Mailing List Archive

judder problems with nvidia 358.16
As the original message got caught in the moderation filter, I'm reformatting it with more taken out to pastebin.

 To me, the likeliest candidates for the problem look like Compositing still being enebled somewhere, despite being disabled in xorg.conf and the xfce desktop configuration??

incorrect rtc permissions, although I don't see this fixing the incorrect refresh rates being consistently chosen.

All feedback and tips gratefully received!

 Thanks!

Marius


On Thursday, March 3, 2016 20:40 CET, "Marius Schrecker" <marius.schrecker@lyse.net> wrote:
  I have followed the judder-free wiki page with
Separate Video modes for GUI and TV Playback

both set to 1920x1080, auto rate, default aspect.

I tried downgrading to nvidia 352.x but had problems compiling /installing against my 4.4.2 kernel. The pre-packaged versions gave me glx problems, so ended up with 358.16, which was the lowest version that built and installed. If this is the likeliest problem I will obviously downgrade the kernel.

Here is an example 25fps video

I was going to add further examples, but refresh is wrong for all framerates I have tried and the mail is getting long.

Compositing is disabled in xorg.conf and in xfce window manager tweaks, although mplayer complains:


 output from mplayer:
VIDEO:  720x304  25.000 fps    0.0 kbps ( 0.0 kB/s)
Aspect ratio is 2.40:1 - scaling to correct movie aspect.
VO: [vdpau] 720x304 => 729x304 Planar YV12 
[vdpau] Compositing window manager detected. Assuming timing info is inaccurate.
A:   8.0 V:   8.1 A-V: -0.004 ct:  0.000   0/  0  1%  1%  0.7% 0 0 
No bind found for key 'Ctrl+c'.
A:  16.0 V:  16.0 A-V:  0.000 ct:  0.000   0/  0  0%  1%  0.7% 0 0 



mythuser@mythbox:/var/lib/mythtv/MOVIES$ xrandr
Screen 0: minimum 8 x 8, current 1920 x 1080, maximum 16384 x 16384
DVI-I-0 disconnected (normal left inverted right x axis y axis)
VGA-0 disconnected (normal left inverted right x axis y axis)
DVI-I-1 disconnected (normal left inverted right x axis y axis)
HDMI-0 connected 1920x1080+0+0 (normal left inverted right x axis y axis) 1600mm x 900mm
   1920x1080     60.00 +  59.94    50.00    29.97    25.00    23.97*   60.05    60.00    50.04

In MythTV I have set up separate GUI and playback modes... as instructed in the judder free wiki page and am using the vdpau high quality profile

Mythfrontend log (verbose)
http://pastebin.com/9hxFm6QU

Note that the refreshrate is not matching the framerate and that frames are being dropped. Also note the permission denied on the rtc

$ xrandr
Screen 0: minimum 8 x 8, current 1920 x 1080, maximum 16384 x 16384
DVI-I-0 disconnected (normal left inverted right x axis y axis)
VGA-0 disconnected (normal left inverted right x axis y axis)
DVI-I-1 disconnected (normal left inverted right x axis y axis)
HDMI-0 connected 1920x1080+0+0 (normal left inverted right x axis y axis) 1600mm x 900mm
   1920x1080     60.00 +  59.94    50.00    29.97    25.00    23.97    60.05    60.00    50.04* 
   1360x768      60.02  
   1280x1024     60.02  
   1280x720      60.00    59.94    50.00  
   1152x864      60.00  
   1024x768      60.00  
   800x600       60.32  
   720x576       50.00    50.08  
   720x480       59.94  
   640x480       59.95    59.94    59.93 

$ cat xorg.conf

 http://pastebin.com/zkm2qjmH


vdpau info:

http://pastebin.com/YqvzX6ea
  

Nvidia module:
# modinfo nvidia
filename:       /lib/modules/4.4.2/kernel/drivers/video/nvidia.ko
alias:          char-major-195-*
version:        358.16
supported:      external
license:        NVIDIA
srcversion:     38681B6CCC0B032F48069CF
alias:          pci:v000010DEd00000E00sv*sd*bc04sc80i00*
alias:          pci:v000010DEd*sv*sd*bc03sc02i00*
alias:          pci:v000010DEd*sv*sd*bc03sc00i00*
depends:        drm
vermagic:       4.4.2 SMP mod_unload modversions 
parm:           NVreg_Mobile:int
parm:           NVreg_ResmanDebugLevel:int
parm:           NVreg_RmLogonRC:int
parm:           NVreg_ModifyDeviceFiles:int
parm:           NVreg_DeviceFileUID:int
parm:           NVreg_DeviceFileGID:int
parm:           NVreg_DeviceFileMode:int
parm:           NVreg_UpdateMemoryTypes:int
parm:           NVreg_InitializeSystemMemoryAllocations:int
parm:           NVreg_UsePageAttributeTable:int
parm:           NVreg_MapRegistersEarly:int
parm:           NVreg_RegisterForACPIEvents:int
parm:           NVreg_CheckPCIConfigSpace:int
parm:           NVreg_EnablePCIeGen3:int
parm:           NVreg_EnableMSI:int
parm:           NVreg_TCEBypassMode:int
parm:           NVreg_MemoryPoolSize:int
parm:           NVreg_RegistryDwords:charp
parm:           NVreg_RmMsg:charp
parm:           NVreg_AssignGpus:charp

$ nvidia-smi -q | grep : | grep -v N/A
Timestamp                           : Thu Mar  3 19:36:14 2016
Driver Version                      : 358.16
Attached GPUs                       : 1
GPU 0000:01:00.0
    Product Name                    : GeForce GT 610
    Product Brand                   : GeForce
    Persistence Mode                : Disabled
    GPU UUID                        : GPU-bda18178-2424-4734-8dd3-d2915077b96f
    Minor Number                    : 0
    VBIOS Version                   : 75.19.55.00.02
        Bus                         : 0x01
        Device                      : 0x00
        Domain                      : 0x0000
        Device Id                   : 0x104A10DE
        Bus Id                      : 0000:01:00.0
        Sub System Id               : 0x840D1043
    Performance State               : P8
        Total                       : 1023 MiB
        Used                        : 42 MiB
        Free                        : 981 MiB
    Compute Mode                    : Default
        GPU Current Temp            : 30 C


glx info:

http://pastebin.com/u9vMxFyF

 
Re: judder problems with nvidia 358.16 [ In reply to ]
On 3/4/16 5:10 AM, Marius Schrecker wrote:
> incorrect rtc permissions, although I don't see this fixing the
> incorrect refresh rates being consistently chosen.


