Announcement

Collapse
No announcement yet.

AMD Secure Encrypted Virtualization Updated For Linux

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

  • #11
    wodencafe
    That AMD thinks that a 3rd party security audit is somehow an adequate substitute to open source PSP code is indeed unfortunate.
    But what do you suggest buying instead?

    Comment


    • #12
      Originally posted by wodencafe View Post
      Who still wants to use AMD after that PSP fiasco?

      They basically give a big middle finger to the Security Community.
      Intel manages to be consistently worse with their ME, btw.

      Comment


      • #13
        Originally posted by Qaridarium
        Not in the same market as EPYC.

        Comment


        • #14
          Originally posted by wodencafe View Post
          Who still wants to use AMD after that PSP fiasco?

          They basically give a big middle finger to the Security Community.
          What fiasco? Ahem, I believe it was intel that just had a critical PSP security flaw, not AMD. Sounds like your facts are a bit off.

          Comment


          • #15
            Originally posted by wodencafe View Post
            Who still wants to use AMD after that PSP fiasco?

            They basically give a big middle finger to the Security Community.
            not sure what PSP fiasco you mean to be honest but in my case it probably won't bother me much since all my hypervisor systems don't have network access only the VMs through VFIO or specialized bridges, so my main server will always be safe(as possible). What i need is the actual RAM per VM encryption because that way i'm 90% sure if someone sniff something it has to be a freaking magician instead of some noob looping a pointer getting lucky.

            Obviously i'm aware this won't magically make everything 100% secure but it does make it harder and in security that is all you can hope achieve, the only way this will make me disable it is if an exploit can run from inside the VM and reach ring -3 below the OS(and as far as i know no one has do so on ARM so far) and is easy enough for any idiot to do it, anyway i doubt it will be worse than intel's "security" extensions on the Xeons

            Comment


            • #16
              Originally posted by chithanh View Post
              Not in the same market as EPYC.
              How are they not in the same market? POWER9 and EPYC are designed to be used in servers and high-end workstations. I don't follow.

              Comment


              • #17
                Originally posted by Dopefish View Post
                How are they not in the same market? POWER9 and EPYC are designed to be used in servers and high-end workstations. I don't follow.
                Talos II is a workstation. EPYC is a server CPU.
                Workstations that compete with Talos will use Ryzen Threadripper.

                Originally posted by Qaridarium
                right. i think he want so say something like: "it is not X86 and because of this my so loved closed source software wlll not run."
                The problem of running x86 software on non-x86 CPUs with good performance has been solved more than two decades ago with DEC FX!32.

                Comment


                • #18
                  Maybe you can read more closely, you missed where I wrote that Talos II is a workstation. Whether Power9 is exclusively a workstation CPU (it is not) is inconsequential to this question.

                  About EPYC workstations, at least the dual socket mobos we have seen so far are decidedly targeted at server and not workstation markets.
                  Last edited by chithanh; 27 July 2017, 02:39 PM.

                  Comment

                  Working...
                  X