Announcement

Collapse
No announcement yet.

The ~200 Line Linux Kernel Patch That Does Wonders

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

  • kerneloftruth, so far, your patches produce the fastest kernel 2.6.37 for me, but, it still not as fast as zen kernel version 2.6.36-zen3 with bfs. it gets pretty close though.

    Comment


    • Originally posted by raj7095 View Post
      the first patchset worked perfectly, the second patch fails at compile time for me. what does the second patch do anyways?
      you patched 2.6.37 with the 1st patchset

      and AFTER THAT applied the 2nd patchset ?

      or did you try to compile 2.6.37 *only with* the 2nd patchset ?

      I've modified it that way that it should apply cleanly *on top* of 2.6.37 + 1st patchset


      in a nutshell it should improve the autogroup (or group scheduling in general) even more:

      [wake_afine fixes/improvements 0/3] Introduction

      I've been looking at the wake_affine path to improve the group scheduling case
      (wake affine performance for fair group sched has historically lagged) as well
      as tweaking performance in general.

      The current series of patches is attached, the first of which should probably be
      considered for 2.6.38 since it fixes a bug/regression in the case of waking up
      onto a previously (group) empty cpu. While the others can be considered more
      forwards looking.

      Originally posted by raj7095
      kerneloftruth, so far, your patches produce the fastest kernel 2.6.37 for me, but, it still not as fast as zen kernel version 2.6.36-zen3 with bfs. it gets pretty close though.
      yeah, interactivity-wise BFS from 2.6.36-zen* still felt better

      but if you take 2.6.37-zen* 's base and use these 2 patchsets (I've built my own similar kernel patchset like zen) it's the smoothest kernel EVER - so far

      the best indicator for that is:

      it almost never interrupts streaming / full throughput of my ethernet card during large i/o operations (meaning copying over large amounts of data between harddrives with device-mapper and encryption) - this also includes listening to webradio

      Comment


      • you can't apply the first patch onto the zen kernel, can you? btw, i patched in the wrong order, thats why it didnt work before. i corrected my mistake and the patches were successful.

        Comment


        • Originally posted by raj7095 View Post
          you can't apply the first patch onto the zen kernel, can you? btw, i patched in the wrong order, thats why it didnt work before. i corrected my mistake and the patches were successful.
          you could try

          but I highly doubt that they would apply cleanly

          therefore you would need to fix those rejects (if that even worked afterwards)

          Comment


          • ok, a few hunks fail, but i can fix that.

            Comment


            • kerneloftruth, i got an interesting investigation. if you compile kernel 2.6.36 with ck patches, instead of the zen for bfs, you will get the exact same amount of interactivity in both 2.6.36 and 2.6.37 versions. so, that means the problem isn't in bfs, but apparently some sorta change in zen kernel. ok, so i tried zen kernel with the second patch (most of the patches in the first one are already there) with cfs autogroup and all those cgroup option, and i am pretty impressed with the performance increase. i can again compile with 4 jobs and still play video games and watch hd videos on youtube at the same time.

              Comment


              • Originally posted by raj7095 View Post
                u say u dont have smt, correct? i do, on a dual core, maybe that makes a difference in the new bfs. oh well, i will just stick to kerneloftruth's patches or the 2.6.36 version of zen kernel.
                sorry, i got confused with all the machines i''ve got. I have a dual core T4300 (2 cores w/o SMT ), i3-530 (2 cores with SMT), i7-920 (4 cores with SMT). All of them seem to run 2.6.37-ck1 just fine, but my desktop machine (the one I wrote about yesterday) is the i3-530 with SMT.

                Comment


                • Originally posted by raj7095 View Post
                  kerneloftruth, i got an interesting investigation. if you compile kernel 2.6.36 with ck patches, instead of the zen for bfs, you will get the exact same amount of interactivity in both 2.6.36 and 2.6.37 versions. so, that means the problem isn't in bfs, but apparently some sorta change in zen kernel. ok, so i tried zen kernel with the second patch (most of the patches in the first one are already there) with cfs autogroup and all those cgroup option, and i am pretty impressed with the performance increase. i can again compile with 4 jobs and still play video games and watch hd videos on youtube at the same time.
                  glad it helps with the interactivity on your desktop

                  but the largest bulk of those patches are NOT included in zen

                  Comment


                  • i found another weird result. if you compile firefox on one kernel, it will work the best just on that specific kernel. i dont get it???

                    Comment


                    • Originally posted by raj7095 View Post
                      i found another weird result. if you compile firefox on one kernel, it will work the best just on that specific kernel. i dont get it???
                      dunno how that could be

                      only in the case that CFLAGS might not have been handed over correctly to gcc and using

                      -march=native

                      if you're still looking for some newer (2.6.37-ish) kernel that is similar to zen

                      try out my patchset: 2.6.37_plus kernel with coordinate flush, kztmem or TOI

                      kztmem seems to improve things (in addition to the patches already offered in zen) even more

                      Comment


                      • Originally posted by kernelOfTruth View Post
                        dunno how that could be

                        only in the case that CFLAGS might not have been handed over correctly to gcc and using

                        -march=native

                        if you're still looking for some newer (2.6.37-ish) kernel that is similar to zen

                        try out my patchset: 2.6.37_plus kernel with coordinate flush, kztmem or TOI

                        kztmem seems to improve things (in addition to the patches already offered in zen) even more
                        co-incidentally, i just stumbled across that patchset today myself. i am currently compiling it with kztmem, and custom optimization flags, -O3 -march=native -pipe -flto -floop-parallelize-all -floop-block -floop-interchange -floop-strip-mine -ftree-loop-linear -fexcess-precision=fast
                        i can see a huge set of nice-sounding patches that i never saw in zen kernel before. this patchset looks amazing looking at the improvement list. just one question, what is kztmem? i remember the zram option said something about increasing performance a lot, and kztmem needs zram, so i am assuming it tries to use zram as much as possible for programs that can make its use, is that correct?

                        Comment


                        • nice, it boots up. interesting, cursor doesn't stop at all. i mean really, i click on firefox and while its loading and dont see a single slowdown while i am moving it. this seems like a very good start. i will see how well it performs compared to the zen kernel. it gave me an amazing first impression, thats for sure.

                          Comment


                          • download link for the patch?

                            Originally posted by dagger View Post
                            Just tested this patch and from what I can see it doesn't make anything faster. It makes everything *MUCH* smoother by reducing resources to processes which use a lot of CPU - (like make -j64 on linux kernel). Compilation takes a bit longer, but system is fully usable during that time.

                            This is by far one of the best patches I've tested so far! Outstanding work. Thanks for bringing that up!
                            Hi, where can I download this patch so that I can try it?

                            Comment


                            • Download 2.6.38-rc3, there is the patch in.

                              Comment


                              • Originally posted by towo2099 View Post
                                Download 2.6.38-rc3, there is the patch in.
                                Hi towo2099,
                                Thanks for that. I'm downloading it now...Will be trying it out. By the way, would I need to recompile the kernel and configure the kernel with any specific settings? Is kernel pre-emption set to ON, timer frequency set 1000Hz and the lot done by default?
                                I have a Toshiba Satellite Pro.
                                Looking forward to experiencing the "smoothness" boost with this patch!

                                Comment

                                Working...
                                X