I'm no expert on this stuff, but these look suspicious:

Mar 3 20:13:32 mythbox mythfrontend.real: mythfrontend[22207]: E
CoreContext vsync.cpp:275 (TryInit) VSYNC: DRMVideoSync: VBlank ioctl
did not work, unimplemented in this driver?
Mar 3 20:13:32 mythbox mythfrontend.real: mythfrontend[22207]: E
CoreContext vsync.cpp:356 (TryInit) VSYNC: RTCVideoSync: Could not open
/dev/rtc: #012#011#011#011eno: Permission denied (13)
Mar 3 20:13:32 mythbox mythfrontend.real: mythfrontend[22207]: E
CoreContext mythplayer.cpp:2395 (VideoStart) Player(0): Video sync
method can't support double framerate (refresh rate too low for 2x deint)
Mar 3 20:13:32 mythbox mythfrontend.real: mythfrontend[22207]: I
CoreContext mythplayer.cpp:1789 (InitAVSync) Player(0): Display Refresh
Rate: 23.971 Video Frame Rate: 25.000



The equivalent on my system:

Nov 18 21:52:35 <HOSTNAME> mythfrontend.real: mythfrontend[2375]: I
CoreContext mythrender_opengl.cpp:76 (Create) OpenGL: Sync to VBlank is
enabled (good!)




I think the former could cause tearing. But I'd think that the nvidia
driver stuff could handle that stuff. I'm using the nvidia 304.131
driver, FWIW. However according to
<https://www.mythtv.org/wiki/Frame_display_timing>, "Contrary to what
Mythfrontend may say, VDPAU does not use vsync/vblank timing. RTC should
be used instead. " So that may be not relevant. My system config has
persisted through many years of MythTV, though, so that could even be an
artifact of a prior pre-VDPAU configuration.

As for the 23.97fps vs 25fps stuff, that surely seems like trouble. It
almost would seem that you're trying to play PAL video on an NTSC TV?


-WD
_______________________________________________
mythtv-users mailing list
mythtv-users@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-users
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org
Re: judder problems with nvidia 358.16 [ In reply to ]
On Friday, March 4, 2016 15:01 CET, Will Dormann <wdormann@gmail.com> wrote:
 On 3/4/16 5:10 AM, Marius Schrecker wrote:
> incorrect rtc permissions, although I don't see this fixing the
> incorrect refresh rates being consistently chosen.


I'm no expert on this stuff, but these look suspicious:

Mar 3 20:13:32 mythbox mythfrontend.real: mythfrontend[22207]: E
CoreContext vsync.cpp:275 (TryInit) VSYNC: DRMVideoSync: VBlank ioctl
did not work, unimplemented in this driver?
Mar 3 20:13:32 mythbox mythfrontend.real: mythfrontend[22207]: E
CoreContext vsync.cpp:356 (TryInit) VSYNC: RTCVideoSync: Could not open
/dev/rtc: #012#011#011#011eno: Permission denied (13)
Mar 3 20:13:32 mythbox mythfrontend.real: mythfrontend[22207]: E
CoreContext mythplayer.cpp:2395 (VideoStart) Player(0): Video sync
method can't support double framerate (refresh rate too low for 2x deint)
Mar 3 20:13:32 mythbox mythfrontend.real: mythfrontend[22207]: I
CoreContext mythplayer.cpp:1789 (InitAVSync) Player(0): Display Refresh
Rate: 23.971 Video Frame Rate: 25.000



The equivalent on my system:

Nov 18 21:52:35 <HOSTNAME> mythfrontend.real: mythfrontend[2375]: I
CoreContext mythrender_opengl.cpp:76 (Create) OpenGL: Sync to VBlank is
enabled (good!)




I think the former could cause tearing. But I'd think that the nvidia
driver stuff could handle that stuff. I'm using the nvidia 304.131
driver, FWIW. However according to
<https://www.mythtv.org/wiki/Frame_display_timing>, "Contrary to what
Mythfrontend may say, VDPAU does not use vsync/vblank timing. RTC should
be used instead. " So that may be not relevant. My system config has
persisted through many years of MythTV, though, so that could even be an
artifact of a prior pre-VDPAU configuration.

As for the 23.97fps vs 25fps stuff, that surely seems like trouble. It
almost would seem that you're trying to play PAL video on an NTSC TV?


-WD
Thanks Will,

The TV should handle most stuff that I throw at it, as the output of xrandr shows. Certainly there are both 25Hz and 50Hz modes so 25Hz material should present nop problem.  Interestingly, though not documented here, if I throw some 24p material at it, it tries to render it at 50Hz and I end up with the same problem.  

Mar  3 20:13:32 mythbox mythfrontend.real: mythfrontend[22207]: E
CoreContext vsync.cpp:275 (TryInit) VSYNC: DRMVideoSync: VBlank ioctl
did not work, unimplemented in this driver?

worries me. If I have to downgrade the nvidia driver, I will surely have to do the same with the kernel, but you also say that:

"Contrary to what
Mythfrontend may say, VDPAU does not use vsync/vblank timing. RTC should
be used instead."

I remember seeing some rtc permissions problems in one of the logs, so that might be a place to start:

https://www.mythtv.org/wiki/MythTV-HOWTO#ss5.4
rlimits
The rlimits method is the preferred method and is included in Linux 2.6.12 and above and requires PAM version 0.79 or above. Assuming anyone running mythfrontend is in the audio group and rlimits are supported, all you need to do is place this in your /etc/security/limits.conf
* - rtprio 0
* - nice 0
@audio - rtprio 50
@audio - nice 0

 
Re: judder problems with nvidia 358.16 [ In reply to ]
On 3/4/16 9:41 AM, Marius Schrecker wrote:
> Assuming anyone
> running mythfrontend is in the audio group and rlimits are supported,
> all you need to do is place this in your /etc/security/limits.conf
> * - rtprio 0
> * - nice 0
> @audio - rtprio 50
> @audio - nice 0/


I've got that in my system, FWIW.


-WD

_______________________________________________
mythtv-users mailing list
mythtv-users@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-users
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org
Re: judder problems with nvidia 358.16 [ In reply to ]
On Friday, March 4, 2016 15:48 CET, Will Dormann <wdormann@gmail.com> wrote:
 On 3/4/16 9:41 AM, Marius Schrecker wrote:
> Assuming anyone
> running mythfrontend is in the audio group and rlimits are supported,
> all you need to do is place this in your /etc/security/limits.conf
> * - rtprio 0
> * - nice 0
> @audio - rtprio 50
> @audio - nice 0/


I've got that in my system, FWIW.


-WD
 
Good to know. I'll certainly try it :)


 
Re: judder problems with nvidia 358.16 [ In reply to ]
Le vendredi 4 mars 2016, Marius Schrecker <marius.schrecker@lyse.net> a
écrit :

> As the original message got caught in the moderation filter, I'm
> reformatting it with more taken out to pastebin.
>
> To me, the likeliest candidates for the problem look like Compositing
> still being enebled somewhere, despite being disabled in xorg.conf and the
> xfce desktop configuration??
>
> incorrect rtc permissions, although I don't see this fixing the incorrect
> refresh rates being consistently chosen.
>
> All feedback and tips gratefully received!


