Announcement

Collapse
No announcement yet.

Intel Open-Sources LLVM Graphics Compiler, Compute Runtime With OpenCL 2.1+

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

  • #11
    Originally posted by andre30correia View Post

    since they have a open driver the things will work with opengl and vulkan if gpu can do that. I have a sandybridge too and it only have ogl 3.3 because it can be more than that, for example Apple make their on driver and intel sandy only have 3.2 and windows 3.1. The support in linux by intel is good
    This discussion is about OpenCL, not OpenGL.

    Comment


    • #12
      Originally posted by willmore View Post
      While I appreciate the effort to bring more advanced OpenCL support to Intel IGPs, I can't help but feeling distressed that every time they make an advance, they reset the processor generation it's attached it. It makes it seem like older chips will never be supported. And by older chips, I mean anything but the current newest generation.
      It often doesnt make a lot of sence to have your primary developers focused on the past. By the time something like this is stabilized the hardware they have targetted will be legacy anyways.

      In any event it is great to see yet another LLVM project come to light. LLVM has become like the Linux kernel with its wide adoption.

      Comment


      • #13
        Originally posted by Jedibeeftrix View Post
        Will Neo support be broadened to include Broadwell in later updates, or this Arch (and earlier) left on Beignet/Closed-Source?
        Broadwell support is already enabled in this driver stack, albeit for OpenCL 2.0.

        Comment


        • #14
          Originally posted by andre30correia View Post
          ok compile fails to me
          Could I ask you to file an issue on our project page @ github? We would like tu understand what is not working for you.

          Direct link: https://github.com/intel/compute-runtime/issues

          #IamIntel

          Comment


          • #15
            Originally posted by AdamCetnerowski View Post

            Could I ask you to file an issue on our project page @ github? We would like tu understand what is not working for you.

            Direct link: https://github.com/intel/compute-runtime/issues

            #IamIntel

            Hi Adam, welcome to the forum!

            Which Intel projects are you involved? So we can tag you, like we do to Kayden when related to graphics or Arjan when related to Clear Linux

            Comment


            • #16
              Originally posted by AdamCetnerowski View Post

              Could I ask you to file an issue on our project page @ github? We would like tu understand what is not working for you.

              Direct link: https://github.com/intel/compute-runtime/issues

              #IamIntel
              I will try again and see if something is wrong with my config and if persists I report

              Comment


              • #17
                "Neo" that is meant to replace both Beignet and their closed-source/classic OpenCL driver.
                So you mean Neo is The One?
                Cool. Everyone loves mister Anderson.

                Comment


                • #18
                  Originally posted by wizard69 View Post

                  It often doesnt make a lot of sence to have your primary developers focused on the past. By the time something like this is stabilized the hardware they have targetted will be legacy anyways.

                  In any event it is great to see yet another LLVM project come to light. LLVM has become like the Linux kernel with its wide adoption.
                  It's a great thing that OpenCL is getting more attention. But I am not sure LLVM being wildly deployed is a good thing to myself. Since nowadays GPU compute is well integrated into mobile platform, BSD license allows vendors redistribute software in closed form. It's getting harder and harder to get a controllable mobile now.

                  Comment


                  • #19
                    Originally posted by wizard69 View Post
                    It often doesnt make a lot of sence to have your primary developers focused on the past. By the time something like this is stabilized the hardware they have targetted will be legacy anyways.
                    I have to disagree. CPUs & iGPUs haven't made a lot of progress, in recent years. This means that, more than ever before, there's a lot of older CPUs kicking around.

                    I've still got two SandyBridge i7's and hadn't even considered upgrading until Coffee Lake or Zen+. But with Meltdown and Spectre, I think I'll just sit tight and wait for a generation with some hardware fixes (maybe Zen 2).

                    Comment


                    • #20
                      Originally posted by trivialfis View Post
                      It's a great thing that OpenCL is getting more attention.
                      I wish Raspberry Pi would hurry up and release something based on VidCore V. I think that'll be a game-changer for OpenCL adoption, with so many people using it for robotics (computer vision) and machine learning.

                      Comment

                      Working...
                      X