Announcement

Collapse
No announcement yet.

SIGGRAPH 2016 Kicks Off Today: What Exciting News Awaits?

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

  • SIGGRAPH 2016 Kicks Off Today: What Exciting News Awaits?

    Phoronix: SIGGRAPH 2016 Kicks Off Today: What Exciting News Awaits?

    SIGGRAPH 2016 kicks off today in Anaheim, California! It will hopefully be an interesting week with news excepted from the likes of The Khronos Group and others...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    Unfortunately, I am not able to attend in person due to the extremely tight budget due to ad-blockers and the like that have prevented me from doing any conferences in person the past few years, but will be providing my coverage remotely.
    Oh come on Michael... Even as a premium member I'm getting tired of these passive-aggressive shots that you seem to do in almost all your articles! I'm following a *lot* of websites and you're the *only* one whining about ad-blockers that much. This kind of behavior won't make the viewer disable its ad-blocker, it will just my him angry at you.

    If you want more premium members, try to do less articles but more relevant ones. For instance, just listing the changes that you can find in the commit list isn't relevant. What's relevant is what the features will bring to the users, what is their impact and who would be interested by them. Same goes for the benchmarks... It's often just a copy/paste of graphs with some "X is better than Y here" pretty useless comments. 5 pages of that is quite unproductive in the end. We don't know why something worked or not, we just have the numbers thrown at us.

    So please, stop saying it's the ad-blocker's fault. First, don't consider that quantity = quality, then maybe try other business models. A lot of websites now use the premium-only model, that might work, but your articles needs to level up a lot, because right now you won't get a lot of members just by having a hundred of articles a day. I'm sorry, but that's not how it works.
    Last edited by Creak; 24 July 2016, 11:09 AM.

    Comment


    • #3
      I would like to see the following:
      - OpenGL 4.6 and introduction of new extensions
      - Vulkan news about future direction and features of Vulkan, no new release of Vulkan is fine to me.
      - Information about the Vulkan-Hpp C++ API announcement.
      Sessions should include features, future plans, how interaction works with other languages such as Dlang, Go, Rust and maybe even Swift?
      More focus on higher level languages like C++ in addition to C is a huge step forward.
      I wonder how the C++ standard library rewrite with ranges and introduction of other features in C++17/20/2x will influence the Vulkan-Hpp C++ API.
      - OpenCL 2.2 information and features.
      - SYCL 2.2 information and features.
      Last edited by plonoma; 24 July 2016, 12:06 PM.

      Comment


      • #4
        I would like to see Sony's and Apple's adoption of Vulkan so that the only platform not on Vulkan would be the XBox from Microsoft. If that happens, I think we could see a major shift of the gaming industry toward embracing Vulkan by default instead of DirectX.

        Comment


        • #5
          Originally posted by plonoma View Post
          I would like to see the following:
          - I wonder how the C++ standard library rewrite with ranges and introduction of other features in C++17/20/2x will influence the Vulkan-Hpp C++ API.
          - OpenCL 2.2 information and features.
          - SYCL 2.2 information and features.
          + I think that the new C++ and STL features will not affect Vulkan-Hpp. At least not the one that they have published now. C++11/14 is widely adopted and is the only thing people can build upon today. C++17 is far from being done, not to mention it brings almost zero new features. See, Modules and Concepts, now those are two features that will greatly alter the way libraries are designed, but none of them made it into C++17, and I'm afraid Khronos will not build upon Technical Specifications. It will not be until 2020 that large or even medium projects will bump their major version number and adopt Concepts and provide Modules-aware libraries. (I'm particularly interested in Qt, how such a monolithic library will redesign itself, especially the Signal-Slot mechanism, which is it's quintessence. Will it be able to deprecate MOCing?)

          + Provisional specs to 2.2 are available on-line. It is a minor update, but that doesn't matter. It looks like nobody will adopt OpenCL 2.x, which is quite sad. Even AMD, one of the forefront adopters pretty much abandoned it and considers it a second-rate API in favor of HCC and HIP. (Extremely sad, really.)

          + Now, imagine the amount of work put into and the priority of the abstraction of an abstraction of an API that nobody implements. SYCL 1.2 is C++11 only and will remain that way, not even C++14 will be implemented. SYCL 2.2 would be kick@ss, but because there is nothing to build on (no functioning CL 2.2 until God knows when), it's quite hard to make a compiler against it.

          Comment


          • #6
            Originally posted by Creak View Post
            *Words*
            There is a consideration about perspective. While I don't agree about the little jabs, the site is operated in an atypical way compared to most other news sites. As a result, it doesn't make nearly as much from advertising I would have to guess than most other news sites, but it allows the site to be free to report on what they want, how they want. This site has the ability to "keep it weird" as they say. Most sites have to answer a lot more to advertisers as a tradeoff to being more profitable, and those news sites are disgustingly profitable, and the advertising even more so.

            As far as content, I personally don't mind the raw numbers benchmark articles but they can be a bit lacking in details. The summary text is almost unnecessary most of the time unless there is some actual idea on what the difference could be from, and I also don't mind the digest articles but again they can be a bit vague. Sometimes even clicking through to the relevant article can be a bit lacking in important details which often result in ending up picking through source code to determine what stuff even does.

            So this site has some shortcomings and definitely could be a bit more refined in some ways, but I can understand. It's not a big site with a huge team of writers and lots of money, so it has the potential to be more honest. I'm not making any commentary about that either way on this site, I really have no specific opinion there, but advertiser influence makes sites dishonest.

            As for the topic at hand, I am looking forward to what comes out. I hope there really isn't anything too crazy. The open source drivers need some time to catch up and honestly even the proprietary drivers are horribly behind on conformance. Software in general is far behind hardware and new "standards".

            Comment


            • #7
              as far as OpenGL goes. i would probably only want SPIR-V support in order to bring Vulkan and GL closer, everything else new should go in Vulkan

              Comment


              • #8
                Originally posted by Meteorhead View Post
                I'm particularly interested in Qt, how such a monolithic library will redesign itself
                Since Qt is a set of libraries, e.g. QtCore, QtGui, etc., did you mean each of them is monolithic?
                But aren't all libraries monolithic in that sense?

                Originally posted by Meteorhead View Post
                Will it be able to deprecate MOCing?)


                Cheers,
                _

                Comment


                • #9
                  Originally posted by paulguy View Post
                  So this site has some shortcomings and definitely could be a bit more refined in some ways, but I can understand. It's not a big site with a huge team of writers and lots of money, so it has the potential to be more honest. I'm not making any commentary about that either way on this site, I really have no specific opinion there, but advertiser influence makes sites dishonest.
                  The world isn't that black and white, it's not just the struggling independent sites against the big wealthy ones... When I was comparing Phoronix with alternatives, I wasn't thinking about these big sites. I value the independence of a sites, and I know and follow quite a few sites that are both independent and that seem to work pretty good for them.

                  Comment


                  • #10
                    Originally posted by anda_skoa View Post
                    Since Qt is a set of libraries, e.g. QtCore, QtGui, etc., did you mean each of them is monolithic?
                    But aren't all libraries monolithic in that sense?




                    Cheers,
                    _
                    Thanks for the tip.

                    Well, if even parts of Qt is not monolithic, then I don't know what is. Sure, it is modularized, but just one module builds for 10-15 minutes on a quad core machine. For example I have no idea why release builds of Qt don't use pre-compiled headers for itself? That would cut the compile times tenfold.

                    Comment

                    Working...
                    X