Your logs are useless as you've trimmed out the actual important stuff.

When you run with -v playback --loglevel=debug

Don't trim, open a bug and attach it all.

How did you configure the GUI / Playback

What did you set the refresh rate too?

When you select the resolution, select 1920x1080 in the list of refresh
rate, which rates do you see appearing?
If it shows 60, 59.94, 50, 30, 25, 24 etc... Then set it to auto.

Then it will automatically set your screen to the right refresh rate,
provided your screen supports it.

(To test if changing refresh rate works, when you play with xrandr do you
see the refresh rate the screen is getting changing?)

>
Re: judder problems with nvidia 358.16 [ In reply to ]
Hi Jean Yves and many thanks for your reply.

On Saturday, March 5, 2016 13:09 CET, Jean-Yves Avenard <jyavenard@gmail.com> wrote:
  Your logs are useless as you've trimmed out the actual important stuff.Sorry about that, will try again this evening (WAF -permitting) When you run with -v playback --loglevel=debug Don't trim, open a bug and attach it all.Not sure where to attach bugs, but will try to find out. Pastebin balked at the last full log I tried to attach. How did you configure the GUI / Playback What did you set the refresh rate too?​​default res. set to 1920x1080 auto When you select the resolution, select 1920x1080 in the list of refresh rate, which rates do you see appearing?If it shows 60, 59.94, 50, 30, 25,  24 etc... Then set it to auto. Actually haven't checked what appears there, but have it set to auto.

Interestingly the resfresh rate to the TV does change, and changes diferently depending on the source material. It just doesn't change correctly.Then it will automatically set your screen to the right refresh rate, provided your screen supports it. (To test if changing refresh rate works, when you play with xrandr do you see the refresh rate the screen is getting changing?)Yes. Xrandr behaves completely as expected and TV reports the same refresh rate as I set using xrandr.


Will continue as soon as I get some time alone with the system. TV is normally in use whenever I have any free time, which makes all MythTV tinkering drag out.

BR.

Marius


 
Re: judder problems with nvidia 358.16 [ In reply to ]
Hello again,
    Sorry for the bad logs last time.

Captured and pasted as per instructions below:

On Saturday, March 5, 2016 13:09 CET, Jean-Yves Avenard <jyavenard@gmail.com> wrote:
 Your logs are useless as you've trimmed out the actual important stuff. When you run with -v playback --loglevel=debug Don't trim, open a bug and attach it all.25Hz Video. The TV now reports the correct refresh rate of 25Hz for this one.

http://pastebin.com/2UfJiVeE

23.97Hz (24p). Tv reports 50Hz
http://pastebin.com/eyAEX5K4
  How did you configure the GUI / Playback What did you set the refresh rate too?Auto When you select the resolution, select 1920x1080 in the list of refresh rate, which rates do you see appearing?If it shows 60, 59.94, 50, 30, 25,  24 etc... Then set it to auto.Stupid of me not to check that! The GUI refresh rates are the NVIDIA "false" rates: 50Hz, 51 Hz, 52 Hz, etc.. Then it will automatically set your screen to the right refresh rate, provided your screen supports it. (To test if changing refresh rate works, when you play with xrandr do you see the refresh rate the screen is getting changing?) Xrandr reports and sets the real refresh rates (confirmed with TV)


In the meantime I also added the mythuser (login user) to audio and set up limits.conf as per the wiki.

BR.

Marius
Re: judder problems with nvidia 358.16 [ In reply to ]
Hi

On Tuesday, 8 March 2016, Marius Schrecker <marius.schrecker@lyse.net>
wrote:

> Stupid of me not to check that! The GUI refresh rates are the NVIDIA
> "false" rates: 50Hz, 51 Hz, 52 Hz, etc..
>
>
>
>
That's what your problem is then, 50Hz is the closest to 23.93. So myth is
doing the right thing as far as its concerned.

Which version of Xrandr do you have installed?

Myth only supports 1.2 or the nvidia X extension(libnvctrl) , but this
needs to be installed and present.
Normally the nvidia installer would install it automatically.

If you completely clear any nvidia drivers and install from nvidia official
installer. Does it work?

When starting X do you see any error about not being able to load the
nvctrl X extension?

Do you see anything in mythtv log about being enable to access it?

That would explain why you are seeing this.

An alternative is to disable nvidia TwinView; then the refresh rates
reported to Xrandr will be the real ones (though as Xrandr 1.4 and earlier
can only deal with integers, you will only get things like 24, 25, 30 50,
60, not 59.997)



> Then it will automatically set your screen to the right refresh rate,
> provided your screen supports it.
>
> (To test if changing refresh rate works, when you play with xrandr do you
> see the refresh rate the screen is getting changing?)
>
>>
>
> Xrandr reports and sets the real refresh rates (confirmed with TV)
>
>
> In the meantime I also added the mythuser (login user) to audio and set up
> limits.conf as per the wiki.


Don't need that IMHO.
Those instructions were written like 10 years ago, all obsolete with the
new audio stack which was added like in 2009 or something!


>
> BR.
>
> Marius
Re: judder problems with nvidia 358.16 [ In reply to ]
On Tuesday, March 8, 2016 09:45 CET, Jean-Yves Avenard <jyavenard@gmail.com> wrote:
 Hi
...
...That's what your problem is then, 50Hz is the closest to 23.93. So myth is doing the right thing as far as its concerned.  Which version of Xrandr do you have installed? Myth only supports 1.2 or the nvidia X extension(libnvctrl) , but this needs to be installed and present. Normally the nvidia installer would install it automatically.  ​I see this on the nvidia release notes for driver 358.16:
Problems that Will Not Be Fixed

NV-CONTROL versions 1.8 and 1.9
Version 1.8 of the NV-CONTROL X Extension introduced target types for setting and querying attributes as well as receiving event notification on targets. Targets are objects like X Screens, GPUs and Quadro Sync devices. Previously, all attributes were described relative to an X Screen. These new bits of information (target type and target id) were packed in a non-compatible way in the protocol stream such that addressing X Screen 1 or higher would generate an X protocol error when mixing NV-CONTROL client and server versions.

This packing problem has been fixed in the NV-CONTROL 1.10 protocol, making it possible for the older (1.7 and prior) clients to communicate with NV-CONTROL 1.10 servers. Furthermore, the NV-CONTROL 1.10 client library has been updated to accommodate the target protocol packing bug when communicating with a 1.8 or 1.9 NV-CONTROL server. This means that the NV-CONTROL 1.10 client library should be able to communicate with any version of the NV-CONTROL server.

NVIDIA recommends that NV-CONTROL client applications relink with version 1.10 or later of the NV-CONTROL client library (libXNVCtrl.a, in the nvidia-settings-358.16.tar.bz2 tarball). The version of the client library can be determined by checking the NV_CONTROL_MAJOR and NV_CONTROL_MINOR definitions in the accompanying nv_control.h.

