Announcement

Collapse
No announcement yet.

What Linux Benchmarks Would You Like To See Next?

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

  • #21
    For me, two benchmarks are very interesting:

    1) Windows vs Linux performance of GTX980 or GTX970 using the commercial steam games (that are available native for Windows and Linux).

    2) SLI vs Single card performance in Linux of Maxwell microarchitecture.

    Comment


    • #22
      Originally posted by xpander View Post
      i think it has

      timedemo1_p -benchmark -seconds=60 -novsync

      will test it if i have time
      Let me know if it works for you.
      Michael Larabel
      https://www.michaellarabel.com/

      Comment


      • #23
        Scientific / Computational Chemistry software?

        I'd like to see benchmarks with CP2K, QuantumEspresso or Abinit open source computational chemistry packages..
        A bit of a niche interest I know, but I run a lot of DFT calculations so it's what matters to me.

        More generally, some measure of power efficiency would be useful. Idle and performace / watt with different load types.

        Anyway, keep up the good work!

        Comment


        • #24
          ]Apps Benchmarking?

          1) How about LibreOffice or OpenOffice, Firefox or Chrome, or other popular productivity apps/suites?

          It would be a challenge to build a comprehensive test suite, but they might already exist in some form with the developers, or perhaps an opportunity to create the framework to build one. From an openbenchmarking perspective, another opportunity to build or validate integration points into other testing tools (if you need to integrate their test suite), or find ways to leverage the community to grow a large test suite (could a libreoffice user easily submit a test case?) [perhaps this is all possible now, I mostly skim Phoronix to have a feel for what's going on, so I might be naive about openbenchmarking]

          I was curious if this would yield useful information - when changing the hardware, software, or configurations, is there a noticeable impact on the apps people use day to day? Does using X vs Wayland vs Mir, or upgrading to newer versions of those, have a noticeable benefit or impact to day to day apps? Do different versions of these end-user applications noticeably impact resource usage, battery life, or performance (has libreoffice been getting faster/more efficient, or slower/more bloated/draining my battery?).

          App tests also should be inherently more portable, able to run on any platform (How fast does LibreOffice perform on Linux, BSD, OSX, Windows)

          2) I like the idea of the battery life tests as well. Not just from an OS/driver perspective, but also with say the top 10-20 apps running. (ie, does some app process keep waking the system?)

          Comment


          • #25
            Originally posted by Michael View Post
            Unfortunately it still looks like they don't have a benchmark mode in Spring, at least per that thread.
            Spring has a replay function allowing to watch any of the 60+ thousands of games stored at http://replays.springrts.com/browse/
            Fly the camera over one of those replays -> benchmark?

            Abma one of the more active Spring developers boiled it down to:
            to get a benchmark imo three things are needed:

            1. check if its possible to run spring on different platforms / hardware with the same config (screen-resolution, graphic-decals, ...) which affects performance
            2. make some lua-ai / gadget which is deterministic (does every run the same). maybe easiest would be a something like a give all / camera fly over the scene.
            3. create a widget that quits the game / dumps stats

            it shouldn't be soooo hard to do that, because the validation-test already does something similar.
            So how do we get this working.. :-)
            (Or who has the time.. yes I know a bit about Spring and can write php-cli code.. but surely someone else is willing to do this :-) right?)

            Comment


            • #26
              I'd love to see some SLI benchmarks (I know it's hard since that requires having two identical cards). The last SLI benchmarks are from 2007 afaik.

              Comment


              • #27
                Some ideas:
                • Timed AOSP build: AOSP has "notorious" build requirements, stretches new hardware, would be interesting to compare 4/8+ core CPUs.
                • Youtube video playback - basically the same as the mplayer test but with Firefox/Chrome playing the video. Compare ff vs chrome and flv vs h264 vs webm players/streams.
                • Timed Bittorrent download (for example) - Run rtorrent server on a LAN PC(s), generate a large file from a random seed and the .torrent file, then time a client downloading the file. It should to be trivial to add different servers or clients, some article ideas "fastest bittorrent client", compare throughput of different ethernet/wifi chipsets, or proprietary vs open drivers (where possible eg. Broadcom chips).
                • Some article/benchmarks comparing performance per $ might be interesting, ie something like https://www.cpubenchmark.net/cpu_value_available.html but including non-x86 CPU/GPU. In many benchmarks it is obvious that Intel will "win" but in the real world cost is a factor.
                • Power consumption on Android/Win/OS X/Linux, on same hardware and across hardware. Is Linux competitive?
                • Wayland vs Mir for whatever works in both (might not be ready yet, but I imagine Chromium, Firefox, Libreoffice, VLC/Mplayer/XBMC, some kind of QT/GTK benchmarks are all candidates?)
                • Gaming benchmarks in a VM; would be interesting to see how the various GPU virtualization projects are doing (Virgil3D, Intel GVT, VirtualBox, VMware etc.)

                Comment


                • #28
                  Originally posted by Michael View Post
                  Let me know if it works for you.
                  doesnt seem to work indeed

                  seems the timedemo was working in borderlands 1..but this isnt the case with 2 and tps anymore..

                  sad

                  Comment


                  • #29
                    Originally posted by hlechner View Post
                    For me, two benchmarks are very interesting:

                    1) Windows vs Linux performance of GTX980 or GTX970 using the commercial steam games (that are available native for Windows and Linux).

                    2) SLI vs Single card performance in Linux of Maxwell microarchitecture.
                    +1 for both !

                    Thx Michael

                    Comment


                    • #30
                      Originally posted by phoronix View Post
                      Phoronix: What Linux Benchmarks Would You Like To See Next?
                      SQL database performance on disk (fsync) limited systems.

                      For instance the disk with the database should ideally be located on a slows SSD, USB or SD Card.

                      Databases would be ideally mysql, postgresql, sqlite.

                      Why? Many e-mail readers, web browsers and other desktop goodies use sqlite (sometimes mysql), but become almost unusably slow when running from a slow SSD or SD Card (my good old Eeepc 900 f.i, or a USB based ubuntu). Reason is the way fsync's affect the performance of the machine (stall the CPU).

                      I would be good to be able to track the progress of file systems handling this corner case.

                      Comment

                      Working...
                      X