Announcement

Collapse
No announcement yet.

Can Ubuntu 9.10 Outperform Mac OS X 10.6?

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

  • Can Ubuntu 9.10 Outperform Mac OS X 10.6?

    Phoronix: Can Ubuntu 9.10 Outperform Mac OS X 10.6?

    Back on Friday we published Mac OS X 10.6 benchmarks and found it to offer some terrific performance improvements, but at the same time, there were a few notable regressions. Apple engineers have been working hard at pushing technologies like Grand Central Dispatch (GCD), OpenCL, full 64-bit support, and other changes to their OS X stack to bolster its performance capabilities and reduce the overall footprint. Now that we have tested Mac OS X 10.6, we are seeing how its performance compares to that of Ubuntu Linux. Ubuntu 9.10 "Karmic Koala" will be out in October and does have some performance improvements as our earlier tests have shown, but Canonical engineers have not been exclusively focusing on performance optimizations with this release. Can the Karmic Koala outperform Snow Leopard? Yes and no.

    http://www.phoronix.com/vr.php?view=14158

  • #2
    Originally posted by phoronix View Post
    Phoronix: Can Ubuntu 9.10 Outperform Mac OS X 10.6?
    I would say that is a "Yes." Ubuntu seem to have won more of the tests than it lost or tied.

    Comment


    • #3
      Most of those tests would be better suite to the server variants of both Ubuntu and Mac OS.

      Morever, since the comparison involves proprietary operating systems, it would be nice to throw in Windows results, too.

      Comment


      • #4
        About timed compilation tests. Like Kano mentioned - dependencies counts? If yes, what's the point? :> Btw. Ubuntu can outperform OS X easily and OS X can outperform Ubuntu - it matters what tests you'll show :> It seems there are some problems with ray-tracing. I wonder what can have influence on the results. GCC maybe?
        Last edited by kraftman; 08-31-2009, 08:55 AM.

        Comment


        • #5
          Originally posted by BlackStar View Post
          Morever, since the comparison involves proprietary operating systems, it would be nice to throw in Windows results, too.
          When Windows becomes a unix, PTS will be ported within days. Until then, no. A benching app running on cygwin won't tell anybody anything.


          It would have been nice to have the two most recent OS X versions here, since we had the two most recent Ubuntu versions.. but oh well.

          Comment


          • #6
            Originally posted by ethana2 View Post
            When Windows becomes a unix, PTS will be ported within days. Until then, no. A benching app running on cygwin won't tell anybody anything.
            Actually, as I have said before, if any individual or company will send me out a few Windows (7) systems at the least, I will work on porting PTS to Windows. But until then, there will be no Windows support, I have access to no Windows installations.
            Michael Larabel
            http://www.michaellarabel.com/

            Comment


            • #7
              8.10?

              I really wish phoronix would include ubu 8.10 in benchmark. Many people have refused to upgrade because of the intel graphics performance regression with 9.04 and 9.10.

              Comment


              • #8
                Originally posted by Michael View Post
                Actually, as I have said before, if any individual or company will send me out a few Windows (7) systems at the least, I will work on porting PTS to Windows. But until then, there will be no Windows support, I have access to no Windows installations.
                If someone buys a laptop and doesn't want the preinstalled windows, is legal to give them?

                Comment


                • #9
                  Originally posted by kraftman View Post
                  About timed compilation tests. Like Kano mentioned - dependencies counts? If yes, what's the point? :> Btw. Ubuntu can outperform OS X easily and OS X can outperform Ubuntu - it matters what tests you'll show :> It seems there are some problems with ray-tracing. I wonder what can have influence on the results. GCC maybe?
                  I suspect this is might be a factor more of usage of LLVM.

                  It would also be interesting to see what the results would be for C projects when compiled with clang.
                  Last edited by deanjo; 08-31-2009, 12:22 PM.

                  Comment


                  • #10
                    Originally posted by deanjo View Post
                    I suspect this is might be a factor more of usage of LLVM.
                    Yes, this is possible.

                    It would also be interesting to see what the results would be for C projects when compiled with clang.
                    Yep, I'm also interested.

                    Comment


                    • #11
                      Originally posted by kraftman View Post
                      Yes, this is possible.



                      Yep, I'm also interested.
                      The reason I say it might be a factor is because openSSL performance in the benchmark are surprising considering that the LLVM project mentions it as being one of the key performance libraries that 10.6 compiled with LLVM.

                      http://llvm.org/Users.html

                      Mac OS X 10.6 (and later): The OpenCL GPGPU implementation is built on Clang and LLVM compiler technology. This requires parsing an extended dialect of C at runtime and JIT compiling it to run on the CPU, GPU, or both at the same time. In addition, several performance sensitive pieces of Mac OS X 10.6 were built with llvm-gcc such as OpenSSL and Hotspot. Finally, the compiler_rt library has replaced libgcc and is now a part of libsystem.dylib.
                      Something is a amiss, and determining the cause of poor performance is probably more important then the specific results. Some investigation is needed. Pehaps a run against 10.6 own openSSL library would provide some insight to this.

                      Comment


                      • #12
                        gcc vs. llvm-gcc on 10.6 results should be out this week.
                        Michael Larabel
                        http://www.michaellarabel.com/

                        Comment


                        • #13
                          Originally posted by Michael View Post
                          gcc vs. llvm-gcc on 10.6 results should be out this week.
                          Thanks Michael.

                          Comment


                          • #14
                            I may be talking out of my arse here, but didn't it used to be that a number of the packages in the alpha releases of Ubuntu used to have lots of extra debug stuff enabled, which would slow their performance down?

                            It seems odd to me to bother benchmarking pre-release software against a completely different, stable, OS?

                            Comment


                            • #15
                              Originally posted by Garp View Post
                              I may be talking out of my arse here, but didn't it used to be that a number of the packages in the alpha releases of Ubuntu used to have lots of extra debug stuff enabled, which would slow their performance down?

                              It seems odd to me to bother benchmarking pre-release software against a completely different, stable, OS?
                              That varies from distro to distro. IIRC Ubuntu doesn't have debug enabled in their alpha or beta releases.

                              Comment

                              Working...
                              X