The only web released NVIDIA Linux driver that is affected by this problem (i.e., the only driver to use either version 1.8 or 1.9 of the NV-CONTROL X extension) is 1.0-8756.

Is that relevant?


BR.

Marius



 
Re: judder problems with nvidia 358.16 [ In reply to ]
Hi again Jean-Yves,

On Tuesday, March 8, 2016 09:45 CET, Jean-Yves Avenard <jyavenard@gmail.com> wrote:
 Hi

On Tuesday, 8 March 2016, Marius Schrecker <marius.schrecker@lyse.net> wrote:Stupid of me not to check that! The GUI refresh rates are the NVIDIA "false" rates: 50Hz, 51 Hz, 52 Hz, etc..  That's what your problem is then, 50Hz is the closest to 23.93. So myth is doing the right thing as far as its concerned.  Which version of Xrandr do you have installed?libxrandr2 : 1.5Myth only supports 1.2 or the nvidia X extension(libnvctrl) , but this needs to be installed and present. Normally the nvidia installer would install it automatically. 
 Hmmmm. This is proving to be black magic.  Explicitly adding the module load directive to xorg conf does result in:

343408.826] (II) LoadModule: "nv-control"
[343408.826] (WW) Warning, couldn't open module nv-control
[343408.826] (II) UnloadModule: "nv-control"
[343408.826] (II) Unloading nv-control
[343408.826] (EE) Failed to load module "nv-control" (module does not exist, 0)


So I started digging.

 There IS a Ubuntu package called:

 libxnvctrl0

so I installed it.
# apt-cache show libxnvctrl0
Package: libxnvctrl0
Priority: optional
Section: libs
Installed-Size: 74
Maintainer: Ubuntu MOTU Developers <ubuntu-motu@lists.ubuntu.com>
Original-Maintainer: Debian NVIDIA Maintainers <pkg-nvidia-devel@lists.alioth.debian.org>
Architecture: amd64
Source: nvidia-settings
Version: 352.21-0ubuntu1
Depends: libc6 (>= 2.14), libx11-6 (>= 2:1.4.99.1), libxext6
Filename: pool/main/n/nvidia-settings/libxnvctrl0_352.21-0ubuntu1_amd64.deb
Size: 11534
MD5sum: 69b4410f3c2428d3e2bfbc851ae3f4b2
SHA1: 1a0ef460048106be1fed88d668d53680767a1a69
SHA256: 9f81e8f9e8c10551e047dad6de23800fd9d69365f32bb2419da16851b4f7c6c8
Description-en: NV-CONTROL X extension (runtime library)
 The NV-CONTROL X extension provides a mechanism for X clients to
 query and set configuration parameters of the NVIDIA X driver.
 State set by the NV-CONTROL X extension is assumed to be persistent
 only for the current server generation.
 .
 This package contains the shared library.
Description-md5: 6d4f8aebac836277193a58e922a49603
Bugs: https://bugs.launchpad.net/ubuntu/+filebug
Origin: Ubuntu
Supported: 9m
​

 The version is  352.21-0ubuntu1

 and itseems to install the library file libXNVCtrl.so  to /usr/lib:

# locate libXNVCtrl.so
/usr/lib/libXNVCtrl.so.0
/usr/lib/libXNVCtrl.so.0.0.0
​

 (NOT to "/usr/lib/xorg/modules") which is where Xorg is expecting to find them.

 Question 1: Should I symlink?

I then looked in the nvidia driver source for traces of the same library, but found nothing, so either it has a different name here or needs to be downloaded separately, but I have yet to find where.

As the packaged version of the livrary was 352.21, I decided to try again to install the same version of the driver, first using the sgfxi script (which refuses), then using the pre-packaged version, which looses my nvidia GLX extension.

So that's where I am so far.

 Maybe the control library doesn't need to be in the same version as the rest of the driver, in which case I could reinstall 358 or 361 from source (sgfxi or manual build) and run against  libXNVCtrl 352.21 symlinked to /usr/lib/xorg/modules 

 Otherwise I do need to work out where to find or how to build the version of the library matching a buildable driver.

any clues?

 BR.

--Marius--

  

  If you completely clear any nvidia drivers and install from nvidia official installer. Does it work? When starting X do you see any error about not being able to load the nvctrl X extension? Do you see anything in mythtv log about being enable to access it? That would explain why you are seeing this.  An alternative is to disable nvidia TwinView; then the refresh rates reported to Xrandr will be the real ones (though as Xrandr 1.4 and earlier can only deal with integers, you will only get things like 24, 25, 30 50, 60, not 59.997)  Then it will automatically set your screen to the right refresh rate, provided your screen supports it. (To test if changing refresh rate works, when you play with xrandr do you see the refresh rate the screen is getting changing?) Xrandr reports and sets the real refresh rates (confirmed with TV)


In the meantime I also added the mythuser (login user) to audio and set up limits.conf as per the wiki. Don't need that IMHO. Those instructions were written like 10 years ago, all obsolete with the new audio stack which was added like in 2009 or something! 
BR.

Marius


 
Re: judder problems with nvidia 358.16 [ In reply to ]
Hi all,

 Short, deliberate top posting to say that the problem remains after following bottom-posted workflow.
 
On Tuesday, 8 March 2016, Marius Schrecker <marius.schrecker@lyse.net> wrote:Stupid of me not to check that! The GUI refresh rates are the NVIDIA "false" rates: 50Hz, 51 Hz, 52 Hz, etc..  That's what your problem is then, 50Hz is the closest to 23.93. So myth is doing the right thing as far as its concerned.  Which version of Xrandr do you have installed?libxrandr2 : 1.5Myth only supports 1.2 or the nvidia X extension(libnvctrl) , but this needs to be installed and present. Normally the nvidia installer would install it automatically. 
 Hmmmm. This is proving to be black magic.  Explicitly adding the module load directive to xorg conf does result in:

343408.826] (II) LoadModule: "nv-control"
[343408.826] (WW) Warning, couldn't open module nv-control
[343408.826] (II) UnloadModule: "nv-control"
[343408.826] (II) Unloading nv-control
[343408.826] (EE) Failed to load module "nv-control" (module does not exist, 0)


So I started digging.

