Announcement

Collapse
No announcement yet.

Linux Benchmarks Of NVMe SSD Performance With Varying I/O Polling Queues

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

  • Linux Benchmarks Of NVMe SSD Performance With Varying I/O Polling Queues

    Phoronix: Linux Benchmarks Of NVMe SSD Performance With Varying I/O Polling Queues

    A Phoronix reader recently pointed out a little known kernel tunable option to adjust the number of I/O polling queues for NVMe solid-state drives that can potentially help improve the performance/latency. Here are some benchmarks from the NVMe poll_queues option...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    I'm sure the average home user or heavy number-crunching server would be better off with 0. Something like a NAS or database server would be better with a higher number.

    Comment


    • #3
      I'm trying to run the Tests on my machine is it normal that there is a sleep 120 between the test runs?

      Comment


      • #4
        Originally posted by dispat0r View Post
        I'm trying to run the Tests on my machine is it normal that there is a sleep 120 between the test runs?
        Hmmm? There shouldn't be any long sleep cycle between tests.
        Michael Larabel
        https://www.michaellarabel.com/

        Comment


        • #5
          Originally posted by Michael View Post

          Hmmm? There shouldn't be any long sleep cycle between tests.
          if [[ "$MAX_STARTUP" != "0" ]]; then
          bash -c "sleep $MAX_STARTUP && \
          echo Timeout: killing command;\
          cat current-pid | xargs -I pid kill -9 -pid ;\
          touch Stop-iterations" &
          KILLPROC=$!
          disown
          fi

          This part here sleeps 120 seconds maybe some test doesn't work like it supposed to be on my system?

          Comment


          • #6
            Originally posted by dispat0r View Post

            if [[ "$MAX_STARTUP" != "0" ]]; then
            bash -c "sleep $MAX_STARTUP && \
            echo Timeout: killing command;\
            cat current-pid | xargs -I pid kill -9 -pid ;\
            touch Stop-iterations" &
            KILLPROC=$!
            disown
            fi

            This part here sleeps 120 seconds maybe some test doesn't work like it supposed to be on my system?
            What code are you looking at? That code doesn't exist in PTS itself at least.
            Michael Larabel
            https://www.michaellarabel.com/

            Comment


            • #7
              installed-tests/pts/startup-time-1.4.1/S-master/comm_startup_lat/comm_startup_lat.sh

              Comment


              • #8
                Originally posted by dispat0r View Post
                installed-tests/pts/startup-time-1.4.1/S-master/comm_startup_lat/comm_startup_lat.sh
                Ah okay that would be an issue then with upstream S program maintained by the Linaro/BFS crew.
                Michael Larabel
                https://www.michaellarabel.com/

                Comment

                Working...
                X