Forums > Jitter

jit.gl.slab v1.6 bug with multiple displays

June 26, 2006 | 9:30 am

Hi,

my setup consists of a G5 DualCore2,3GHz, QT7.11,OSX4.6,
3x GForce6600.

Just try the jit.gl.slab help file.

1) Start qmetro
2) Start metro

As soon as you start the metro the following message is sent for each bang to the MAX window,

error: jit.gl.texture: error binding readback mechanism for capture!
error: jit.gl.pbuffer: error binding invalid pbuffer!

Does not happen on a MacBook (no second screen attached).

Bernd


August 26, 2006 | 11:54 am

Quote: Bernd wrote on Mon, 26 June 2006 03:30
—————————————————-
> Hi,
>
> my setup consists of a G5 DualCore2,3GHz, QT7.11,OSX4.6,
> 3x GForce6600.
>
> Just try the jit.gl.slab help file.
>
> 1) Start qmetro
> 2) Start metro
>
> As soon as you start the metro the following message is sent for each bang to the MAX window,
>
>
> error: jit.gl.texture: error binding readback mechanism for capture!
> error: jit.gl.pbuffer: error binding invalid pbuffer!
>
> Does not happen on a MacBook (no second screen attached).
>
> Bernd
>
—————————————————-

Am I the only one with this problem?
I tried with 4.6.1/1.6.1 same result?!

Thanks for any help

Bernd


August 26, 2006 | 9:07 pm

On Aug 26, 2006, at 4:54 AM, Bernd wrote:

> my setup consists of a G5 DualCore2,3GHz, QT7.11,OSX4.6,
> 3x GForce6600.
> Am I the only one with this problem?
> I tried with 4.6.1/1.6.1 same result?!

Unfortunately I don’t have a machine with a similar setup to test.
You’ll need to provide us more information about how you’re using the
multiple displays.

After it reports that error, does it actually draw fine? There have
been instances where this problem happens the first time rendering,
but afterwards without problem.

Anyone else with a similar setup to Bernd?

-Joshua


August 27, 2006 | 11:09 am

> You’ll need to provide us more information about how you’re using the
> multiple displays.

Currently 4 monitors are sitting side by side in the monitor setup pref pane, no mirroring. No matter on which I move the Finder bar, always the same result with the jit.gl.slab helpfile.

>
> After it reports that error, does it actually draw fine?

No drawing, each bang creates the error messages.
It looks like this:

jit.gl.render: building GL on window "foo"…
? error: jit.gl.pbuffer: unable to create pbuffer context!
? error: jit.gl.texture: error attaching texture to readback mechanism for capture!
? error: jit.gl.pbuffer: unable to create pbuffer context!
? error: jit.gl.texture: error attaching texture to readback mechanism for capture!
? error: jit.gl.pbuffer: unable to create pbuffer context!
? error: jit.gl.texture: error attaching texture to readback mechanism for capture!
? error: jit.gl.pbuffer: error binding invalid pbuffer!
? error: jit.gl.texture: error binding readback mechanism for capture!
? error: jit.gl.pbuffer: error binding invalid pbuffer!
? error: jit.gl.texture: error binding readback mechanism for capture!
? error: jit.gl.pbuffer: error binding invalid pbuffer!
? error: jit.gl.texture: error binding readback mechanism for capture!
? error: jit.gl.pbuffer: error binding invalid pbuffer!
? error: jit.gl.texture: error binding readback mechanism for capture!
? error: jit.gl.pbuffer: error binding invalid pbuffer!
? error: jit.gl.texture: error binding readback mechanism for capture!
? error: jit.gl.pbuffer: error binding invalid pbuffer!
? er

Bernd


August 27, 2006 | 12:31 pm

Further investigation:

Starting up with only one monitor, jit.gl.slab works well.

As soon as having a second monitor connected at the SECOND video card (which is in the 8-lane slot), the error occurs.

Hth

Bernd


August 27, 2006 | 2:55 pm


August 27, 2006 | 10:30 pm

On Aug 27, 2006, at 5:31 AM, Bernd wrote:

> Further investigation: Starting up with only one monitor,
> jit.gl.slab works well.
> As soon as having a second monitor connected at the SECOND video
> card (which is in the 8-lane slot), the error occurs.

