Announcement

Collapse
No announcement yet.

GCC 4.5 vs. 4.6 On AMD's FX-4100 Bulldozer

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

  • GCC 4.5 vs. 4.6 On AMD's FX-4100 Bulldozer

    Phoronix: GCC 4.5 vs. 4.6 On AMD's FX-4100 Bulldozer

    In continuing from yesterday's AMD FX-4100 "Bulldozer" Linux benchmarks, here are more Ubuntu test results from this system comparing the stock GCC 4.5.2 and GCC 4.6.1 compilers for the new Bulldozer platform.

    http://www.phoronix.com/vr.php?view=16563

  • #2
    Is the table for the system specification really supposed to be far outside of the main page? It's been like that for months, so I assume so. I'm using Firefox 7.0.1, so it's not like it's an uncommon browser.

    Comment


    • #3
      Originally posted by Azpegath View Post
      Is the table for the system specification really supposed to be far outside of the main page? It's been like that for months, so I assume so. I'm using Firefox 7.0.1, so it's not like it's an uncommon browser.
      I've been working on an appropriate fix and had one working for some browsers but not others, still working on it.
      Michael Larabel
      http://www.michaellarabel.com/

      Comment


      • #4
        another incomplete article! why?!

        Michael, I personally don't really like that you are pumping out articles with bits and pieces of information instead of doing all benchmarks and writing proper performance analysis in a single, complete review?! I would appreciate the latter much more and I've heard similar comments from others as well.

        To be more specific, in this case without benchmarks with proper compiler tuning for Barcelona architecture (-march=bdvar1 and others), IMO this article is just incomplete and kindof sucks.

        You should really consider starting a blog section and have clear separation between these blog-like posts and proper reviews. Alternatively, if you want to still keep on pumping out partial reviews without calling them blog posts you could at least call them "pre-review" or something similar.

        Cheers,
        Sz.

        Comment


        • #5
          Summing up the most notable changes on that configuration:
          GCC 4.6 yields a 25% improvement in the GraphicsMagick Sharpen test, and a 50% improvement in the GCrypt Library Camelia-256 ECB Cipher test.
          GCC 4.5 has a 14% advantage in the N-Queens test though.

          That's basically it.
          If you don't know what that means for your usage cases, well... Welcome to the club.

          Comment


          • #6
            Originally posted by pszilard View Post
            Michael, I personally don't really like that you are pumping out articles with bits and pieces of information instead of doing all benchmarks and writing proper performance analysis in a single, complete review?! I would appreciate the latter much more and I've heard similar comments from others as well.

            To be more specific, in this case without benchmarks with proper compiler tuning for Barcelona architecture (-march=bdvar1 and others), IMO this article is just incomplete and kindof sucks.

            You should really consider starting a blog section and have clear separation between these blog-like posts and proper reviews. Alternatively, if you want to still keep on pumping out partial reviews without calling them blog posts you could at least call them "pre-review" or something similar.

            Cheers,
            Sz.
            why do you not set "pre-review" in your mind for phoronix .com ?

            on phoronix.com i mostly read only raw results most of the time thereis no deep analysis about why is something slow and why is something fast and what is the best way to go...

            Michael Larabel isn't your computer decision consultant brain prosthesis.

            if you want a brain prosthesis I'm sure Michael can help you if you pay the bill.

            bill thats a good idea why not pay him a co-worker to do proper reviews ?

            Comment


            • #7
              Originally posted by Qaridarium View Post
              why do you not set "pre-review" in your mind for phoronix .com ?

              on phoronix.com i mostly read only raw results most of the time thereis no deep analysis about why is something slow and why is something fast and what is the best way to go...

              Michael Larabel isn't your computer decision consultant brain prosthesis.

              if you want a brain prosthesis I'm sure Michael can help you if you pay the bill.

              bill thats a good idea why not pay him a co-worker to do proper reviews ?
              You've obviously just jumped at defending Michael without actually thinking through what I said. I'm not going to do you a favor by continuing the smart-ass style discussion you started.

              I was not demanding any kind of deep analysis. Please re-read what I wrote and realize that I was criticizing the way bits and pieces of information are released over several articles. This obviously introduces redundancy (introduction, hardware specs, etc. repeated in multiple articles) and fragmentation of the topic.

              I can only guess about the reasons why he doesn't hold off with an article for a day or two (or even a week) and publish at once a complete set of benchmarks.


