Announcement

Collapse
No announcement yet.

The Linux 2.6.36 Kernel Is Now Out There

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

  • The Linux 2.6.36 Kernel Is Now Out There

    Phoronix: The Linux 2.6.36 Kernel Is Now Out There

    The Linux 2.6.36 kernel is now out there on the Internet. After an unexpected delay and some other slowdowns in the 2.6.36 development cycle, Linus tagged the 2.6.36 kernel this afternoon...

    http://www.phoronix.com/vr.php?view=ODY5NA

  • #2
    I'm reasonably sure the heavy I/O responsiveness patches are already in 2.6.36. I used one of the .36-rc's and it was phenomenal with large file operations.

    Comment


    • #3
      Originally posted by etnlWings View Post
      I'm reasonably sure the heavy I/O responsiveness patches are already in 2.6.36. I used one of the .36-rc's and it was phenomenal with large file operations.
      Not all of them, the more untested were held back for 2.6.37.

      Comment


      • #4
        No fanotify!

        Originally posted by phoronix View Post
        Phoronix: The Linux 2.6.36 Kernel Is Now Out There

        The Linux 2.6.36 kernel is now out there on the Internet. After an unexpected delay and some other slowdowns in the 2.6.36 development cycle, Linus tagged the 2.6.36 kernel this afternoon...

        http://www.phoronix.com/vr.php?view=ODY5NA
        Fanotify system calls were pulled at the last minute over ABI concerns. See the release announcement.

        Comment


        • #5
          Originally posted by Xake View Post
          Not all of them, the more untested were held back for 2.6.37.
          Well that will be interesting, considering current performance (on relatively old hardware, even). I stand corrected.

          Comment


          • #6
            Oh God please desktop responsiveness patches.

            Comment


            • #7
              This seems to be the topic of the day, so here it goes once again:

              Desktop responsiveness patches FTW \m/

              Comment


              • #8
                typo: debuging should be debugging

                Comment


                • #9
                  I have been a Linux junky for many years. But recently I came across a data point which really baffled me for and put shameful dent in my pride of Linux.

                  At our company, one of the things that was forced down our throats were laptops with Windows 7 running. We had to use VMs to run Linux for our development.

                  Being a Linux zealot, I reversed the roles. Ran Linux on host and Windows 7 in VM. I noticed that just one Windows 7 VM running, with 1GB given to it, a project build (which is largish) can make the desktop completely unresponsive: mouse freezes, desktop can't be changed, Windows7 VM doesn't refresh its window etc. And this is with 2.6.36-rc7.

                  Reverse the roles, run 3 Linux VMs inside Windows 7 with 1GB of RAM each. Build the same project inside any of those VMs and everybody (the host as well as all the VMs) keeps chugging along nicely. And surprisingly, the build (which is pretty linear single thread) inside VM finished faster.

                  That hurt bad! Why is it that we can't find this bottleneck and fix it after so many years?

                  Comment


                  • #10
                    I have been a Linux junky for many years. But recently I came across a data point which really baffled me for and put shameful dent in my pride of Linux.

                    At our company, one of the things that was forced down our throats were laptops with Windows 7 running. We had to use VMs to run Linux for our development.

                    Being a Linux zealot, I reversed the roles. Ran Linux on host and Windows 7 in VM. I noticed that just one Windows 7 VM running, with 1GB given to it, a project build (which is largish) can make the desktop completely unresponsive: mouse freezes, desktop can't be changed, Windows7 VM doesn't refresh its window etc. And this is with 2.6.36-rc7.

                    Reverse the roles, run 3 Linux VMs inside Windows 7 with 1GB of RAM each. Build the same project inside any of those VMs and everybody (the host as well as all the VMs) keeps chugging along nicely. And surprisingly, the build (which is pretty linear single thread) inside VM finished faster.

                    Laptop has 4GB of RAM and Windows 7 (32-bit) sees only 3.4GB while Linux sees all 4GB of it, so technically Linux is working with more RAM.

                    That hurt bad! Why is it that we can't find this bottleneck and fix it after so many years?

                    Comment


                    • #11
                      I can't seem to edit my posts on this forum...I just wanted to add the 4GB RAM part.

                      Comment


                      • #12
                        Originally posted by devsk View Post
                        That hurt bad! Why is it that we can't find this bottleneck and fix it after so many years?
                        Well, maybe after all the desktop responsiveness patches are applied it will be better. OTOH disk performance is better with linux: http://www.phoronix.com/scan.php?pag..._win7_ws&num=4 If only the user experience didn't suffer so much...

                        Comment


                        • #13
                          Originally posted by devius View Post
                          Well, maybe after all the desktop responsiveness patches are applied it will be better. OTOH disk performance is better with linux: http://www.phoronix.com/scan.php?pag..._win7_ws&num=4 If only the user experience didn't suffer so much...
                          The problem is that because Linux insists on caching too much to improve throughput, it likes to swap stuff than give up on caches. Which immediately introduces a big bottleneck. Swapping path is extremely slow and has probably not been looked into for years.

                          Note that even if you swapoff all your devices, kswapd will still hunt down pages which are on disk and read-only (excutable pages e.g.) and evict them thinking that it can read them back at any time. Of course, that time comes when I move my mouse and X/kwin/plasma-desktop needs that page.

                          My issue was seen with 2.6.36-rc7, so some of the desktop responsiveness patches are in there.

                          Comment


                          • #14
                            RS690 HDMI audio support

                            It was said here: http://www.phoronix.com/forums/showthread.php?t=25084 that the 2.6.36 will provide RS690 hdmi audio support. I tried it and it is still not working for me. Does anybody know which git changeset provided the fix ? Is a new version of the xorg driver needed ? Everything seems ok for me, the device is listed in aplay -l and enabled in the mixer but still no sound...

                            Comment

                            Working...
                            X