Announcement

Collapse
No announcement yet.

Phoronix Test Suite 0.2

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

  • Phoronix Test Suite 0.2

    Phoronix: Phoronix Test Suite 0.2

    Less than a week ago we pushed out the first public release of the Phoronix Test Suite. This GPLv3 Linux benchmarking software had received a great deal of interest, but v0.1 did have a few bugs and a number of uncompleted features. Yesterday though we pushed out version 0.2 of the Phoronix Test Suite with a number of improvements, new benchmarking profiles, and other changes. In addition to noting some of the improvements made in this release, there are also other Phoronix Test Suite highlights to share from this past week.

    http://www.phoronix.com/vr.php?view=12165

  • #2
    I am watching the progress of the PTS since its launch, and it made some great improvements already in only about a week. However, I haven't tried it for myself yet, due to several reasons that don't have something to do with the PTS. I am very looking forward to comparing sidux 2008-01 (coming this week, hopefully) vs. Mandriva 2008.1 (coming this week) vs. Kubuntu 8.04 (coming in 2-3 weeks) vs. KANOTIX (even if uses Ubuntu's kernel, but the kernel isn't all a distribution offers).

    Nothing else to say than "keep it up!"

    Comment


    • #3
      As long as you test a hd install you basically bench your hd performance. Therefore using a live cd (with 2gb ram or more) and

      sudo swapoff -a

      You get the max performance possible. For other live cds exept the pts special one you usually would need about 3gb ram because this one has nexuiz preinstalled - was needed because the standard binary does not run with etch. I hope the url will be updated soon to:

      http://debian.tu-bs.de/project/kanot...tix-ng-pts.iso
      http://debian.tu-bs.de/project/kanot...ng-pts.iso.md5

      Warning: this iso image has german language by default! You can switch the keyboard with ease to us and use the console, so basically it should not really matter if kde is in german or not, if you dont like german kde then best try to remember where to change the language in kcontrol, it is always at the same position.

      Something about the nexuiz benchmark:

      this is for all owners of a midrange or better card just a cpu benchmark, as you can not get to the gpu limits with normal.cfg, you would need high or even ultra settings to use it as gfx benchmark. So don't think you card is as fast as another one - the cpu is just too slow to push the gfx card to the limit.
      Last edited by Kano; 04-07-2008, 11:23 AM.

      Comment


      • #4
        I'm wondering where I can submit a bug report for the test suite:

        When I tried to install the benchmark super-pi I got this error:

        Code:
        metzgerh@stu:~/phoronix-test-suite$ ./phoronix-test-suite install super-pi
        PHP Warning:  mkdir(): No such file or directory in /home/2008/metzgerh/phoronix-test-suite/pts-core/functions/pts-functions_config.php on line 8
        PHP Warning:  file_put_contents(/home/metzgerh/.phoronix-test-suite/user-config.xml): failed to open stream: No such file or directory in /home/2008/metzgerh/phoronix-test-suite/pts-core/functions/pts-functions_config.php on line 27
        PHP Warning:  mkdir(): No such file or directory in /home/2008/metzgerh/phoronix-test-suite/pts-core/functions/pts-functions.php on line 219
        PHP Warning:  mkdir(): No such file or directory in /home/2008/metzgerh/phoronix-test-suite/pts-core/functions/pts-functions.php on line 221
        PHP Warning:  file_put_contents(/home/metzgerh/pts-benchmark-env/.processes/phoronix-test-suite.p): failed to open stream: No such file or directory in /home/2008/metzgerh/phoronix-test-suite/pts-core/functions/pts-functions.php on line223
        PHP Warning:  mkdir(): No such file or directory in /home/2008/metzgerh/phoronix-test-suite/pts-core/functions/pts-functions-install.php on line 59
        PHP Warning:  mkdir(): No such file or directory in /home/2008/metzgerh/phoronix-test-suite/pts-core/functions/pts-functions-install.php on line 63
        PHP Warning:  mkdir(): No such file or directory in /home/2008/metzgerh/phoronix-test-suite/pts-core/functions/pts-functions-install.php on line 67
        
        =================================
        Installing Benchmark: super-pi
        =================================
        Can't open display
        Can't open display
        install.sh: line 3: cd: /home/metzgerh/pts-benchmark-env/super-pi: No such file or directory
        pi
        super_pi
        Readme.txt
        