Interesting. There might be some aspect of using these cards in this
configuration that actually prevents using the appropriate driver for
these cards, and it might be falling back on the software based
renderer, or some other more basic OpenGL driver. Are there any other
messages posted in the max window about using a software renderer?

A few random questions:

- Do other OpenGL patches work fine?

- Does FSAA work? (jit.window @fsaa 1)

- Does typing the following into a message box and clicking it result
in a functional jit.gl.slab.help? ("; jitter glreadback fbo", which
is the message to use an FBO rather than a pbuffer backend for each
slab rendering pass).

We may need to try and get our hands on this HW to better track down
these particular issues.

-Joshua


August 28, 2006 | 9:30 am

Quote: jkc wrote on Sun, 27 August 2006 16:30
—————————————————-

> Interesting. There might be some aspect of using these cards in this
> configuration that actually prevents using the appropriate driver for
> these cards, and it might be falling back on the software based
> renderer, or some other more basic OpenGL driver. Are there any other
> messages posted in the max window about using a software renderer?

No, no other messges in the Max window.

>
> A few random questions:
>
> – Do other OpenGL patches work fine?

As far as I tested all works fine excluding jit.gl.slab.

>
> – Does FSAA work? (jit.window @fsaa 1)

No error messages when using the jit.gl.videoplane.help
and sending fsaa 0/1 messages to jit.window.

>
> – Does typing the following into a message box and clicking it result
> in a functional jit.gl.slab.help? ("; jitter glreadback fbo", which
> is the message to use an FBO rather than a pbuffer backend for each
> slab rendering pass).

No, unfortunately not. The following error messages appear:

? error: jit.gl.fbo: unsupported framebuffer format!
? error: jit.gl.fbo: unsupported framebuffer format!
? error: jit.gl.fbo: unsupported framebuffer format!
? error: jit.gl.fbo: unsupported framebuffer format!
? error: jit.gl.fbo: unsupported framebuffer format!
? error: jit.gl.fbo: unsupported framebuffer format!
? error: jit.gl.fbo: unsupported framebuffer format!
? error: jit.gl.fbo: unsupported framebuffer format!
? error: jit.gl.texture: error updating readback mechanism for capture!

>
> We may need to try and get our hands on this HW to better track down
> these particular issues.

Would be great as currently multi-screen work with jit.gl.slab is completely broken here.

Ah, and I forgot to mention that it?s the same behaviour with the 1.5.3 beta version.

Thanks

Bernd

OSX 10.4.7, G5, QT 7.1.2, Max/MSP 4.6.1/Jitter 1.6.1


September 8, 2006 | 2:06 pm

We’ve got the same problem here. Just upgraded to jitter 1.6. A patch that worked fine in 1.5 gives the following error:

? error: jit.gl.pbuffer: unable to create pbuffer context!
? error: jit.gl.texture: error attaching texture to readback mechanism for capture!

after sending ; jitter glreadback fbo, we get

? error: jit.gl.fbo: unsupported framebuffer format!
? error: jit.gl.fbo: unsupported framebuffer format!
? error: jit.gl.fbo: unsupported framebuffer format!
? error: jit.gl.fbo: unsupported framebuffer format!
? error: jit.gl.fbo: unsupported framebuffer format!
? error: jit.gl.fbo: unsupported framebuffer format!
? error: jit.gl.fbo: unsupported framebuffer format!
? error: jit.gl.fbo: unsupported framebuffer format!
? error: jit.gl.texture: error updating readback mechanism for capture!

The setup is:
Quad G5 2,5 GHz (PPC), 1x GeForce 7800 (two of two monitors connected), 1x GeForce 6600 (one of two monitors connected), OSX 10.4.7, Max 4.6.1, Jitter 1.6.1

Regards,
Mattijs


September 8, 2006 | 3:38 pm

On Sep 8, 2006, at 7:06 AM, Mattijs Kneppers wrote:

>
> We’ve got the same problem here. Just upgraded to jitter 1.6. A
> patch that worked fine in 1.5 gives the following error:

Thanks for the report. We’ll try to solve this one in the near
future, once we get some HW to test on.

