Announcement

Collapse
No announcement yet.

NVIDIA Makes PhysX Open-Source

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

  • #21
    Originally posted by GruenSein View Post

    While that would be helpful, it seems almost like starting from scratch. Why not just contribute to Bullet Physics?
    Mainly existing game compatibility.

    Comment


    • #22
      This seems like little more than a token gesture, for the purpose of capturing another headline. As has already been mentioned, NVidia's history of open sourcing code leaves a lot to be desired.

      Comment


      • #23
        Originally posted by uxmkt View Post
        And nobody tested this. First thing that happens is that a gcc8 linux64 build dies at some point because of -Werror.
        It's possible that they just tested it with a different compiler/version that has different warnings enabled by default. The fact that 'Werror' is part of the build indicates that they at least take/took compiler warnings fairly seriously.

        I've run into this at work frequently enough after java updates. Upgrade to the latest java version, get another batch of warnings to fix due to a new warning type included in the jdk.

        Comment


        • #24
          Originally posted by TemplarGR View Post

          Didn't know about that. Still, it could use AVX(2)....
          Well cpu physX cloth have used AVX codepath since 2014... But it's like every physics simulations over there, some are running just fine without AVX while some does not even run well on cpu no matter what instruction is in use, but need more parallelism to run it well(thus GPU).

          Comment


          • #25
            Third page of comments and still nobody's mentioned that BSD license does not say a word about patents?

            Sure, the code is open-source, but they still can sue you for patent infringement. And guess what, Nvidia actually has patents on PhysX.

            Comment


            • #26
              Originally posted by GruenSein View Post
              While that would be helpful, it seems almost like starting from scratch. Why not just contribute to Bullet Physics?
              Because you might as well write your own physics library with hookers and blackjack.

              Comment


              • #27
                Originally posted by TemplarGR View Post

                Didn't know about that. Still, it could use AVX(2)....
                It already make use of AVX for certain solver, contrary to popular belief PhysX is not just competitive but actually faster than both Bullet and Havok (the Havok license prohibits the publication of performance comparisons).
                When doing such kind of consideration you have to keep Amdahl's law in mind, the improvement could be modest at the end and not worth the effort when there are better law hanging fruit to take

                Comment

                Working...
                X