        PHP Warning:  file_put_contents(/home/metzgerh/pts-benchmark-env/super-pi/pts-install): failed to open stream: No such file or directory in /home/2008/metzgerh/phoronix-test-suite/pts-core/functions/pts-functions-install.php on line 75
        My home directory is /home/2008/metzgerh, but sometimes it assumes that it is /home/metzgerh. Looking though the code I came across this:

        Code:
        function pts_find_home($path)
        {
                if(strpos($path, '~') !== FALSE)
                {
                        $whoami = trim(shell_exec("whoami"));
        
                        if($whoami == "root")
                                $home_path = "/root";
                        else
                                $home_path = "/home/$whoami";
        
                        $path = str_replace('~', $home_path, $path);
                }
                return $path;
        }
        As you can see here, that does not calculate it correctly.
        As you already shell out, I changed the function to be:
        Code:
        function pts_find_home($path)
        {
                if(strpos($path, '~') !== FALSE)
                {
                        $homedir = trim(shell_exec("getent passwd `whoami` | cut -d':' -f6"));
                        $path = str_replace('~', $homedir, $path);
                }
                return $path;
        }
        This appears to work on the few computers I tested it on, can anyone verify that it works for them as well?

        Harry

        Comment


        • #5
          I've just tried it and it (as all other software does) incorrectly detects my overclocked cpu speed, because of speedstep.

          Wasn't this supposed to be fixed?

          Comment


          • #6
            Originally posted by [Knuckles] View Post
            I've just tried it and it (as all other software does) incorrectly detects my overclocked cpu speed, because of speedstep.

            Wasn't this supposed to be fixed?
            I think overclockers should add to the bug reports. Propably developers with corporate linux in mind don't feel this is an important issue to fix. However it is very annoying for a lot of home users as SpeedStep doesn't work at all under linux with overclocking.

            Comment


            • #7
              This incorrect report is basically an issue for V too, when you check it. It happens when you lower the multi of your cpu. Then the reported value is always: FSB*original multi - it seems that the BIOS restriction can not read out correctly. /proc/cpuinfo just reports this.

              Comment


              • #8
                I've already tried reporting this ( http://bugzilla.kernel.org/show_bug.cgi?id=10144 ), and submitted a possible solution, maybe other people could see if my solution was correct, and confirm the report, so we could get some attention on this problem.

                Btw puntarenas, speedstep *does* work, this is just a visual issue (i.e, only the speed displayed is wrong, all is working correctly otherwise), but a very confusing one, especially if you don't understand what might be going on "huh? my bios said 3ghz and now it says 2ghz!?".

                Comment


                • #9
                  The reported lower value is no real problem for me, for me the problem is a HIGHER value than normal. You can easyly disable speedstep for max. performance.

                  Comment


                  • #10
                    Seems there are different peculiarities around the speedstep issue. I also have no problem with speedstep reporting too low frequencies (in my case CPU Frequency Scaling Monitor applet), the issue is that frequency scaling support is completely gone as soon as I overclock the FSB.

                    So when I raise the FSB on my Gigabyte 965P-DS4 from 266Mhz (default) to 333Mhz, my E6600 always runs at 333x9=3000Mhz under Linux. The same setup gives me 333x9=3000Mhz under load and 333x6=2000Mhz in idle under Windows XP and that's the way it should be, not to forget the voltage reduction in idle.

                    Comment


                    • #11
                      Originally posted by puntarenas View Post
                      Seems there are different peculiarities around the speedstep issue. I also have no problem with speedstep reporting too low frequencies (in my case CPU Frequency Scaling Monitor applet), the issue is that frequency scaling support is completely gone as soon as I overclock the FSB.

                      So when I raise the FSB on my Gigabyte 965P-DS4 from 266Mhz (default) to 333Mhz, my E6600 always runs at 333x9=3000Mhz under Linux. The same setup gives me 333x9=3000Mhz under load and 333x6=2000Mhz in idle under Windows XP and that's the way it should be, not to forget the voltage reduction in idle.
                      Oh, sorry, I thought you had the same one as me, that seems to be common enough to deserve a faq entry: http://www.codemonkey.org.uk/project...req/common.php ("I overclocked my CPU, and now cpufreq gets the frequency wrong.").

                      Have you tried playing with the cpufreq apps and kernel modules?

                      And if you haven't, do open a bug report, we need the kernel people to see that there are people out there with problems like these.

                      Comment


                      • #12
                        Have you tried playing with the cpufreq apps and kernel modules?

                        And if you haven't, do open a bug report, we need the kernel people to see that there are people out there with problems like these.
                        There are lots of bugreports and I already added to one, but it seems to be considered a minor problem: Click!

                        Comment

                        Working...
                        X