 There IS a Ubuntu package called:

 libxnvctrl0

so I installed it.
# apt-cache show libxnvctrl0
Package: libxnvctrl0
Priority: optional
Section: libs
Installed-Size: 74
Maintainer: Ubuntu MOTU Developers <ubuntu-motu@lists.ubuntu.com>
Original-Maintainer: Debian NVIDIA Maintainers <pkg-nvidia-devel@lists.alioth.debian.org>
Architecture: amd64
Source: nvidia-settings
Version: 352.21-0ubuntu1
Depends: libc6 (>= 2.14), libx11-6 (>= 2:1.4.99.1), libxext6
Filename: pool/main/n/nvidia-settings/libxnvctrl0_352.21-0ubuntu1_amd64.deb
Size: 11534
MD5sum: 69b4410f3c2428d3e2bfbc851ae3f4b2
SHA1: 1a0ef460048106be1fed88d668d53680767a1a69
SHA256: 9f81e8f9e8c10551e047dad6de23800fd9d69365f32bb2419da16851b4f7c6c8
Description-en: NV-CONTROL X extension (runtime library)
 The NV-CONTROL X extension provides a mechanism for X clients to
 query and set configuration parameters of the NVIDIA X driver.
 State set by the NV-CONTROL X extension is assumed to be persistent
 only for the current server generation.
 .
 This package contains the shared library.
Description-md5: 6d4f8aebac836277193a58e922a49603
Bugs: https://bugs.launchpad.net/ubuntu/+filebug
Origin: Ubuntu
Supported: 9m
​

 The version is  352.21-0ubuntu1

 and itseems to install the library file libXNVCtrl.so  to /usr/lib:

# locate libXNVCtrl.so
/usr/lib/libXNVCtrl.so.0
/usr/lib/libXNVCtrl.so.0.0.0
​

 (NOT to "/usr/lib/xorg/modules") which is where Xorg is expecting to find them.

 Question 1: Should I symlink?

I then looked in the nvidia driver source for traces of the same library, but found nothing, so either it has a different name here or needs to be downloaded separately, but I have yet to find where.

As the packaged version of the livrary was 352.21, I decided to try again to install the same version of the driver, first using the sgfxi script (which refuses), then using the pre-packaged version, which looses my nvidia GLX extension.

So that's where I am so far.

 Maybe the control library doesn't need to be in the same version as the rest of the driver, in which case I could reinstall 358 or 361 from source (sgfxi or manual build) and run against  libXNVCtrl 352.21 symlinked to /usr/lib/xorg/modules 

 Otherwise I do need to work out where to find or how to build the version of the library matching a buildable driver.

any clues?

 BR.

--Marius--

  

  If you completely clear any nvidia drivers and install from nvidia official installer. Does it work? When starting X do you see any error about not being able to load the nvctrl X extension? Do you see anything in mythtv log about being enable to access it? That would explain why you are seeing this.  An alternative is to disable nvidia TwinView; then the refresh rates reported to Xrandr will be the real ones (though as Xrandr 1.4 and earlier can only deal with integers, you will only get things like 24, 25, 30 50, 60, not 59.997)  Then it will automatically set your screen to the right refresh rate, provided your screen supports it. (To test if changing refresh rate works, when you play with xrandr do you see the refresh rate the screen is getting changing?) Xrandr reports and sets the real refresh rates (confirmed with TV)


In the meantime I also added the mythuser (login user) to audio and set up limits.conf as per the wiki. Don't need that IMHO. Those instructions were written like 10 years ago, all obsolete with the new audio stack which was added like in 2009 or something! 
BR.

Marius


Latest news:

  I symlinked the libXNVCtrl.so basefile symlinked to /usr/lib/xorg/modules replicating symlinks and filenames from /usr/lib.

I built nvidia 352.63 direct from the Nvidia sources. It compiled without errors and loads fine.

All working except that NV_CONTROL is still not loading. Nvidia "false" refresh-rates still listed in Mythfrontend setup.

BR.

--Marius--
 

 


 
Re: judder problems with nvidia 358.16 [ In reply to ]
maybe nvctrl extensions was removed from the nvidia drivers.

Would have to implement the new xrandr

which version of xrandr do you have ?


On 15 March 2016 at 01:34, Marius Schrecker <marius.schrecker@lyse.net> wrote:
> Hi all,
>
> Short, deliberate top posting to say that the problem remains after
> following bottom-posted workflow.
>
>
>
>
> On Tuesday, 8 March 2016, Marius Schrecker <marius.schrecker@lyse.net>
> wrote:
>>
>> Stupid of me not to check that! The GUI refresh rates are the NVIDIA
>> "false" rates: 50Hz, 51 Hz, 52 Hz, etc..
>>
>>
>
>
> That's what your problem is then, 50Hz is the closest to 23.93. So myth is
> doing the right thing as far as its concerned.
>
> Which version of Xrandr do you have installed?
>
> libxrandr2 : 1.5
>
> Myth only supports 1.2 or the nvidia X extension(libnvctrl) , but this needs
> to be installed and present.
> Normally the nvidia installer would install it automatically.
>
>
> Hmmmm. This is proving to be black magic. Explicitly adding the module load
> directive to xorg conf does result in:
>
> 343408.826] (II) LoadModule: "nv-control"
> [343408.826] (WW) Warning, couldn't open module nv-control
> [343408.826] (II) UnloadModule: "nv-control"
> [343408.826] (II) Unloading nv-control
> [343408.826] (EE) Failed to load module "nv-control" (module does not exist,
> 0)
>
>
> So I started digging.
>
> There IS a Ubuntu package called:
>
> libxnvctrl0
>
> so I installed it.
> # apt-cache show libxnvctrl0
> Package: libxnvctrl0
> Priority: optional
> Section: libs
> Installed-Size: 74
> Maintainer: Ubuntu MOTU Developers <ubuntu-motu@lists.ubuntu.com>
> Original-Maintainer: Debian NVIDIA Maintainers
> <pkg-nvidia-devel@lists.alioth.debian.org>
> Architecture: amd64
> Source: nvidia-settings
> Version: 352.21-0ubuntu1
> Depends: libc6 (>= 2.14), libx11-6 (>= 2:1.4.99.1), libxext6
> Filename: pool/main/n/nvidia-settings/libxnvctrl0_352.21-0ubuntu1_amd64.deb
> Size: 11534
> MD5sum: 69b4410f3c2428d3e2bfbc851ae3f4b2
> SHA1: 1a0ef460048106be1fed88d668d53680767a1a69
> SHA256: 9f81e8f9e8c10551e047dad6de23800fd9d69365f32bb2419da16851b4f7c6c8
> Description-en: NV-CONTROL X extension (runtime library)
> The NV-CONTROL X extension provides a mechanism for X clients to
> query and set configuration parameters of the NVIDIA X driver.
> State set by the NV-CONTROL X extension is assumed to be persistent
> only for the current server generation.
> .
> This package contains the shared library.
> Description-md5: 6d4f8aebac836277193a58e922a49603
> Bugs: https://bugs.launchpad.net/ubuntu/+filebug
> Origin: Ubuntu
> Supported: 9m
>
>
> The version is 352.21-0ubuntu1
>
> and itseems to install the library file libXNVCtrl.so to /usr/lib:
>
> # locate libXNVCtrl.so
> /usr/lib/libXNVCtrl.so.0
> /usr/lib/libXNVCtrl.so.0.0.0
>
>
> (NOT to "/usr/lib/xorg/modules") which is where Xorg is expecting to find
> them.
>
> Question 1: Should I symlink?
>
> I then looked in the nvidia driver source for traces of the same library,
> but found nothing, so either it has a different name here or needs to be
> downloaded separately, but I have yet to find where.
>
> As the packaged version of the livrary was 352.21, I decided to try again to
> install the same version of the driver, first using the sgfxi script (which
> refuses), then using the pre-packaged version, which looses my nvidia GLX
> extension.
>
> So that's where I am so far.
>
> Maybe the control library doesn't need to be in the same version as the
> rest of the driver, in which case I could reinstall 358 or 361 from source
> (sgfxi or manual build) and run against libXNVCtrl 352.21 symlinked to
> /usr/lib/xorg/modules
>
> Otherwise I do need to work out where to find or how to build the version
> of the library matching a buildable driver.
>
> any clues?
>
> BR.
>
> --Marius--
>
>
>
>
>
>
> If you completely clear any nvidia drivers and install from nvidia official
> installer. Does it work?
>
> When starting X do you see any error about not being able to load the nvctrl
> X extension?
>
> Do you see anything in mythtv log about being enable to access it?
>
> That would explain why you are seeing this.
>
> An alternative is to disable nvidia TwinView; then the refresh rates
> reported to Xrandr will be the real ones (though as Xrandr 1.4 and earlier
> can only deal with integers, you will only get things like 24, 25, 30 50,
> 60, not 59.997)
>
>
>>
>> Then it will automatically set your screen to the right refresh rate,
>> provided your screen supports it.
>>
>> (To test if changing refresh rate works, when you play with xrandr do you
>> see the refresh rate the screen is getting changing?)
>>>
>>>
>>
>> Xrandr reports and sets the real refresh rates (confirmed with TV)
>>
>>
>> In the meantime I also added the mythuser (login user) to audio and set up
>> limits.conf as per the wiki.
>
>
> Don't need that IMHO.
> Those instructions were written like 10 years ago, all obsolete with the new
> audio stack which was added like in 2009 or something!
>
>>
>>
>> BR.
>>
>> Marius
>
>
>
>
> Latest news:
>
> I symlinked the libXNVCtrl.so basefile symlinked to /usr/lib/xorg/modules
> replicating symlinks and filenames from /usr/lib.
>
> I built nvidia 352.63 direct from the Nvidia sources. It compiled without
> errors and loads fine.
>
> All working except that NV_CONTROL is still not loading. Nvidia "false"
> refresh-rates still listed in Mythfrontend setup.
>
> BR.
>
> --Marius--
>
>
>
>
>
>
>
>
> _______________________________________________
> mythtv-users mailing list
> mythtv-users@mythtv.org
> http://lists.mythtv.org/mailman/listinfo/mythtv-users
> http://wiki.mythtv.org/Mailing_List_etiquette
> MythTV Forums: https://forum.mythtv.org
>
_______________________________________________
mythtv-users mailing list
mythtv-users@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-users
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org
Re: judder problems with nvidia 358.16 [ In reply to ]
On Tuesday, March 15, 2016 09:11 CET, Jean-Yves Avenard <jyavenard@gmail.com> wrote:
 maybe nvctrl extensions was removed from the nvidia drivers.

