Yes. As I mentioned earlier in the thread (August 27), I had filed a Khronos bug on this issue after realizing that this behavior would be problematic. We decided to simply remove the restriction...

Jakob,


I think I've root-caused it. It should be easy to fix.

It has nothing to do with the absence of barrier() in your shader, though you will need barrier() calls for more complex...

Yes, that's correct, though it would certainly be possible to create a NVIDIA extension to GLSL that relaxes this restriction.

The limitation on barrier() is inherited from tessellation control...

Jakob,

Thanks for the feedback.

I agree that the first issue (not treating gl_WorkGroupSize as a constant expression) looks like a driver compiler bug, where it is treating it as an "in"...

Sorry I'm behind on looking at these forum updates.

The error with #extension GL_NV_shader_buffer_load is a compiler bug that I was able to reproduce. The fix was an oversight in the logic...

Someone already spotted that, and Piers root-caused and fixed the bug. Thanks for the helpful report and reproducer. If you have more issues that you suspect are likely driver things, feel free to...

You're also welcome to send me bug reports. My NVIDIA email ID is the same as my user ID on these forums.

I've confirmed that this is a recent regression, and might only be in the OpenGL 4.3 beta driver. I'll try to get this into the next beta driver.

Yes; the extension is not supported on Fermi or earlier NVIDIA GPUs.

That's about right. The extension spec was published on the NVIDIA developer site this morning:

http://developer.download.nvidia.com/opengl/specs/GL_NV_bindless_texture.txt

I've submitted it...

@Chris Lux: I've dug into our GLSL compiler and root caused your issue related to the "binding" layout qualifier and have a fix undergoing testing. As I mentioned above, you may be able to get...

It is legal. The "binding" part isn't covered in the ARB_shader_image_load_store extension, because the binding feature for samplers and images is in ARB_shading_language_420pack.

I'll be...

No problem. I'm sorry for any difficulty this bug may have caused you.

Chris --

Your bug report has been brought to my attention. You didn't mention it in your report, but I assume that you are running with a pre-Fermi GPU? I was able to reproduce the failure with...

Yes, that is exactly the problem. From the GLSL 4.00 spec (p. 41):

To get an array of structures in a single buffer, you need to do something like:



struct Sector
{
vec4...

There shouldn't be any difference in BindSampler behavior on NVIDIA GPUs between hardware versions, GeForce/Quadro, or OpenGL context versions. The first 3.3 driver did erroneously accept TEXTURE0...

On the site it's listed as only supported on OpenGL 4.0 hardware. Is there any technical reason for not making this available on 3.3 hardware?[/QUOTE]

Yes, these formats require decompression...

That's correct. It has already been fixed and will appear in our next OpenGL driver release.



As far as I can tell, an implementation is not required to report unreferenced uniforms in...

It appears that this is my fault. I will be fixing this today and it should appear in NVIDIA's next OpenGL driver release.

We accept GL_TEXTURE0 because of the error in the "New Procedures and...

One of my colleagues at NVIDIA pointed me at this. I can't see anything obviously wrong from your description.

It sounds like your basic approach works, since you can capture 8 vec4's in...

This dimension limitation on the GeForce 7 GPUs is baked into the hardware. Texture address calculations are fairly complicated and the more address lines you have, the more chip area and design...

Crud, I forgot to comment on the crash issue you described...

From a quick scan of the relevant driver code, it appears that we are enforcing the wrong size limit for TEXTURE_3D targets. So we...

NVIDIA's GeForce 7 series GPUs do not support 3D textures larger than 512 texels in any dimension, regardless of the amount of available video memory. That should be true under both OpenGL and...

Yes, it looks like glGenerateMipmapEXT is doing the wrong thing for array textures in the 97.02. It's supposed to work by generating mipmaps layer-by-layer, but it doesn't. I've filed a bug report,...

Right, it seems that index offsetting is easy enough in such cases that it's only a minor annoyance. Perhaps a stronger argument might be to allow developers to unify OpenGL and D3D renderers, if...