Announcement

Collapse
No announcement yet.

Google Posts KVM-CPUFreq Driver To Dramatically Boost VM Performance, Power Efficiency

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

  • Google Posts KVM-CPUFreq Driver To Dramatically Boost VM Performance, Power Efficiency

    Phoronix: Google Posts KVM-CPUFreq Driver To Dramatically Boost VM Performance, Power Efficiency

    Google engineers on Thursday posted initial "request for comments" patches on their KVM-CPUFreq driver that is part of their effort to improve the dynamic voltage and frequency scaling behavior and task placement within KVM-based virtual machines. This effort is leading to big improvements in raw performance and performance-per-Watt for tasks running within Linux VMs...

    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
    Whenever I read "Google works on driver x", I end up thinking, "I wonder what PRISM-style backdoor they've added to it, and how long it will take to find it and to find out how much damage it did?"

    It would almost be better if Snowden had never red-pilled the world and I could live in ignorant bliss. Like Cypher in the original Matrix movie - it would be nice to be able to just plug back into the Matrix and enjoy juicy steak dinners.

    Comment


    • #3
      Originally posted by andyprough View Post
      ...
      You must be fun at parties

      Comment


      • #4
        I would imagine that this only yields better performance if the host is set to decrease performance during times of a perceived lower load (schedutil, balanced, etc). But I wonder how many systems running VM(s) are actually power constrained.

        Comment


        • #5
          Originally posted by bachchain View Post
          You must be fun at parties
          I am. I complain about systemd and Rust and non-free blobs in the kernel and then I grab the microphone and sing karaoke. Just like everyone else here.

          Comment


          • #6
            Originally posted by EphemeralEft View Post
            I would imagine that this only yields better performance if the host is set to decrease performance during times of a perceived lower load (schedutil, balanced, etc). But I wonder how many systems running VM(s) are actually power constrained.
            Well, they offered a chromebook example (which are power/battery constrained). But for the Hyperscalers running millions of systems even a small improvement in power efficiency can result in substantial power (and cost) savings.

            Comment


            • #7
              Originally posted by andyprough View Post

              I am. I complain about systemd and Rust and non-free blobs in the kernel and then I grab the microphone and sing karaoke. Just like everyone else here.
              Karaoke:
              "Allllllll my ex's use X11.
              That's why I hang my hat in Wayland DEs"

              Comment


              • #8
                Originally posted by andyprough View Post
                then I grab the microphone and sing karaoke
                No, you'd go on a 20 minute rant about how the karaoke machine uses non-FSF approved firmware and then leave out of protest

                Comment


                • #9
                  Originally posted by bachchain View Post
                  No, you'd go on a 20 minute rant about how the karaoke machine uses non-FSF approved firmware and then leave out of protest
                  I'm not Richard Stallman. And none of this "you suck at parties" stuff has anything to do with the verified fact that Google built backdoors into their search and cloud products. And PRISM wasn't the only program, there was also MUSCULAR, which apparently exposed far more data to the surveillance dragnets, and was offshore, so not answerable even to the FISA courts.

                  Comment


                  • #10
                    Originally posted by EphemeralEft View Post
                    I would imagine that this only yields better performance if the host is set to decrease performance during times of a perceived lower load (schedutil, balanced, etc). But I wonder how many systems running VM(s) are actually power constrained.
                    Most of them, I would think. Power=heat. It's also extra workload that you might be able to wedge onto the node, if the node were more efficient.

                    Comment

                    Working...
                    X