Announcement

Collapse
No announcement yet.

New Mesa Patch To Improve CPU-Bound Applications

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

  • #11
    Originally posted by dungeon View Post
    And who said i don't do that, if i have time Thing is i alone don't have a time to bisect and filling bugs about everything.
    For a start, I think it would be great if you posted what/how you measured.

    Comment


    • #12
      Originally posted by nanonyme View Post
      You could probably get some dev bisect for you if you send them your computer. More seriously, the problem might be specific to a single chipset which devs do not necessarily have available. If you want any guarantee of a fix, you need to contribute time
      I contributed my time as a user more then enough, right now i discuss something other on bugzilla :... well not this one, i am waiting for llvm scheduler to be turned on and if performance is still lower then mesa at begining of this month i will probably bisect that .

      Comment


      • #13
        Originally posted by dungeon View Post
        I contributed my time as a user more then enough, right now i discuss something other on bugzilla :... well not this one, i am waiting for llvm scheduler to be turned on and if performance is still lower then mesa at begining of this month i will probably bisect that .
        like this one ? http://llvm.org/viewvc/llvm-project?...evision=227461

        Comment


        • #14
          Originally posted by log0 View Post
          For a start, I think it would be great if you posted what/how you measured.
          Sadly we already know.

          Originally posted by dungeon View Post
          Benchmarking by eye GALLIUM_HUD=cpu,fps

          Comment


          • #15
            Originally posted by tarceri View Post
            Sadly we already know.
            Why not that is fine enough for on eye benchmarking, only sadly unreproducable on intel because it does not using gallium.
            Last edited by dungeon; 29 January 2015, 05:09 PM.

            Comment


            • #16
              Originally posted by tarceri View Post
              Sadly we already know.
              He, he, and please stop trolling about that ... because even Kristian does that:

              Code:
              Most CPU bound cases are helped and some of our internal
              bind_buffer_object heavy benchmarks gain up to 10%.
              He does something internaly and i does something internally - we are equal Do you know with what/how we measure, you don't
              Last edited by dungeon; 29 January 2015, 05:42 PM.

              Comment


              • #17
                Originally posted by dungeon View Post
                He, he, and please stop trolling about that ... because even Kristian does that:

                Code:
                Most CPU bound cases are helped and some of our internal
                bind_buffer_object heavy benchmarks gain up to 10%.
                He does something internaly and i does something internally - we are equal Do you know with what/how we measure, you don't
                What part of this statement his him saying he uses gallium hud and checks by eye? This sounds more like Mesa has it's own benchmarking test suite (which it should have) and the numbers were coming out to up to 10% improvement.

                Comment


                • #18
                  Originally posted by Luke_Wolf View Post
                  What part of this statement his him saying he uses gallium hud and checks by eye? This sounds more like Mesa has it's own benchmarking test suite (which it should have) and the numbers were coming out to up to 10% improvement.
                  Tarceri seems offended by me previosly with my comments about his SIMD optimizations which for sure does not give any gain in mesa and usually leave sarcastic comments whenever he see me in the thread, but what should i do about that it has a right to feel how he like

                  And about those virtual up to tha 10% we don't know much about, only we can happily assume when someone from intel said "some of our internal heavy benchmarks gain up to 10%" that besides that is publicly unknown benchmark also usually means using intel compiler, so most of the usuall distribution users won't see that gain - it is more likely around 1.5% that is what i said in first post of this thread, etc...
                  Last edited by dungeon; 29 January 2015, 06:56 PM.

                  Comment


                  • #19
                    Originally posted by dungeon View Post
                    Oh thank you, people can't say a word about their experience it seems.
                    "Oh, I'm sorry. I thought this was AMERICA!" </RandyMarsh>

                    Comment


                    • #20
                      Wait, does Kristian imply here that futexes have too much overhead? On ARM Cortex A9 I recall the uncontended futex lock case to be around 260 cycles (vs 1600 for kernel-arbitrated mutexes). What insane amount of locks are taken that a reduction of 100 cycles max. per lock/unlock op results in a 10% speed improvement? Or was Mesa bypassing the regular futex codepaths?

                      Comment

                      Working...
                      X