Announcement

Collapse
No announcement yet.

Another Potential Performance Optimization For KPTI Meltdown Mitigation

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

  • Another Potential Performance Optimization For KPTI Meltdown Mitigation

    Phoronix: Another Potential Performance Optimization For KPTI Meltdown Mitigation

    Now that the dust is beginning to settle around the Meltdown and Spectre mitigation techniques on the major operating systems, in the weeks and months ahead we are likely to see more performance optimizations come to help offset the performance penalties incurred by mitigations like kernel page table isolation (KPTI) and Retpolines. This week a new patch series was published that may help with KPTI performance...

    http://www.phoronix.com/scan.php?pag...I-Global-Pages

  • #2
    It's incredible how easily Intel can get away with it.

    Comment


    • #3
      Originally posted by eydee View Post
      It's incredible how easily Intel can get away with it.
      That remains to be seen. They're being slapped left and right with law suits, of which the results remain to be seen. I don't think they're going to get away with it that easily, actually. It's not going to kill off Intel, obviously. But still, I'd imagine this is going to end up costing them, quite a lot actually.

      First in the settlements which are going to be the obvious result of all the lawsuits against them and secondly in the loss of market share. AMD gave them a blow with Ryzen and the whole Meltdown-debacle gave them a blow as well. And yes, I explicitly mentioned only Meltdown there. Since Spectre is more global, not solely limited to Intel. For a lot of people, there simply are no BIOS updates available to help with the mitigation, they will be forced to purchase new hardware and with Ryzen now also featuring desktop APUs in its lineup, I'd wager to guess that a lot of people will be eye-ing AMD.

      I know I am.

      Comment


      • #4
        Having trouble understanding this from the article: "Intel developer Dave Hansen discovered that the back when KPTI was known as KAISER allowed the user/kernel shared areas to be marked global that would reduce the TLB overhead." Should it be "[...] known as KAISER, allowing the user/kernel shared areas to be marked as global would reduce the TLB overhead."?

        Comment


        • #5
          Originally posted by mmmbop View Post
          Having trouble understanding this from the article: "Intel developer Dave Hansen discovered that the back when KPTI was known as KAISER allowed the user/kernel shared areas to be marked global that would reduce the TLB overhead." Should it be "[...] known as KAISER, allowing the user/kernel shared areas to be marked as global would reduce the TLB overhead."?
          Agreed, the original sentence:
          Intel developer Dave Hansen discovered that the back when KPTI was known as KAISERallowed the user/kernel shared areas to be marked global that would reduce the TLB overhead.
          is quite off.

          It should probably read:
          Intel developer Dave Hansen discovered that when KPTI was still known as KAISER it allowed the user/kernel shared areas to be marked global which reduces the TLB overhead.

          Comment


          • #6
            Originally posted by iyxwsoekthsv View Post
            Agreed, the original sentence:is quite off.

            It should probably read:

            Intel developer Dave Hansen discovered that when KPTI was still known as KAISER it allowed the user/kernel shared areas to be marked global which reduces the TLB overhead.
            I agree. Michael if you could amend the article, iyxwsoekthsv said it in much more natural english.

            Comment

            Working...
            X