make any sense. Is this a real problem? Why the aforementioned fact would stop anybody from implementing it?

It would work only for Intel iGPUs. Wonderful. Isn't it nice? It wouldn't work on nVidia or AMD cards. Then what? Those users would probably use madVR or the default EVR-CP renderer with Bicubic resizer.

I feel there has to be more behind your stance concerning this issue.

[kopija]

Greetings,
currently I am on 1.4.6 (build 1590), which is latest stable.
How come then I am getting prompt to upgrade to latest 1.5.0 (build 1783), which has beta tag?