Would have to implement the new xrandr

which version of xrandr do you have ?​1.5

BR.

Marius
Re: judder problems with nvidia 358.16 [ In reply to ]
Hi again,

On Tuesday, March 15, 2016 09:11 CET, Jean-Yves Avenard <jyavenard@gmail.com> wrote:
 maybe nvctrl extensions was removed from the nvidia drivers.

Would have to implement the new xrandr

which version of xrandr do you have ?​1.5
On 15 March 2016 at 01:34, Marius Schrecker <marius.schrecker@lyse.net> wrote:
...
>
> On Tuesday, 8 March 2016, Marius Schrecker <marius.schrecker@lyse.net>
> wrote:
>>
>> Stupid of me not to check that! The GUI refresh rates are the NVIDIA
>> "false" rates: 50Hz, 51 Hz, 52 Hz, etc..
>>
>>
>
>
> That's what your problem is then, 50Hz is the closest to 23.93. So myth is
> doing the right thing as far as its concerned.
>
> Which version of Xrandr do you have installed?
>
> libxrandr2 : 1.5
>
> Myth only supports 1.2 or the nvidia X extension(libnvctrl) , but this needs
> to be installed and present.
> Normally the nvidia installer would install it automatically.
>
>The strange thing is that the nvidia settings panel displays The NV-CONTROL version - 1.29 I think (will check), but I can't find any reference to the library file that it's using and X-org is certainly failing to find/load it.

Neither can I find a library that is recognisable as the control extension in any of the Nvidia installer tarballs that I nhave looked at - latest attempt being 352.x which was reported as working by another user.  So I think that the problem may still exist between the chair and the keyboard ;).

yesterday I decided to give up and disable dynamic twin view, by setting it "false" in xorg.conf

