Announcement

Collapse
No announcement yet.

Hands On With The Most Open-Source, High-Performance System For 2018

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

  • Hands On With The Most Open-Source, High-Performance System For 2018

    Phoronix: Hands On With The Most Open-Source, High-Performance System For 2018

    While there are several vendors working on open-source hardware systems with goals of fully open designs and open-source software down to the firmware, there is only one vendor that has achieved that mission while delivering server/workstation class performance as we approach the end of 2018... Raptor Computing Systems' Talos II. We finally have this dual POWER9 system in our labs for some interesting benchmarks ahead...

    http://www.phoronix.com/scan.php?pag...itial-Hands-On

  • #2
    Would like to know about compilation performance and idle power consumption.

    Comment


    • #3
      My wishlist using debian buster as main system

      a) run with virt-manager a linux distro, i suggest xfce or lxde as DE because they are lightw, and try to watch a youtube video (with x86 it work with no problem)
      b) test if firejail, qbittorent, amule, vlc (or another video player) and libre office works properly
      c) firefox esr 52 should start, i'm curious to know how it work, if there is lags or if is usable, firefox quantum actually do not work from repo, need to be built from source if you have time to do it (this could help) https://www.talospace.com/2018/09/mo...irefox-62.html



      with blackbird i think people like me could be interested on desktop programs, we already saw, thanks to phoronix, the "power" numbers, probably now is time to more desktop user experience info, because there is no info about it on internet

      Comment


      • #4
        That system is a beautiful beast.

        Comment


        • #5
          The usual mix of benchmakrs for the use cases these machines are build for are fine, I guess. But I liked also your past media and content creation benchmarks (btw. is the brand new AOM encoder benchmark friendly? the usual x264 and x265 are popular, too) even though it is forseeable that POWER and ARM might lack behind due to missing optimizations. But it is a valid use case and a fair point to be made there. It might be an incentive for these companies to invest more into these projects.

          I don't know if it is feasible to run the most recent or an older SPEC suite on these machines or in a comparable manner, but as their workloads are considered to be standard in the industry, this would be nice to see in a fair fight across ISA boundaries (maximum compiler optimization vs. equal set of optimization level on all targets) as well as performance/power metrics. But I don't know if they charge an arm and a leg for a license.

          Comment


          • #6
            Man my back started to ache just looking at that machine. This is no Mac Mini🤣🤣🤣😂😒😜

            in any event Micheal where do we stand on Alt-Vec in these systems? Maybe a brief review with testing against Intel and maybe even a few GPUs. I ask because frankly I haven’t heard much at since the PowerPC days. I would have to think that some improvements have been made. The imagination runs wild thinking about 44 cores of vector processors running.

            Comment


            • #7
              Originally posted by Blahblah View Post
              That system is a beautiful beast.
              Haha, no fancy RGB, no fancy passive cooling even not a custom pcb color. Instead, the specs speak for themselves. I agree, there is beauty in purism.

              Comment


              • #8
                Originally posted by MaxToTheMax View Post
                Would like to know about compilation performance and idle power consumption.
                Here's a few I ran just as some sanity checks this morning - https://openbenchmarking.org/result/...SK-POWER9COM35
                Michael Larabel
                http://www.michaellarabel.com/

                Comment


                • #9
                  Originally posted by Michael View Post

                  Here's a few I ran just as some sanity checks this morning - https://openbenchmarking.org/result/...SK-POWER9COM35
                  Just as a quick note, kernel compile time is not normally an apples to apples comparison across architectures, as the kernel configuration varies significantly between architectures and this alters the number of files compiled, etc. A better comparison is one of the compiler compilation benchmarks IMO.

                  Also, this machine has maximum spectre protections enabled -- far more stringent than modern "mitigated" x86 boxes. Protection level is a tunable on POWER9; all systems ship with it cranked up all the way for maximum security.

                  Comment


                  • #10
                    Originally posted by madscientist159 View Post

                    Just as a quick note, kernel compile time is not normally an apples to apples comparison across architectures, as the kernel configuration varies significantly between architectures and this alters the number of files compiled, etc. A better comparison is one of the compiler compilation benchmarks IMO.

                    Also, this machine has maximum spectre protections enabled -- far more stringent than modern "mitigated" x86 boxes. Protection level is a tunable on POWER9; all systems ship with it cranked up all the way for maximum security.
                    Yep will be running POWER tests at various Spectre levles.

                    And yes the defconfig does differ for Linux. As stated, this was basically sanity check after I got Ubuntu freshly installed and just wanted to fire off something to verify no immediate stability or thermal problems.
                    Michael Larabel
                    http://www.michaellarabel.com/

                    Comment

                    Working...
                    X