              And btw I'm lucky enough that I have access to tons of computer hardware and expertise in most of the decisions on choosing, some even before the official release, so your idiotic comments don't even have much relevance...

              Cheers,
              Sz

              Comment


              • #8
                As has been said, pretty much useless with default compiler flags. At least your binary packages won't speed up...

                But so are most bulldozer tests I have seen... I think the most useful article would compare 3.0 and 3.1-rc10 with and without the patch and gcc 4.5, gcc 4.6, open64 and path64/ekopath nightly first with default settings and then the difference with -march=native and -Ofast or the equivalent of the compiler.

                It's strange anyway how phoronix has hyped the ekopath compiler a bit with the dirndl stories and then never again picked th topic up...

                Comment


                • #9
                  Originally posted by pszilard View Post
                  You've obviously just jumped at defending Michael without actually thinking through what I said. I'm not going to do you a favor by continuing the smart-ass style discussion you started.

                  I was not demanding any kind of deep analysis. Please re-read what I wrote and realize that I was criticizing the way bits and pieces of information are released over several articles. This obviously introduces redundancy (introduction, hardware specs, etc. repeated in multiple articles) and fragmentation of the topic.

                  I can only guess about the reasons why he doesn't hold off with an article for a day or two (or even a week) and publish at once a complete set of benchmarks.


                  And btw I'm lucky enough that I have access to tons of computer hardware and expertise in most of the decisions on choosing, some even before the official release, so your idiotic comments don't even have much relevance...

                  Cheers,
                  Sz
                  He has to post as many articles as he can, regardless of if it makes sense or even matters, it's how a content mill laden with advertisement works.

                  The RSS feed is useless because it only includes a sentence or two, to get you to the site full of advertisements.

                  About 1 in 10 articles is something useful.

                  So to recap, if he slowed down and made better articles, he'd put out less pages full of ads and his pay would decrease.

                  Other content mills, such as ZDNet and PC World and others do the same thing, only worse...One author writes an "Ubuntu sucks" article, another writes an "ZOMG, Ubuntu is awesome article", that way they get maximum clicks from people who already decided one way or another how they felt about Ubuntu.

                  Phoronix posts ceaseless benchmarks on things that don't really change enough to be noticeable over the span of weeks or a few months (like video drivers). In some cases, including this one, the benchmarks make no sense because they're not being compared to anything else. In other cases, the benchmarks are simply botched (In one file system benchmark, Phoronix was corrected by Eric Sandeen of Red Hat, who actually wrote parts of XFS and Ext4).

                  In yet another one, that refuses to die, he screams every day about a few minutes less battery life because of a patch made so that systems wouldn't crash.

                  And every time he benchmarks VDrift, he claims that Mesa is faster because it doesn't handle shaders correctly, and is then corrected by Marek Olsak who obviously knows what Mesa does and doesn't do who tells him that it is Vdrift that is to blame for shaders that shouldn't even compile on standards-compliant OpenGL drivers in the first place.

                  https://en.wikipedia.org/wiki/Content_farm

                  Comment


                  • #10
                    Originally posted by DaemonFC View Post
                    He has to post as many articles as he can, regardless of if it makes sense or even matters, it's how a content mill laden with advertisement works.

                    The RSS feed is useless because it only includes a sentence or two, to get you to the site full of advertisements.

                    About 1 in 10 articles is something useful.

                    So to recap, if he slowed down and made better articles, he'd put out less pages full of ads and his pay would decrease.

                    Other content mills, such as ZDNet and PC World and others do the same thing, only worse...One author writes an "Ubuntu sucks" article, another writes an "ZOMG, Ubuntu is awesome article", that way they get maximum clicks from people who already decided one way or another how they felt about Ubuntu.

                    Phoronix posts ceaseless benchmarks on things that don't really change enough to be noticeable over the span of weeks or a few months (like video drivers). In some cases, including this one, the benchmarks make no sense because they're not being compared to anything else. In other cases, the benchmarks are simply botched (In one file system benchmark, Phoronix was corrected by Eric Sandeen of Red Hat, who actually wrote parts of XFS and Ext4).

                    In yet another one, that refuses to die, he screams every day about a few minutes less battery life because of a patch made so that systems wouldn't crash.