 To my dismay the problem persists and I still see the "false" refresh rates in mythtv settings
> Hmmmm. This is proving to be black magic. Explicitly adding the module load
> directive to xorg conf does result in:
>
> 343408.826] (II) LoadModule: "nv-control"
> [343408.826] (WW) Warning, couldn't open module nv-control
> [343408.826] (II) UnloadModule: "nv-control"
> [343408.826] (II) Unloading nv-control
> [343408.826] (EE) Failed to load module "nv-control" (module does not exist,
> 0)
>
>
> So I started digging.
>
> There IS a Ubuntu package called:
>
> libxnvctrl0
>
> so I installed it.
> # apt-cache show libxnvctrl0
> Package: libxnvctrl0
> Priority: optional
> Section: libs
> Installed-Size: 74
> Maintainer: Ubuntu MOTU Developers <ubuntu-motu@lists.ubuntu.com>
> Original-Maintainer: Debian NVIDIA Maintainers
> <pkg-nvidia-devel@lists.alioth.debian.org>
> Architecture: amd64
> Source: nvidia-settings
> Version: 352.21-0ubuntu1
> Depends: libc6 (>= 2.14), libx11-6 (>= 2:1.4.99.1), libxext6
> Filename: pool/main/n/nvidia-settings/libxnvctrl0_352.21-0ubuntu1_amd64.deb
> Size: 11534
> MD5sum: 69b4410f3c2428d3e2bfbc851ae3f4b2
> SHA1: 1a0ef460048106be1fed88d668d53680767a1a69
> SHA256: 9f81e8f9e8c10551e047dad6de23800fd9d69365f32bb2419da16851b4f7c6c8
> Description-en: NV-CONTROL X extension (runtime library)
> The NV-CONTROL X extension provides a mechanism for X clients to
> query and set configuration parameters of the NVIDIA X driver.
> State set by the NV-CONTROL X extension is assumed to be persistent
> only for the current server generation.
> .
> This package contains the shared library.
> Description-md5: 6d4f8aebac836277193a58e922a49603
> Bugs: https://bugs.launchpad.net/ubuntu/+filebug
> Origin: Ubuntu
> Supported: 9m
>
>
> The version is 352.21-0ubuntu1 >> and itseems to install the library file libXNVCtrl.so to /usr/lib:
>
> # locate libXNVCtrl.so
> /usr/lib/libXNVCtrl.so.0
> /usr/lib/libXNVCtrl.so.0.0.0
>
>
> (NOT to "/usr/lib/xorg/modules") which is where Xorg is expecting to find
> them.
>
> Question 1: Should I symlink?
>
> I then looked in the nvidia driver source for traces of the same library,
> but found nothing, so either it has a different name here or needs to be
> downloaded separately, but I have yet to find where.
>
> As the packaged version of the livrary was 352.21, I decided to try again to
> install the same version of the driver, first using the sgfxi script (which
> refuses), then using the pre-packaged version, which looses my nvidia GLX> extension.>
> So that's where I am so far.
>
> Maybe the control library doesn't need to be in the same version as the
> rest of the driver, in which case I could reinstall 358 or 361 from source
> (sgfxi or manual build) and run against libXNVCtrl 352.21 symlinked to
> /usr/lib/xorg/modules
>
> Otherwise I do need to work out where to find or how to build the version
> of the library matching a buildable driver.
>
> any clues?
>
> BR.
>
> --Marius--
>
>
>
>
>
>
> If you completely clear any nvidia drivers and install from nvidia official
> installer. Does it work?
>
> When starting X do you see any error about not being able to load the nvctrl
> X extension?
>
> Do you see anything in mythtv log about being enable to access it?
>
> That would explain why you are seeing this.
>
> An alternative is to disable nvidia TwinView; then the refresh rates
> reported to Xrandr will be the real ones (though as Xrandr 1.4 and earlier
> can only deal with integers, you will only get things like 24, 25, 30 50,
> 60, not 59.997)
>
>
>>
>> Then it will automatically set your screen to the right refresh rate,
>> provided your screen supports it.
>>
>> (To test if changing refresh rate works, when you play with xrandr do you
>> see the refresh rate the screen is getting changing?)
>>>
>>>
>>
>> Xrandr reports and sets the real refresh rates (confirmed with TV)
>>
>>
>> In the meantime I also added the mythuser (login user) to audio and set up
>> limits.conf as per the wiki.
>
>
> Don't need that IMHO.
> Those instructions were written like 10 years ago, all obsolete with the new
> audio stack which was added like in 2009 or something!
>
>>
 


 
Re: judder problems with nvidia 358.16 [ In reply to ]
On 17 March 2016 at 18:25, Marius Schrecker <marius.schrecker@lyse.net> wrote:

> yesterday I decided to give up and disable dynamic twin view, by setting it
> "false" in xorg.conf
>
> To my dismay the problem persists and I still see the "false" refresh rates
> in mythtv settings

obviously didn't :) I rememver having issue disabling it, there were
more than one option to set.
_______________________________________________
mythtv-users mailing list
mythtv-users@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-users
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org
Re: judder problems with nvidia 358.16 [ In reply to ]
On Thursday, March 17, 2016 13:23 CET, Jean-Yves Avenard <jyavenard@gmail.com> wrote:
 On 17 March 2016 at 18:25, Marius Schrecker <marius.schrecker@lyse.net> wrote:

> yesterday I decided to give up and disable dynamic twin view, by setting it
> "false" in xorg.conf
>
> To my dismay the problem persists and I still see the "false" refresh rates
> in mythtv settings

obviously didn't :) I rememver having issue disabling it, there were
more than one option to set.
_______________________________________________Thanks!
That would explain it... will investigate further ;)
Re: judder problems with nvidia 358.16 [ In reply to ]
On 17 March 2016 at 18:25, Marius Schrecker <marius.schrecker@lyse.net>
wrote:
>On March 17, 2016 13:23 CET, Jean-Yves Avenard <jyavenard@gmail.com>
wrote:
>On 17 March 2016 at 18:25, Marius Schrecker <marius.schrecker@lyse.net>
wrote:

> yesterday I decided to give up and disable dynamic twin view, by setting
it
> "false" in xorg.conf
>
> To my dismay the problem persists and I still see the "false" refresh
rates
> in mythtv settings
>
>obviously didn't :) I rememver having issue disabling it, there were
>more than one option to set.
>_______________________________________________Thanks!
>That would explain it... will investigate further ;)

I believe this is the same issue in thread " Incorrect Refresh with Nvidia
361.28 - Long "

Any news on this issue? The only workaround that seems to provide the
correct refresh rate is to choose 51 Hz in setup instead of Auto. 51 Hz is
60p and works for most things, but I do have some 24p and 30p sources and
it would be nice to have them automatically switch refresh rates.

Something changed in the Nvidia drivers and it is clear they have not
changed it back, as the latest drivers exhibit the same issue.


Mark
Re: judder problems with nvidia 358.16 [ In reply to ]
On Tue, Aug 1, 2017 at 3:14 PM, Mark Goldberg <marklgoldberg@gmail.com> wrote:
>>That would explain it... will investigate further ;)
>
> I believe this is the same issue in thread " Incorrect Refresh with Nvidia
> 361.28 - Long "
>
> Any news on this issue? The only workaround that seems to provide the
> correct refresh rate is to choose 51 Hz in setup instead of Auto. 51 Hz is
> 60p and works for most things, but I do have some 24p and 30p sources and it
> would be nice to have them automatically switch refresh rates.
>
> Something changed in the Nvidia drivers and it is clear they have not
> changed it back, as the latest drivers exhibit the same issue.
>
>
> Mark

Yup, this is an issue and nvidia is aware of it though they don't have
a viable solution yet. They were able to reproduce it and have a
workaround but I've not tried it. See the forum for details:

https://devtalk.nvidia.com/default/topic/898310/linux/352-63-regression-quot-noedidmodes-quot-breaks-custom-quot-modeline-quot-modes-ubuntu-14-04-/

I found that I was able to get the 340.98 drivers to work
correctly.... nothing after that. Fortunately this was able to be
install on a 4.8.0 kernel. Until Nvidia has a true fix I'll stay with
this driver.

Neil
_______________________________________________
mythtv-users mailing list
mythtv-users@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-users
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org
Re: judder problems with nvidia 358.16 [ In reply to ]
On Tue, Aug 2, 2017, you wrote:

>Yup, this is an issue and nvidia is aware of it though they don't have
>a viable solution yet. They were able to reproduce it and have a
>workaround but I've not tried it. See the forum for details:
>
>https://devtalk.nvidia.com/default/topic/898310/linux/352-63-regression-quot-noedidmodes-quot-breaks-custom-quot-modeline-quot-modes-ubuntu-14-04-/
<https://devtalk.nvidia.com/default/topic/898310/linux/352-63-regression-quot-noedidmodes-quot-breaks-custom-quot-modeline-quot-modes-ubuntu-14-04-/>
>
>I found that I was able to get the 340.98 drivers to work
>correctly.... nothing after that. Fortunately this was able to be
>install on a 4.8.0 kernel. Until Nvidia has a true fix I'll stay with
>this driver.
>
>Neil

Are you sure this is the same issue? Did it also cause the refresh rates to
be wrong in Mythtv? The Nvidia forum discusses modelines and custom EDID
and does not mention the Mythtv refresh.

