Netgate Discussion Forum
    • Categories
    • Recent
    • Tags
    • Popular
    • Users
    • Search
    • Register
    • Login

    AMD Ryzen pfSense

    Scheduled Pinned Locked Moved Off-Topic & Non-Support Discussion
    4 Posts 2 Posters 2.9k Views
    Loading More Posts
    • Oldest to Newest
    • Newest to Oldest
    • Most Votes
    Reply
    • Reply as topic
    Log in to reply
    This topic has been deleted. Only users with topic management privileges can see it.
    • H
      Harvy66
      last edited by

      I use a Haswell i5 3.2ghz for my pfSense box, but I wonder how a Ryzen, possibly a single module (Ryzen 5 1500) version would perform.

      1 Reply Last reply Reply Quote 0
      • W
        W4RH34D
        last edited by

        @Harvy66:

        I use a Haswell i5 3.2ghz for my pfSense box, but I wonder how a Ryzen, possibly a single module (Ryzen 5 1500) version would perform.

        pfsense benchmark utility would be BA, as most synthetics don't accurately show what can be done.

        I find that I prefer 3.0+ Ghz for pfsense but I have no metric data to show why or what that is.

        Lately my machine has bogged down in responsiveness.  I think it's better on the mind to be a bit slower because it was so fast that I didn't even perceive the refresh or website change.

        Ryzen was supposed to have better single thread performance from articles on the web but reality shows it is application specific.

        Did you really check your cables?

        1 Reply Last reply Reply Quote 0
        • H
          Harvy66
          last edited by

          FreeBSD pre-11.x didn't have the best SMP performance, but was much better than 9.x. Even 11 leaves us wanting. This makes few-core high-frequency better. As they tune the network stack and drivers to make better use of SMP, we will see many cores becoming viable, which will be great because as you reduce the frequency you can reduce the voltage.

          I'm not sure if 3ghz+ is required for any specific reason. In my testing, I was able to get about 1.4Mpps UDP around 20% cpu usage. To give an idea of the scaling of this, 400Kpps was about 15% usage. A 3.5x increase in pps resulted in a 30% increase in cpu usage. Not bad. That was with HFSC+Codel.

          1 Reply Last reply Reply Quote 0
          • W
            W4RH34D
            last edited by

            I wasn't measuring bandwidth but user experience latency.

            Stupid fast!

            Did you really check your cables?

            1 Reply Last reply Reply Quote 0
            • First post
              Last post
            Copyright 2025 Rubicon Communications LLC (Netgate). All rights reserved.