Announcement

Collapse
No announcement yet.

Intel Makes First Release Of Linux OpenCL Project

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

  • Intel Makes First Release Of Linux OpenCL Project

    Phoronix: Intel Makes First Release Of Linux OpenCL Project

    While Intel has previously shipped its OpenCL SDK for Linux and Windows, this SDK is closed-source and on Linux was limited to compute support only on the processor rather than any graphics support with Ivy Bridge and newer hardware. Fortunately, Intel has finally managed to put out a first release of Beignet, an open-source Linux project that supports OpenCL...

    http://www.phoronix.com/vr.php?view=MTM1Mjc

  • #2
    redhat

    red hat dev team is a cancer on linux, they want all people do what they want. everething is bad for them even the good things

    Comment


    • #3
      Originally posted by rikkinho View Post
      red hat dev team is a cancer on linux, they want all people do what they want. everething is bad for them even the good things
      Sorry, but no. Intel is writing an LLVM-based Ivy Bridge compute system for OpenCL. We already have one in Mesa that they could extend.... it even uses LLVM as a back-end, and we've implemented part of the OpenCL run-time library using OpenCL C and some LLVM Assembly.. The libclc/clover projects already support multiple architectures and architecture-specific overrides when necessary.

      Why wouldn't Intel re-use the existing code that we already have and improve/extend that? I agree with Dave completely.

      Comment


      • #4
        NIH syndrome? Different liscenses? Control. Going out of the way to avoid helping competing cards any more than they have to?

        Comment


        • #5
          Originally posted by rikkinho View Post
          red hat dev team is a cancer on linux, they want all people do what they want. everething is bad for them even the good things
          Huh? You must be very pleased to see different vendors' libOpenCL fighting with each other at /usr/lib(64).

          Comment


          • #6
            Originally posted by rikkinho View Post
            red hat dev team is a cancer on linux, they want all people do what they want. everething is bad for them even the good things
            did your mom turn back on your Internet for an hour?

            Comment


            • #7
              Are you at least cooping in regard to piglit tests? (or whatever you use for unit testing) (you - mesa/intel folks alike)

              Comment


              • #8
                From the mainling list

                Zou, Nanhai (Intel OpenCL developer):
                "We can merge or hook the backend into gallium when there are other mature gallium based OpenCL implementations. "

                WTF intel?!? You said the same for Gallium3D. Now there are r600g and radeonsi, and you still stay at "classics"!

                Comment


                • #9
                  Merge already ?

                  Maybe it would be good time, to merge Beignet into Gallium3D/Clover *right now*, before its too late ?

                  Beignet is open source (and I suppose their respective license won't prevent merging).
                  Beignet is still in its infancy (so that won't require the massive work that a merge of the 3D driver from classics into gallium3d requires today).

                  Clover is also under development so it is still easier to adapt if some changes are needed to make it better multi-architecture ?

                  Better do it now, than when it's to late...

                  ...hmmm, too bad that GSoC's projects are set.

                  Can Xorg's equivalent (EVoC) finance such a thing ?

                  Comment


                  • #10
                    Originally posted by DrYak View Post
                    Maybe it would be good time, to merge Beignet into Gallium3D/Clover *right now*, before its too late ?

                    Beignet is open source (and I suppose their respective license won't prevent merging).
                    Beignet is still in its infancy (so that won't require the massive work that a merge of the 3D driver from classics into gallium3d requires today).

                    Clover is also under development so it is still easier to adapt if some changes are needed to make it better multi-architecture ?

                    Better do it now, than when it's to late...

                    ...hmmm, too bad that GSoC's projects are set.

                    Can Xorg's equivalent (EVoC) finance such a thing ?
                    Yes. EVoC can finance such things as long as some core dev is willing to mentor such work. Though EVoC is not about financing work on Gallium3D/Mesa/LLVM, but about giving incentive to students for engaging into FLOSS.

                    Comment


                    • #11
                      As the previous developper of it, there is really no evil plan here from Intel. I just did it the way it is because it was more convenient for me and it was self-contained. Actually, the code actually uses the Nvidia PTX 32 bits ABI / Triple from LLVM :-) and uses a module pass instead of a full by-the-book backend because it was easier.
                      The goal was really to have something small and not too hard to write and that I can use for other purposes than pure compilation (i.e. HW architecture analysis)

                      Cheers,
                      Ben
                      Last edited by bouliiii; 04-16-2013, 08:36 AM.

                      Comment


                      • #12
                        Considering the future of GPU's it's a good thing to have a driver that has both OpenGL and OpenCL.
                        Especially since with OpenGL 4.3 you can use OpenCL and OpenGL together.

                        Seems quite the case of not invented here syndrome.

                        Comment


                        • #13
                          "Why does this exist anymore?"

                          Probably because the Gallium OpenCL support is in a very, very early stage and didn't really exist back when Intel began working on OpenCL. Also, it's often easier to go your own way instead of constantly arguing with developers, which is typical in community projects. Just saying, it's not only NIH.

                          Comment


                          • #14
                            ...

                            Originally posted by zxy_thf View Post
                            Huh? You must be very pleased to see different vendors' libOpenCL fighting with each other at /usr/lib(64).
                            I do not live in the basement of my mother's house as you piece of shit, if you do not like to hear truths is your problem

                            Comment


                            • #15
                              ....

                              Originally posted by airlied View Post
                              did your mom turn back on your Internet for an hour?

                              I do not live in the basement of my mother's house as you piece of s****, if you do not like to hear truths is your problem

                              Comment

                              Working...
                              X