In any case, if it is related, I have a newer 1050 card and cannot use the
older drivers as it is not supported.

Thanks for the info,

Mark
Re: judder problems with nvidia 358.16 [ In reply to ]
Actually I think this is something a little different than that issue. I
believe it boils down to xrandr (X Resize, Rotate and Reflection extension)
only reporting refresh rates as integers. If memory serves, Nvidia ended
up "overriding" the refresh rate to values like 50, 51 in order to report
them uniquely. I think it also had something to do with Twinview/Xinerama
and multiple monitor support. In the past, you could turn this off and get
the "real" refresh rates out, but I think later drivers always ended up
reporting these fake values. But without the fake values, you can't even
distinguish between 59.94 and 60.00 HZ as they both come out as 60,

I thought maybe a later update to X.Org fixed this, but it seems like it
remains that way. However, I *have* noticed the xrandr command line
utility does report 59.94 and 60.00 accurately! It looks like it actually
does a little calculation using some of the raw modeline values to get
around this flaw, but the numbers seem to check out.

I will probably try monkeying around in the code to see if I can get
Mythfrontend to spit out the true refresh rates. I'm not sure how far down
this integer/double confusion goes and if there are other workarounds in
place.

Thanks,
Alex
Re: judder problems with nvidia 358.16 [ In reply to ]
On Wed, Aug 2, 2017 you wrote:

> Actually I think this is something a little different than that issue. I
> believe it boils down to xrandr (X Resize, Rotate and Reflection
extension)
> only reporting refresh rates as integers. If memory serves, Nvidia ended
> up "overriding" the refresh rate to values like 50, 51 in order to report
> them uniquely. I think it also had something to do with Twinview/Xinerama
> and multiple monitor support. In the past, you could turn this off and get
> the "real" refresh rates out, but I think later drivers always ended up
> reporting these fake values. But without the fake values, you can't even
> distinguish between 59.94 and 60.00 HZ as they both come out as 60,
>
> I thought maybe a later update to X.Org fixed this, but it seems like it
> remains that way. However, I *have* noticed the xrandr command line
> utility does report 59.94 and 60.00 accurately! It looks like it actually
> does a little calculation using some of the raw modeline values to get
> around this flaw, but the numbers seem to check out.
>
> I will probably try monkeying around in the code to see if I can get
> Mythfrontend to spit out the true refresh rates. I'm not sure how far down
> this integer/double confusion goes and if there are other workarounds in
> place.
>
> Thanks,
> Alex
>

It used to work for years before Nvidia changed something. Myth could
automatically choose the appropriate framerate. See this other thread for
more info:

https://lists.gt.net/mythtv/users/596480

and these:

https://code.mythtv.org/trac/ticket/5643
https://code.mythtv.org/trac/ticket/11493

and I am sure a few more.

Basically, this:

https://www.mythtv.org/wiki/User_Manual:JudderFree

does not work any more.


Thanks for looking.

Mark
Re: judder problems with nvidia 358.16 [ In reply to ]
On Wed, Aug 2, 2017 at 9:20 PM, Mark Goldberg <marklgoldberg@gmail.com> wrote:
>
> It used to work for years before Nvidia changed something. Myth could
> automatically choose the appropriate framerate. See this other thread for
> more info:
>
> https://lists.gt.net/mythtv/users/596480
>
> and these:
>
> https://code.mythtv.org/trac/ticket/5643
> https://code.mythtv.org/trac/ticket/11493
>
> and I am sure a few more.
>
> Basically, this:
>
> https://www.mythtv.org/wiki/User_Manual:JudderFree
>
> does not work any more.
>
>
> Thanks for looking.
>
> Mark
>

Exactly!!!! Once I tried a newer nvidia driver something broke.
Moving back to 340.98 allowed me to use the modelines and techniques
on the JudderFree page. With the newer drivers (post 340.98) MythTV
would default to a 51 hz (or something odd line that) and would never
select the correct refresh rate. I'd often have interlaced menus that
looked horrible.
_______________________________________________
mythtv-users mailing list
mythtv-users@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-users
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org
Re: judder problems with nvidia 358.16 [ In reply to ]
I don't doubt that something changed in the NVIDIA driver that broke it.
Digging into it some, I've found that MythTV handles its list of screen
sizes/refresh rates the following way:

1. Query XRandr for the screen sizes
2. Query XRandr for each of the available refresh rates for each screen
size
- These are the integer values that would either be rounded numbers
(if you're not using the NVIDIA driver) or the "fake" ones like
50, 51 for
59.94 and 60.00.
3. Query the NV-CTRL extension (if available) for its available video
modes
4. Tokenize and parse out the mode data to modelines, while determining
if they match the proper connected device
5. Calculate the precise refresh rates using these modelines
6. Take these "real" refresh rates and marry them to the "fake" refresh
rates
- I am still very unclear how this matching succeeds. Or if it was
always getting it right.
7. When trying to switch refresh rates/resolutions, look for the "real"
refresh rate, then feed the corresponding "fake" refresh rate to XRandr
since that's the only value it understands.

Step 4 is clearly the one that is failing, I believe on one of the many
checks needed to match the raw mode value: forcing some of the checks gets
at least some difference in the reported refresh rates, although not the
right ones. No doubt NVIDIA changed something in what is getting reported
back by the NV-CTRL extension. Unfortunately I do not currently have an
older driver version to compare against and this parsing logic is pretty
involved.

IMO this is a brittle system that is likely to break again and again as it
seems to rely on undocumented NVIDIA driver behavior. I think I've made a
*little* bit of progress on a different approach that relies on XRandr
alone to query/set modelines, but I admit I am no expert at this.


Thanks,
Alex
Re: judder problems with nvidia 358.16 [ In reply to ]
On Aug 5, 2017 you wrote:
>
> I don't doubt that something changed in the NVIDIA driver that broke it.
> Digging into it some, I've found that MythTV handles its list of screen
> sizes/refresh rates the following way:
>
> Step 4 is clearly the one that is failing, I believe on one of the many
> checks needed to match the raw mode value: forcing some of the checks gets
> at least some difference in the reported refresh rates, although not the
> right ones. No doubt NVIDIA changed something in what is getting reported
> back by the NV-CTRL extension. Unfortunately I do not currently have an
> older driver version to compare against and this parsing logic is pretty
> involved.
>

My earlier posting at https://lists.gt.net/mythtv/users/596480 has some log
snippets that may be of help. It shows the difference in front end log
output with old and new Nvidia drivers.

Thanks for looking into this.

Mark

1 2  View All