How to understand"CUDA support OpenCL"?

I found NV announced these days that CUDA will support OpenCL, but I’m little confused how to understand this sentence, I list three possible meanings ,in my thought, below, please help to check which one is reasonable.

  1. There’ll be OpenCL included in CUDA as an extension of CUDA Packet.
  2. Cuda will change it language to fully identical to OpenCL, in other words, there will be no CUDA any more.
  3. Cuda will transform it’s NVCC to OpenCL compiler,but also support current CUDA language.

Any of your comments will be appreciated

1.OpenCL looks like CUDA driver programing.

2.OpenCL is the protocol, which almost content all the CUDA’s driver API.

but, how do I use OpenCL under CUDA?

Will OpenCL release an OpenCL compiler later? Consider the benefits of OpenCL, does this mean CUDA would disapear in the GPU market?

Please give me a reason to let me have confidence in using CUDA.

any one can help?

1.OpenCL is only a protocol in this version, don’t have memory control platform, don’t have thread control platform…just some APIs…in which device the driver should have their own works.

2.If you want to make a optimal Nvidia GPU programing, you should understand CUDA model very well.

3.In the OpenCL examples, we can see…They are almost the CUDA programing…

Anyway, in the future, the Function programing, new version C++, python, haskell…more and more program language may support parallel programing…

Which one will you choose?

first, we should know device…Nvidia GTX2xx, AMD r7xx, Intel Lrb…

then may be we can write optimal program…

so nvidia’s announcement means cuda’s compiler will compiler openCL’s API also? what if openCL release it’s compiler in the near future?

Please link to the anouncement.

NVIDIA will support OpenCL, but I’ve never seen that it will be part of CUDA. They’ve said nvcc won’t compile OpenCL. Possibly someone else will write the compiler and most libraries, but NVIDIA will do the low-level stuff of making OpenCL work with its architecture.

Thanks. But does that mean cuda will fade out this maket by saying nvcc won’t compile OpenCL ?

The anouncement: http://www.embedded-computing.com/news/db/?14618

you can find the anouncement at: http://biz.yahoo.com/prnews/081208/aqm116.html?.v=61

BTW, does that mean CUDA will fade out the market by saying “nvcc won’t compile OpenCL”

If you would simply read the other OpenCL post on this forum (just a few threads down from yours), you would learn a lot more than by asking the same question over and over again here.

Since I doubt you’ll actually pull it up, here is a quote from Tim Murray at NVIDIA which indicates CUDA will be further supported and could potentially have more features added to it than OpenCL supports.

Thank you so much. your advice is very helpful

For OpenCL I needed to download whole CUDA SDK 4.2.9 as sschaetz’s git does not provide shrUtils.h
and CUDA 5.0 does not contain ocl at all :(
Explanation in http://streamcomputing.eu/blog/2012-09-10/nvidias-industry-leading-support-for-opencl/
It complied and run on Fedora 18 x64 and Fedora 16 x64 just fine.
OCL examples here for your convenience :
http://uloz.to/xfwNX2K/nvidia-opencl-examples-cuda-4-2-9-sdk-zip
yum -y install gcc libXmu-devel