Announcement

Collapse
No announcement yet.

Early OpenCL 2.0 Specification Brings New Features

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Early OpenCL 2.0 Specification Brings New Features

    Phoronix: Early OpenCL 2.0 Specification Brings New Features

    Besides publishing the OpenGL 4.4 specification today, the Khronos Group also published their provisional draft of the OpenCL 2.0 specification for GPGPU comptuing. OpenCL 2.0 brings several new compute-focused features over OpenCL 1.2...

    http://www.phoronix.com/vr.php?view=MTQxNzM

  • #2
    That's great to hear.

    Any information available which GPUs have the required capabilities to support OpenCL 2.0?

    Comment


    • #3
      Have no fear! Nvidia will sit on 1.1 for eternity and keep throwing up CUDA as if the rest of the world wants it.

      Comment


      • #4
        Originally posted by Marc Driftmeyer View Post
        Have no fear! Nvidia will sit on 1.1 for eternity and keep throwing up CUDA as if the rest of the world wants it.
        Well, I'm more concerned about the currently available GCN ASICs by AMD.
        You might have a point about NVidia. Their OpenCL implementation is more an alibi than anything else.

        Comment


        • #5
          Originally posted by Marc Driftmeyer View Post
          Have no fear! Nvidia will sit on 1.1 for eternity and keep throwing up CUDA as if the rest of the world wants it.
          But they do.

          Comment


          • #6
            Originally posted by Marc Driftmeyer View Post
            Have no fear! Nvidia will sit on 1.1 for eternity and keep throwing up CUDA as if the rest of the world wants it.
            If you've ever written algorithms in both CUDA and OpenCL, it's very easy to see why the world would want CUDA ...

            Comment


            • #7
              Originally posted by nslay View Post
              If you've ever written algorithms in both CUDA and OpenCL, it's very easy to see why the world would want CUDA ...
              The world isn't buying into CUDA, now that OpenCL is here to stay. All one has to do is follow the adoption rate by the big companies where hundreds of billions in consumer dollars are the market.

              Don't take my word for it. Take Nvidia's:
              “The OpenCL working group has combined developer feedback with emerging hardware capabilities to create a state-of–the-art parallel programming platform - OpenCL 2.0,” said Neil Trevett, chair of the OpenCL working group, president of the Khronos Group and vice president of mobile content at NVIDIA. “OpenCL continues to gather momentum on both desktop and mobile devices. In addition to enabling application developers it is providing foundational, portable acceleration for middleware libraries, engines and higher-level programming languages that need to take advantage of heterogeneous compute resources including CPUs, GPUs, DSPs and FPGAs.”
              Sorry, but Nvidia will be moving to OpenCL sooner rather than later leaving their own in-house proprietary stuff behind, just like AMD did.

              Comment


              • #8
                Originally posted by nslay View Post
                If you've ever written algorithms in both CUDA and OpenCL, it's very easy to see why the world would want CUDA ...
                Ever since NVIDIA bribed some Academics to use CUDA, we hear the same flawed argument...

                Seriously, you state that opinion on a LINUX forum?

                With your logic, using Windows is more user-friendly, why use Linux/BSD?

                NVIDIA will eventually have the fate of 3dfx. Both wanted to create a monopoly.

                Comment


                • #9
                  Originally posted by TemplarGR View Post
                  Ever since NVIDIA bribed some Academics to use CUDA, we hear the same flawed argument...

                  Seriously, you state that opinion on a LINUX forum?

                  With your logic, using Windows is more user-friendly, why use Linux/BSD?

                  NVIDIA will eventually have the fate of 3dfx. Both wanted to create a monopoly.
                  You're right, CUDA is easier to develop with, much easier. But CUDA supports more C/C++ language features than OpenCL does (for now). Both of these reasons contribute to CUDA's popularity.

                  Let's twist "my logic" into:
                  With your logic, Linux has more features and support, why use BSD?
                  Although, I'm a BSD guy myself

                  Comment


                  • #10
                    Originally posted by Marc Driftmeyer View Post
                    The world isn't buying into CUDA, now that OpenCL is here to stay. All one has to do is follow the adoption rate by the big companies where hundreds of billions in consumer dollars are the market.

                    Don't take my word for it. Take Nvidia's:

                    Sorry, but Nvidia will be moving to OpenCL sooner rather than later leaving their own in-house proprietary stuff behind, just like AMD did.
                    I wasn't aware that AMD ever used anything other then OpenCL. Their entire HSA architecture is based upon maximizing OpenCL performance by eliminating the performance hits incurred by having to do multiple copies to and from CPU and GPU memory over the PCI bus by having both have direct access to the same memory pool.

                    Now all we need is for OpenCL to take off in the OSS drivers so we start seeing it get picked up by every OS project.

                    Need the Gallium3D guys to hammer out multiple GPUs, gaming would be great to get an AMD APU to do all of the physics and AI processing and a fast dedicated GPU pushing 3-5 screens.

                    Comment

                    Working...
                    X