Announcement

Collapse
No announcement yet.

Intel Preparing Linux Kernel Support For "Unaccepted Memory"

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

  • Intel Preparing Linux Kernel Support For "Unaccepted Memory"

    Phoronix: Intel Preparing Linux Kernel Support For "Unaccepted Memory"

    The latest patch series from Intel engineers worth noting for the Linux kernel is around implementing support for "unaccepted memory". Unaccepted memory is supported by the latest-generation AMD EPYC processors but not yet supported under Linux while Intel is preparing the kernel support for their next-gen Xeon CPUs having this capability...

    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
    This is unacceptable!

    Comment


    • #3
      They couldn't use some slightly different jargon?

      "Unaccepted Memory" makes it sound like the VM is a petulant child refusing to eat its greens.

      Comment


      • #4
        Originally posted by ssokolow View Post
        They couldn't use some slightly different jargon?
        This is Linux. Why would they want to?

        Comment


        • #5
          Sounds like a hardware-assisted implementation of memory ballooning. Maybe the author could mention that, since memory ballooning is the industry term for how it is done with VM-OS cooperative platform memory driver.
          Last edited by linuxgeex; 11 August 2021, 03:41 PM.

          Comment


          • #6
            Originally posted by ssokolow View Post
            They couldn't use some slightly different jargon?
            The UEFI forum chose the term (and embedded it into the spec). If you wanted a different name, the time for action would have been being part of the UEFI forum processes.

            Comment


            • #7
              Originally posted by linuxgeex View Post
              Sounds like a hardware-assisted implementation of memory ballooning.
              Not exactly the same, but trying to achieve some of the same benefits of being able to over-allocate real memory to guests. Memory ballooning lets the guest indicate when memory is no longer needed, so the underlying pages can be used elsewhere by the host, while unaccepted memory is memory that the guest has not yet even added to its environment to be used, and may not request allocation until actually needed by that guest. Since memory setup (especially for guests with a very large memory allocation) can take some time, being able to start the host faster to start the applications more quickly can have benefit for at least some use cases. As with much else, this appears to target/benefit the hyper-scalers.

              Comment


              • #8
                Originally posted by CommunityMember View Post
                The UEFI forum chose the term (and embedded it into the spec). If you wanted a different name, the time for action would have been being part of the UEFI forum processes.
                I guessed that. You misunderstood which "they" I was talking about.

                (Not that it matters. It's there and it's dumb, regardless of who's at fault.)

                Comment


                • #9
                  Originally posted by ssokolow View Post

                  I guessed that. You misunderstood which "they" I was talking about.

                  (Not that it matters. It's there and it's dumb, regardless of who's at fault.)
                  I skimmed the archives, and did not find your comment(s) to the UEFI forum. I probably just missed it. Can you point at where you told the forum they were wrong, and at fault, and proposed your approved name(s)?

                  Comment


                  • #10
                    Originally posted by CommunityMember View Post

                    I skimmed the archives, and did not find your comment(s) to the UEFI forum. I probably just missed it. Can you point at where you told the forum they were wrong, and at fault, and proposed your approved name(s)?
                    Do you really want to hold everyone to that standard? ...especially in this forum? ...because, if not, it just makes you sound like a hypocrite, wielding the "you didn't participate in the design" hammer whenever someone observes a flaw in anything, thus making it impossible for anyone to criticize flaws in the future.

                    ...but I'm sure you aren't that foolish... especially when I was trying to be humorous. That'd just make you humorless... and nobody likes a humorless person.

                    Comment

                    Working...
                    X