                    And every time he benchmarks VDrift, he claims that Mesa is faster because it doesn't handle shaders correctly, and is then corrected by Marek Olsak who obviously knows what Mesa does and doesn't do who tells him that it is Vdrift that is to blame for shaders that shouldn't even compile on standards-compliant OpenGL drivers in the first place.

                    https://en.wikipedia.org/wiki/Content_farm
                    A Content farm do outperform traditional journalism.

                    Or in Wikipedia words: "that proponents represent as "true market demand", a feature that traditional journalism lacks."

                    and the Quality is "just good enough"

                    and Michael do real research and a pure content farm do not make research.

                    and wikipedia and Google also do not count research as a content farm.

                    Comment


                    • #11
                      Originally posted by DaemonFC View Post
                      And every time he benchmarks VDrift, he claims that Mesa is faster because it doesn't handle shaders correctly, and is then corrected by Marek Olsak who obviously knows what Mesa does and doesn't do who tells him that it is Vdrift that is to blame for shaders that shouldn't even compile on standards-compliant OpenGL drivers in the first place.
                      He should be running the latest VDrift (2011-09) instead of the one from 2010-06, perhaps that is a lot better. I've taken the old GLSLValidator from 3DLabs and updated it to compile on a "modern" Linux distribution with wxWidgets 2.8. I was planning on running their shaders through that since it only supports up to GLSL1.2. I tested some of their shaders and some of those fragment shaders failed.
                      The code/program can be found at https://github.com/AzP/GLSL-Validate/

                      Comment


                      • #12
                        Originally posted by Azpegath View Post
                        He should be running the latest VDrift (2011-09) instead of the one from 2010-06, perhaps that is a lot better. I've taken the old GLSLValidator from 3DLabs and updated it to compile on a "modern" Linux distribution with wxWidgets 2.8. I was planning on running their shaders through that since it only supports up to GLSL1.2. I tested some of their shaders and some of those fragment shaders failed.
                        The code/program can be found at https://github.com/AzP/GLSL-Validate/
                        I only had a few minutes before to look into VDrift 2011 but the source package now for it is only 1MB (it should be several hundred MB), and haven't had the time to look into see what changed or how the VDrift build system was altered.
                        Michael Larabel
                        http://www.michaellarabel.com/

                        Comment


                        • #13
                          Originally posted by Michael View Post
                          I only had a few minutes before to look into VDrift 2011 but the source package now for it is only 1MB (it should be several hundred MB), and haven't had the time to look into see what changed or how the VDrift build system was altered.
                          You can have a look at the archlinux PKGBUILDS from the repositories. They have version 2011.09.01. Should be quite easy to read and guess what it does. (the source data gets extracted to $srcdir and the files in $pkgdir will end up in the package).
                          game binary: http://projects.archlinux.org/svntog...86_64/PKGBUILD
                          game data: http://projects.archlinux.org/svntog...y-any/PKGBUILD
                          Last edited by ChrisXY; 10-21-2011, 06:12 AM.

                          Comment


                          • #14
                            Originally posted by Michael View Post
                            I only had a few minutes before to look into VDrift 2011 but the source package now for it is only 1MB (it should be several hundred MB), and haven't had the time to look into see what changed or how the VDrift build system was altered.
                            Yes they split the packages into src and data, and haven't released a data package for it. Sadly they just seem to want to distribute data via svn since their previous version. I've been trying to get them to release a data tar file matching the source release so we can get an updated package in Gentoo, but the devs doesn't reply on IRC. Their channel is just quiet. I've only tried for 2 whole days, but you'd think that somebody would reply.

                            There's a bug on the Gentoo bugzilla about it describing the issue furter (https://bugs.gentoo.org/show_bug.cgi?id=351409#c7)

                            Comment


                            • #15
                              Originally posted by Azpegath View Post
                              Yes they split the packages into src and data, and haven't released a data package for it. Sadly they just seem to want to distribute data via svn since their previous version. I've been trying to get them to release a data tar file matching the source release so we can get an updated package in Gentoo, but the devs doesn't reply on IRC. Their channel is just quiet. I've only tried for 2 whole days, but you'd think that somebody would reply.

                              There's a bug on the Gentoo bugzilla about it describing the issue furter (https://bugs.gentoo.org/show_bug.cgi?id=351409#c7)
                              I'll check with the main developer when I have the time. Last I knew I even had SVN commit access to VDrift from when I was doing the benchmark mode for it, so hopefully he will listen.
                              Michael Larabel
                              http://www.michaellarabel.com/

                              Comment

                              Working...
                              X