glmark2 errors

glmark2 errors

[John Weber]

I'm testing master-next with core-image-x11 on Wandboard (Solo, Dual, Quad).

glxgears and es2gears_x11 test out fine.

The GL benchmark glmark2 throws the following warnings reapeatedly:

** GLX does not support GLX_EXT_swap_control or GLX_MESA_swap_control!
** Failed to set swap interval. Results may be bounded above by refresh rate.

It will eventually segfault during the 'shadow' benchmark.

See log here (for Wandboard Dual, with 1080P resolution screen):

http://pastebin.com/SnyPvzsK

glmark2_es2 works beautifully.

Thanks,
John

Re: glmark2 errors

[John Weber]

Hi all -

On 4/28/14, 1:40 PM, John Weber wrote:
> I'm testing master-next with core-image-x11 on Wandboard (Solo, Dual, Quad).
>
> glxgears and es2gears_x11 test out fine.
>
> The GL benchmark glmark2 throws the following warnings reapeatedly:
>
> ** GLX does not support GLX_EXT_swap_control or GLX_MESA_swap_control!
> ** Failed to set swap interval. Results may be bounded above by refresh rate.
>
> It will eventually segfault during the 'shadow' benchmark.
>
> See log here (for Wandboard Dual, with 1080P resolution screen):
>
> http://pastebin.com/SnyPvzsK
>
> glmark2_es2 works beautifully.
Also note that this problem exists before the latest patch series by Lauren was 
merged into master-next (i.e. in 3.10.17_1.0.0_beta as well).
>
> Thanks,
> John

Re: glmark2 errors

[Diego]

John Weber wrote:
> The GL benchmark glmark2 throws the following warnings reapeatedly:
> 
> ** GLX does not support GLX_EXT_swap_control or GLX_MESA_swap_control!
> ** Failed to set swap interval. Results may be bounded above by refresh
> rate.
> 
> It will eventually segfault during the 'shadow' benchmark.

Hi John,

as I previously reported plain OpenGL is not officially supported by Vivante 
drivers. While hardware supports OpenGL, binary drivers support it only 
unofficially, but there are no headers or explicitly documented support; and I 
think situation will continue to be that way.

See also:
https://bugzilla.yoctoproject.org/show_bug.cgi?id=5098
https://bugzilla.yoctoproject.org/show_bug.cgi?id=6083#c1

Bests,
Diego

Re: glmark2 errors

[Carlos Rafael Giani]

On 2014-04-29 08:51, Diego wrote:
> John Weber wrote:
>> The GL benchmark glmark2 throws the following warnings reapeatedly:
>>
>> ** GLX does not support GLX_EXT_swap_control or GLX_MESA_swap_control!
>> ** Failed to set swap interval. Results may be bounded above by refresh
>> rate.
>>
>> It will eventually segfault during the 'shadow' benchmark.
> Hi John,
>
> as I previously reported plain OpenGL is not officially supported by Vivante
> drivers. While hardware supports OpenGL, binary drivers support it only
> unofficially, but there are no headers or explicitly documented support; and I
> think situation will continue to be that way.
>
> See also:
> https://bugzilla.yoctoproject.org/show_bug.cgi?id=5098
> https://bugzilla.yoctoproject.org/show_bug.cgi?id=6083#c1
>
> Bests,
> Diego
>

That makes sense. Regular OpenGL just isn't very well suited for 
embedded GPUs. Thats why OpenGL ES exists.
Also, IIRC the Vivante libGL binaries just emulate it by using OpenGL ES 
calls internally.

Carlos

Re: glmark2 errors

[Prabhu S]

[-- Attachment #1: Type: text/plain, Size: 1291 bytes --]

depth buffer is not supported in the current opengl driver and supported in
GLES driver. It will be fixed in the next release.
GLX_EXT_swap_control extensions is not supported as well.

On Mon, Apr 28, 2014 at 1:42 PM, John Weber <rjohnweber@gmail.com> wrote:

> Hi all -
>
>
> On 4/28/14, 1:40 PM, John Weber wrote:
>
>> I'm testing master-next with core-image-x11 on Wandboard (Solo, Dual,
>> Quad).
>>
>> glxgears and es2gears_x11 test out fine.
>>
>> The GL benchmark glmark2 throws the following warnings reapeatedly:
>>
>> ** GLX does not support GLX_EXT_swap_control or GLX_MESA_swap_control!
>> ** Failed to set swap interval. Results may be bounded above by refresh
>> rate.
>>
>> It will eventually segfault during the 'shadow' benchmark.
>>
>> See log here (for Wandboard Dual, with 1080P resolution screen):
>>
>> http://pastebin.com/SnyPvzsK
>>
>> glmark2_es2 works beautifully.
>>
> Also note that this problem exists before the latest patch series by
> Lauren was merged into master-next (i.e. in 3.10.17_1.0.0_beta as well).
>
>
>> Thanks,
>> John
>>
>
> --
> _______________________________________________
> meta-freescale mailing list
> meta-freescale@yoctoproject.org
> https://lists.yoctoproject.org/listinfo/meta-freescale
>

[-- Attachment #2: Type: text/html, Size: 2227 bytes --]

Re: glmark2 errors

[John Weber]

Hi Carlos and Diego -

On 4/29/14, 1:56 AM, Carlos Rafael Giani wrote:
> On 2014-04-29 08:51, Diego wrote:
>> John Weber wrote:
>>> The GL benchmark glmark2 throws the following warnings reapeatedly:
>>>
>>> ** GLX does not support GLX_EXT_swap_control or GLX_MESA_swap_control!
>>> ** Failed to set swap interval. Results may be bounded above by refresh
>>> rate.
>>>
>>> It will eventually segfault during the 'shadow' benchmark.
>> Hi John,
>>
>> as I previously reported plain OpenGL is not officially supported by Vivante
>> drivers. While hardware supports OpenGL, binary drivers support it only
>> unofficially, but there are no headers or explicitly documented support; and I
>> think situation will continue to be that way.
>>
>> See also:
>> https://bugzilla.yoctoproject.org/show_bug.cgi?id=5098
>> https://bugzilla.yoctoproject.org/show_bug.cgi?id=6083#c1
>>
>> Bests,
>> Diego
>>
>
> That makes sense. Regular OpenGL just isn't very well suited for embedded 
> GPUs. Thats why OpenGL ES exists.
> Also, IIRC the Vivante libGL binaries just emulate it by using OpenGL ES calls 
> internally.
>
> Carlos
No problem, and thanks for the information.  Glmark2 was fully functional (no 
warnings) in older versions of the FSL BSP.  I run this benchmark as a way of 
checking the health of a given image build, so when I see warnings like this I 
wonder if there has been a regression.