Announcement

Collapse
No announcement yet.

Initial Benchmarks Of The Spectre "SWAPGS" Mitigation Performance Impact

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

  • #11
    Originally posted by skeevy420 View Post

    Both. Some we have, some we don't, some have been fixed, some won't be fixed.

    It's dumb to run unmitigated outside of benchmarks.
    It depends, if you have a high performance science cluster where only local people can run their massive parallel load, maybe they do not need that extra security. Similar to servers without remove shell logins or remote code execution I would assume, too, ..?

    Comment


    • #12
      Originally posted by DoMiNeLa10 View Post

      Every chip since the Pentium Pro is vulnerable to spectre.
      Atom? ,-)

      "…without any instruction reordering, speculative execution, or register renaming. The [] microarchitecture therefore represents a partial revival of the principles used in earlier Intel designs such as P5 and the i486 …" https://www.youtube.com/watch?v=suCj4eulTJg
      Last edited by rene; 07 August 2019, 12:42 PM.

      Comment


      • #13
        We proudly present:
        The speculative execution vulnerability of the week

        Comment


        • #14
          Microsoft patched this vulnerability on July 9, a full fucking month ago. While most Linux distributions only got a hold of the patches today. It is well-known that attackers monitor Windows Updates and can deduce security vulnerabilites from patches, so they can then go attack unpatched Windows systems and other OSes.

          This tells you everything you need to know about how much Microsoft "loves Linux", and how much Intel cares about security.

          Comment


          • #15
            Originally posted by DoMiNeLa10 View Post

            Every chip since the Pentium Pro is vulnerable to spectre.
            Correction: Every x86 based chip since the Pentium Pro. Once again, Itanium is immune.

            Comment


            • #16
              Originally posted by gamerk2 View Post

              Correction: Every x86 based chip since the Pentium Pro. Once again, Itanium is immune.
              Itanium could very well have it's own series of vulnerabilities. We are just not going to find out, ever.

              Comment


              • #17
                Originally posted by rene View Post
                Atom? ,-)
                Atom has been out-of-order since Silvermont in 2013.

                Comment


                • #18
                  Originally posted by hotaru View Post

                  Atom has been out-of-order since Silvermont in 2013.
                  I believe it's the speculative execution part that's required, not ooo. Even when linearly executed, the problem is a younger op executing before the older retires and flushes the bad path op, but it's not required to be out of order, or even supported.

                  Comment


                  • #19
                    Getting an AMD CPU doesn't sound too shabby right now.

                    Comment


                    • #20
                      Originally posted by jrdoane View Post
                      Getting an AMD CPU doesn't sound too shabby right now.
                      At least with Zen 2, no firmware update needed for Spectre and Spectre V4 fixed in hardware... and like else AMD not affected by Meltdown, Foreshadow, Spoller, Lazy FPU, MDS...

                      By their own EPYC Horizon live presentation that i am watching right now now https://www.youtube.com/watch?v=9Jn9NREaSvc

                      Last edited by dungeon; 07 August 2019, 06:44 PM.

                      Comment

                      Working...
                      X