-Joshua


October 2, 2006 | 9:33 am


hello,
i saw a posting from late august into september that addressed the same issue,

http://www.cycling74.com/forums/index.php?t=msg&goto=74345&rid=3287&S=3fe129ab65b1094c47bac3c7e45f9f44&srch=%22error+binding+invalid+pbuffer%21%22#msg_74345

the issue being the following errors in a setup with 4 screens:

jit.gl.texture: error binding readback mechanism for capture!
jit.gl.pbuffer: error binding invalid pbuffer!

my symptoms are just like those reported by bernd and others; i.e., i have a patch that works fine with one two gforce 6600′s installed in a quad G5, but only so long as nothing is connected to the 2nd graphics card. as soon as i connect a monitor to the second graphics card, i get lots of errors and i only appear to get one of my openGL context rendered.

did anyone come up with a solution or a work around for this?

thank you,

ali


October 2, 2006 | 5:16 pm

On Oct 2, 2006, at 2:33 AM, Ali Momeni wrote:

> the issue being the following errors in a setup with 4 screens:
>
> jit.gl.texture: error binding readback mechanism for capture!
> jit.gl.pbuffer: error binding invalid pbuffer!

We are actively investigating this issue. The current workaround is
to send jitter the message "; jitter glreadback fbo" to use the frame
buffer object backend. Unfortunately with multiple graphics cards
there seems to be trouble sharing resources with any windowed
context. Not only window contexts which span multiple cards.

The frame buffer object backend isn’t perfect, but largely works
without problem. You can put the above line inside a file called
jitter-config.txt which resides in the search path if you want this
to always be the case.

-Joshua


October 3, 2006 | 11:50 am

hi joshua,
like the previous posters,
when i tell jitter glreadback fbo,
i get this error:

error: jit.gl.fbo: unsupported framebuffer format!
error: jit.gl.fbo: unsupported framebuffer format!

error: jit.gl.fbo: unsupported framebuffer format!
error: jit.gl.fbo: unsupported framebuffer format!
errir: jit.gl.texture:error updating mechanism for capture!

and now, i don’t know what glreadback mechanism was the default becuase with the current, some of my programmable shaders no longer work at all :(

i like all the !’s in the errors,
as if it’s as surprised as i am heehee.

a


October 3, 2006 | 12:14 pm

just a note to say, i learned from the release notes that the default for glreadback is ‘rtt’ so i changed it back to ‘rtt’ and now my shaders work, but still lots of errors. in fact, aside from the errors, i don’t see any harm; all screens works; did the others also find that the only problem is the posted errors?

a


October 3, 2006 | 4:30 pm

Hi Ali,

Fix for this is in the next version (finally figured out the real
problem). I also hope to solve the jitter glreadback fbo issues
claiming to not be supported. I was able to solve this problem with
1.6.2 by sending the message sequence depthbuffer 0, depthbuffer 1 to
jit.window to force a rebuild, however. I’m surprised that sending
the message glreadback rtt made things work for you, as I didn’t see
such behavior as long as screens were attached to both cards and the
readback mechanism was rtt.

So the recommended steps if you have multiple graphics cards on the
Mac until we get 1.6.3 out the door is:

- "; jitter glreadback fbo"
- "depthbuffer 0, depthbuffer 1"->jit.window (or toggle off if you
don’t want one)

-Joshua


October 3, 2006 | 5:23 pm

On 03.10.2006, at 18:30, Joshua Kit Clayton wrote:

> Fix for this is in the next version (finally figured out the real
> problem).

ah, i’ll follow yacine’s example as in
> hip hip hurray! I didn’t notice spigot~ was back from the dead.
> for those in Brussels, I guess I m going to open a champagne bottle
> to celebrate.
> i am so happy ;)

- but it will be in groningen, and it will just be sekt instead of
champagne!! (well, i also have some beer, gary;-)
THANKS!!!!!

jan


October 3, 2006 | 5:49 pm

On Oct 3, 2006, at 10:23 AM, Jan Klug wrote:

> – but it will be in groningen, and it will just be sekt instead of
> champagne!! (well, i also have some beer, gary;-)

Most disgusting drink award goes to Ben who, while in Berlin, somehow
through miscommunication ordered sekt with tequila. Ick! Everytime I
think about sekt I’m reminded of this horrible combo.

> THANKS!!!!!

But of course.

-Joshua


Viewing 17 posts - 1 through 17 (of 17 total)