Announcement

Collapse
No announcement yet.

8-Way AMD Gallium3D vs. Catalyst Ubuntu GPU Benchmarks

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

  • #11
    Originally posted by przemoli View Post
    Maybe AMD can provide ppa for LLVM with their work?

    It would speed up testing. (And Oibaf could add that to his drivers then)
    Does all the changes are in separate repo or in upstream LLVN svn?
    Because there are a llvm debs:
    deb http://llvm.org/apt/unstable/ llvm-toolchain main

    Comment


    • #12
      Originally posted by stqn View Post
      The names of the cards are unreadable…
      I had minimum font size set to 12 in Firefox’ preferences; turning it to none makes the text readable (while at the same time making some other text barely readable because too small.)

      Comment


      • #13
        Thank you. Interessting article.

        Comment


        • #14
          Thanks for your reviews Michael, It's great to see how the open driver keeps getting better and better. I'll become a premium subscriber this upcoming Friday in thanks for your efforts.
          The numbers look great only the 7850 looks odd in the Xonotic tests using the open driver I wonder what's different about this card? Number seems to be about half of what it should be compared to the rest of the cards in it's range.
          Those who would give up Essential Liberty to purchase a little Temporary Safety,deserve neither Liberty nor Safety.
          Ben Franklin 1755

          Comment


          • #15
            Does Oibaf drivers use LLVM in r600g?

            Just curious if the LLVM 3.4 update would make a difference there or not. Might be an interesting test for Michael to show which backend is performing better for that driver, as well.

            Comment


            • #16
              These results make it a good Saturday for me. The free driver is constantly getting better. Once it was like a better VESA driver and now it even manages to beat fglrx sometimes on 3D workload. Awesome. Just the 6450 shouldn't be that slow. Well, maybe the devs have to add some quirks here and there for some cards.
              I was also surprised that SI was already that much supported. If that also works with the APUs that use SI (Kabini?) then it would be really awesome since I could soon buy myself a nice Kabini board as HTPC. Or just a HD7750 for my main box as an upgrade. I'm sure there's also a passive version of that one around.

              Comment


              • #17
                Originally posted by Adarion View Post
                Just the 6450 shouldn't be that slow.
                It's the only card being tested that has 64-bit memory interface I believe, that is what cripples it. Honestly even the lowest of the low GPU should have at LEAST 128-bit interface for the memory.

                Comment


                • #18
                  Originally posted by sobkas View Post
                  Does all the changes are in separate repo or in upstream LLVN svn?
                  Because there are a llvm debs:
                  deb http://llvm.org/apt/unstable/ llvm-toolchain main
                  All of Tom's changes are upstream trunk svn/git LLVM.

                  Every couple of weeks Debian rolls a new 3.4 update in experimental.

                  Comment


                  • #19
                    It's best to put RadeonSI+ cards on hold until LLVM 3.4 becomes stable and we can actually use it in Oibaf and Trusty Tahr. I can't even play a single Source game without open source drivers on my HD 7950 either CTDing or locking up the system.

                    Comment


                    • #20
                      For us who do not care at all about the games being used in benchmarking here, would it be possible to add a flightsim? Flightgear and X-Plane? The day X-Plane performs on par with FLOSS drivers as with closed drivers, I'm going fully FLOSS!

                      Comment

                      Working...
                      X