NI Vision and Graphics Card/ Hardware Acceleration

[Andrey_Dmitriev]

Hi, Shane,

What is difference between NVidia OpenGL drivers and NVidia CUDA technology?

I have already played with CUDA a little bit and it seems to be OK.

For example, if I want to calculate histogramm of the image - how it can be done with OpenGL?

With CUDA its pretty simple: - init CUDA->Upload image into GPU->Perform calc in GPU->Get results back from GPU->Close GPU.

Something similar possible with OpenGL, or here other ideology behind?

Andrey.

[Intaris]

It's Open C L not G L.

From what I have read OpenCl has adopted a lot from CUDA with the major advantage being that OpenCL being an OPEN standard as opposed to CUDA being an NVidia-only affair.

So ATI, Intel, NVidia and others will be able to utulise the ability of Open CL (Or rather, supply OpenCL compatible drivers for their hardware).

Shane.

Message Edited by Intaris on 04-24-2009 04:18 AM

[Andrey_Dmitriev]

Ah, just one letter and so much differences... Sorry, was confused with that.

Found the answer to my question above:

OpenCL 

OpenCL JumpStart Guide

Andrey.

Message Edited by Andrey Dmitriev on 04-24-2009 11:22 AM

[falkpl]

Are there plans for gpu processing support in labview.  This would be a game changer in advanced graphics proccessing.  I use the vision package often but am sometimes limited by cpu speed in the algorithms I develope.  I would love to see this feature.

[Intaris]

Well that's the great thing about OpenCL, it's a game-changer.

It provides a single interface for heterogenous processing whether the target it a video card, a custom-solution, a DSP or whatever.  Whatever has an OpenCL driver will be used by the system to compute.  Plug and play processing power.  I want some of that!

Shane.

[Kjell-E]

. . . and so would I! - It seems that NI might be cookin' something related to CUDA and NVIDIA, but I cant find any other reference to it than this recently published NI article.