Announcement

Collapse
No announcement yet.

Gallium3D's LLVMpipe Driver Is Now Much Faster

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

  • Gallium3D's LLVMpipe Driver Is Now Much Faster

    Phoronix: Gallium3D's LLVMpipe Driver Is Now Much Faster

    The Gallium3D LLVMpipe driver that's commonly used as the fallback software rasterizer on Linux desktop systems when no GPU hardware driver is present, is a heck of a lot faster with the current Mesa development code. The gains are surprising and quite remarkable.

    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
    as OpenGL isn't meant for running on a CPU.
    Can anyone predict how it will run on a parallela (the 64 core version)?

    Introduction The E64G401 is 64-core microprocessor/coprocessor reference design based on the 4th generation of the Epiphany multicore architecture that was designed as a direct replacement of the E…

    Comment


    • #3
      Such results makes me wanting to see desktop environments related performance results (are KDE and Gnome tests gone with latest versions ?)

      Comment


      • #4
        I guess it's a step in the right direction. I'm actually kind of surprised the frame rates are so low. I remember when quad cores came out and the talk of raytraced scenes and software rendering seemed so promising...

        Comment


        • #5
          And we have a winner : 1.15 fps

          Comment


          • #6
            @Michael
            You can add that some ppl with not funds for better GPU's may use LLVMpipe for learning OpenGL versions not supported by their GPU's.

            For learning (which usually involve simpler gfx) LLVMpipe is quite nice. If only it had modern OpenGL support.

            Comment


            • #7
              It would be far more useful to see benchmarks of how well WMs run against weaker hardware.

              Comment


              • #8
                Originally posted by nightmarex View Post
                I guess it's a step in the right direction. I'm actually kind of surprised the frame rates are so low. I remember when quad cores came out and the talk of raytraced scenes and software rendering seemed so promising...
                The CPU he used for the benchmark is an hyper-threaded dual-core (= 4 threads), not a quad core.

                Desktop Celeron/Pentium: dual-core
                Desktop i3: dual-core + HT
                Desktop i5: quad-core
                Desktop i7: quad-core + HT, some hexa-core + HT

                On the mobile market, all the i3, i5 and i7s are hyperthreaded, all i5s are only dual core, some i7s are dual core (those without a "Q" or a "X" in their name).

                Comment


                • #9
                  Originally posted by nightmarex View Post
                  I remember when quad cores came out and the talk of raytraced scenes and software rendering seemed so promising...
                  I remember reading that we would all have flying cars by now. That seemed really promising too

                  That said, the frame rates are pretty impressive (except for Nexuiz, of course). GPUs still have a *lot* more processing power than a typical CPU.
                  Last edited by bridgman; 14 April 2013, 10:57 AM.
                  Test signature

                  Comment


                  • #10
                    Originally posted by ChrisXY View Post
                    Can anyone predict how it will run on a parallela (the 64 core version)?

                    Introduction The E64G401 is 64-core microprocessor/coprocessor reference design based on the 4th generation of the Epiphany multicore architecture that was designed as a direct replacement of the E…

                    http://www.kickstarter.com/projects/...r-for-everyone
                    If it worked then you could imagine a second attempt at the open graphics card project, with a much higher chance of producing something useful at a reasonable cost.

                    Comment

                    Working...
                    X