Hello everyone,

I'm working on a rasterizer that renders triangles in 2d space, all additively (no depth checking). I'm looking for tips on performance optimization of all types. I don't have a lot of experience with OpenCL/memory management etc so I'm very happy to learn about all the ways to impro…

There is no real technical difference between gfx ‘shaders’ and GPGPU 'compute', both use the same cores and will equally benefit from higher frequency. 
Dedicated gfx HW is mainly the non programmable stuff like ROPs or TMUs, and now RT, and those units usually also depend on the same core clo…

I would highly highly highly recommend Game Code Complete 4! It goes over many many aspects of game engine creation in C++ (Audio, Resource Cache, Rendering, Physics, Input, Scripting and then some). On top of that, it discusses making a game editor in C#. It's the best book I've seen for game engi…

Today, Collabora is excited to announce a partnership with Microsoft to build OpenCL and OpenGL mapping layers on DirectX, in order to bring OpenCL 1.2 and OpenGL 3.3 support to all Windows and DirectX 12 enabled devices. Support for OpenGL is realised through the Mesa3D project's Gallium layer.

htt…

Thanks for the help. I got to the bottom of the problem

The issue was not with the texture coordinates but with multisampling or lack thereof when rendering off-screen.

I hadn't set up my framebuffer configuration to allow for multisampling when rendering to the texture, so the edges of the sh…

Yep, and i think they are as well limited to memory sizes, like 4GB or so, which would be an RGB image of around 30.000/30.000. Also, users of GIMP or PS accept long waiting times for operations to execute. Like, hours ? ? The latest graphics cards are actually better than that (32k textures), b…

jhbk

Hi @_Flame_, I'm glad that you have the decision

This is the downside of OpenCL, it needs to have most of the data you use passed as kernel arguments. The constant buffers can only be used for pre-defined tables, you don't really write into them.

Most of the time you should be okay with few dozen kernel arguments (I don't think we ever got…

FusionCL.Net is a new wrapper around OpenCL for C#(4.7.2) -

The git is online, and it is working. It's not 100% complete but you can create programs, run kernels, create membuffers and read/write them. It works fine for me on a Nvidia 1060.

There are two projects on the git repo.

Fus…

25 minutes ago, Alex Garcin said:

OK so how should I do to know which of the two contexts is the one used by opengl? 

I have never had same task.. I think you should create separate OpenCL context for each OpenGL context. 

Usually,  the different OpenGL contexts can be…

My bet would be either the sync problem, or you are creating new  buffers each frame, or you process bigger amount of data instead of one frame per draw? Or even you somewhere add some kind of information like (this is an object 1, save it for further processing) now in another frame you add x mo…

Hello.

What would be the best way to handle such error of enqueue_kernel, let's say for sort algorithm? Loop until it returns success?

One problem with this approach that parent kernels can grab all resources and hence loops will be infinite.

35 minutes ago, Andrey OGL_D3D said:

Have you done it without CL<->GL/D3D interop ?

So, in some cases OpenCL can be faster than VK/GL/D3D compute shaders in case where we needn't share resources ?

Yes, i have never used interop, used graphics just to visual…

When scaling, only the number 0 is used as an anchor. Your initial position of (1.0, 1.0) was multiplied by the scaling factor of (image width, image height), resulting in the image being translated by that amount. For your vertices, have the position coordinates be centered around 0, and not 1.…

I tried alternative suggested but no success.

I soon got to learn from this Stack Overflow thread that there is no direct key binding for GLCanvas which is the canvas JOGL uses. This suggests why the key binding for my code in the original post failed. In the same thread there was a hint (li…

I have another problem. While resizing, view updates lag behind, so there is white (MacOS 10.8) or black (MacOS 10.12) frame near the border which I'm resizing. I'm using WindowDidResize event, is this correct? I've also tried StartLiveResize and EndLiveResize, but it's even worse.
Oh, and I loo…