Announcement

Collapse
No announcement yet.

The "Dirndl" On AMD Opterons Are Impressive

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

  • #31
    Originally posted by Welsh Dwarf View Post
    If it is EKOPath, I'm going to regret not using Gentoo :P

    Anyhow, the PKGBUILD for mesa will get a tweak, that's for sure ^^

    David
    I hope it happens for every single PKGBUILD.

    Do you think Arch will switch to EKOPath any time soon?

    Comment


    • #32
      Originally posted by Viper_Scull View Post
      I hope it happens for every single PKGBUILD.

      Do you think Arch will switch to EKOPath any time soon?
      They didn't switch to systemd, so I doubt if they'll switch to EKOPath soon.

      Comment


      • #33
        Originally posted by clavko View Post
        They already released some of their code under BSD licence,

        http://www.prweb.com/releases/2011/5/prweb8464380.htm

        It would be nice if this code dump would be BSD too.
        This was for freebsd. This time it's different, so I hope they'll choose a GPL.

        Comment


        • #34
          Verry impressive results. But those results do remind me of the "compiler deathmatch" use the search function to find it. In that deathmatch gcc could be made A LOT faster than the stock settings. Archlinux four example it's using roughly the stock compiler settings. If this new compiler is just by default having all those optimizing things turned on than the current phoronix comparisons aren't even fair..

          Just my 5 cents..

          Comment


          • #35
            Originally posted by markg85 View Post
            Verry impressive results. But those results do remind me of the "compiler deathmatch" use the search function to find it. In that deathmatch gcc could be made A LOT faster than the stock settings. Archlinux four example it's using roughly the stock compiler settings. If this new compiler is just by default having all those optimizing things turned on than the current phoronix comparisons aren't even fair..

            Just my 5 cents..
            Gentoo guys will mostly tell you to stick to -march-native -O2 -pipe, anything more aggressive leads to breakage somewhere.

            Aggressive optimisations should be used per-package, when you know that they won't break a particular package.

            Comment


            • #36
              Any compiler guys around to tell whether a 2-3x increase like this over latest gcc is considered possible/doable? I was under the impression it's already very good, with icc only gaining 10-30% and visual studio less.

              Comment


              • #37
                Originally posted by pingufunkybeat View Post
                Gentoo guys will mostly tell you to stick to -march-native -O2 -pipe, anything more aggressive leads to breakage somewhere.

                Aggressive optimisations should be used per-package, when you know that they won't break a particular package.
                In some cases it also leads to slower performance.

                Comment


                • #38
                  Originally posted by pingufunkybeat View Post
                  Gentoo guys will mostly tell you to stick to -march-native -O2 -pipe, anything more aggressive leads to breakage somewhere.

                  Aggressive optimisations should be used per-package, when you know that they won't break a particular package.
                  That's -march=native

                  I've been playing with -O3 and -Ofast using GCC 4.6

                  -Ofast has problems with SQlite and wine won't start programs with -O3

                  I think I may switch back to my tried and tested -Os though

                  Comment


                  • #39
                    Originally posted by curaga View Post
                    Any compiler guys around to tell whether a 2-3x increase like this over latest gcc is considered possible/doable? I was under the impression it's already very good, with icc only gaining 10-30% and visual studio less.
                    I personally find it very unlikely, which is why I'm leaning towards this being about a cpu+gpu compiler. I haven't seen icc reach anything near 2x against gcc/llvm in my (admittedly few) benchmark tests so I doubt ekopath would be able to generate so much better code as to result in 2x, 2.3x better performance (although it sure would be awesome!). As for Visual Studio, last time I benchmarked it against GCC, GCC generated faster code for Mame atleast (which was the only test I did), but that was VS 2008 though.

                    Comment


                    • #40
                      Originally posted by FireBurn View Post
                      That's -march=native

                      I've been playing with -O3 and -Ofast using GCC 4.6

                      -Ofast has problems with SQlite and wine won't start programs with -O3

                      I think I may switch back to my tried and tested -Os though
                      Weird that sqlite has problems with -Ofast (assuming that it works with -O3) since -Ofast only turns on -ffast-math and I can't see why sqlite would depend on high-precision for it's floating point math.

                      As for -Os, it prefers code size over code speed, so unless you are starved for ram I would suggest using -O2 where -O3 causes problems rather than -Os.

                      Comment


                      • #41
                        Originally posted by deanjo View Post
                        In some cases it also leads to slower performance.
                        True, I noticed this when doing some Mame benchmarks here on the Phoronix forums, someone told me that -O2 often beat -O3, and while it didn't happen often the time it did was the one where it made the largest difference.

                        Obviously this is due to how complex it is to make correct heuristics regarding those more advanced optimizations. This can be rectified by using PGO (profile-guided optimization) where the compiler gathers all necessary runtime data during an initial run which is then used to correctly determine when and when not to use certain optimizations in order to maximize performance.

                        When using PGO I've never encountered a situation where -O3 hasn't beaten -O2, the downside of course is that compilation becomes more complex since you need to perform a information gathering execution of the code in between two compiles.

                        Comment


                        • #42
                          Originally posted by Aleve Sicofante View Post
                          Why don't you feel the need to tell your readers what's this all about? Why do you take it for granted everyone knows what you're talking about?

                          I have no idea what Dirndl is or what it is all about. If you have explained it before, put a link to that earlier article. If you haven't, put a link to an external source that explains it.

                          This is web journalism 101 and you miss it all the time.
                          This.
                          I have no idea what this article is even about.

                          Comment


                          • #43
                            Originally posted by pvtcupcakes View Post
                            This.
                            I have no idea what this article is even about.
                            It's explained in the article.

                            Comment


                            • #44
                              For who is wondering about the compiler deathmatch.. here it is: http://phoronix.com/forums/showthrea...339#post175339

                              From some months ago but the compiler world doesn't move that fast so the results are probably still valid.

                              Note the interesting thing in C-Ray. They seem to stay the same with every compiler in that deathmatch! Now if you look at that "Drindl" from this thread you see that C-Ray performs WAY better so, that "Drindl" thing must be doing something magic to perform so much better.

                              Interesting speculation in here
                              Last edited by markg85; 06-10-2011, 03:35 PM.

                              Comment


                              • #45
                                Well, I think that since michael is comparing dirndl to ubuntu, it is something that goes inside the kernel or the os in general. Maybe I'm wrong, but it's the only hint I can guess from the article.

                                Comment

                                Working...
                                X