Announcement

Collapse
No announcement yet.

Finally, Reiser4 Benchmarks Against EXT4 & Btrfs

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

  • #31
    Originally posted by FireBurn View Post
    I think the biggest problem here is Michael don't know how to make a kernel for Ubuntu, so he's used Zen from a PPA / http://liquorix.net/

    Correct me if I'm wrong here
    I spin my own kernels frequently... It was Zen patches from source and not a PPA.
    Michael Larabel
    http://www.michaellarabel.com/

    Comment


    • #32
      Hi,

      Can you please test on hard drives. I know SSDs are the way to the future, but HDs are still everywhere right now and HD testing can show filesystems with better allocation/locality optimizations

      Also, SSDs will 'flatten' the benchmarks, plus most computers STILL have hard drives on them, right?

      xoxo

      Comment


      • #33
        Originally posted by Michael View Post
        I spin my own kernels frequently... It was Zen patches from source and not a PPA.
        So why on earth did you test Zen rather than stock + reiser4?

        Comment


        • #34
          Originally posted by FireBurn View Post
          So why on earth did you test Zen rather than stock + reiser4?
          I'm quite interested in knowing this.

          Also, do you take into consideration the suggestions in comments?

          I think SSDs are fine, but tests done to HDDs are necessary too.

          Also, recommended tweaks from each FS should be tested too.

          I think Phoronix tests show too much statistics but also lacks a lot more depth, that could make it more interesting both for developers and advanced users/admins.

          Even if my nerd growing inside me loves to see statistics and tests, I think the graphical representation you use is quite bloated. I would prefer one big table with all on it instead tons of them with bars.

          Do you know if PTS is used in Linux kernel development and distros and accepting suggestions or patches from them? What about Linux Foundation and this project? Maybe your project with tons more hardware diversity and help from other developers could get into a more powerful project.

          Michael: While I appreciate your work here, I would like to see more complete replies so we can learn and know your opinion to different messages posted here. Maybe this could benefit to both sides. I know this sometimes can be a lot time consuming, but replying fast is not mandatory. You could do one big reply to all topics instead just one.

          Comment


          • #35
            Reiser 3 also lost data regularly for me. What bothered me even more than that was the "why should we care" reaction from the developers to the bug report. I can't find it anymore but the carelessness was shocking. Unless the Reiser 4 developers don't even remotely know the Reiser 3 ones, I wouldn't use it if it were the only FS other than FAT16.

            Also, is nobody irritated by this sentence from the article?
            "Aside from some of the tests failing, we ran into no other issues with Reiser4."

            Well, sometimes it just doesn't work for no apparent reason, but when it does, it mostly does its job. WTF?

            Comment


            • #36
              Originally posted by mudd1 View Post
              Also, is nobody irritated by this sentence from the article?
              "Aside from some of the tests failing, we ran into no other issues with Reiser4."

              Well, sometimes it just doesn't work for no apparent reason, but when it does, it mostly does its job. WTF?
              Get a clue. It's been said over and over in this thread that michael used broken reiser4 patches instead of "normal" ones provided by edward shishkin.

              Comment


              • #37
                Originally posted by Michael View Post
                I spin my own kernels frequently... It was Zen patches from source and not a PPA.
                so you used crap - one question, do you try to hurt reiser4 on purpose?

                this is the place to get stable, up to date and WORKING patches. Straight from Edward:
                ftp://ftp.kernel.org/pub/linux/kerne...eiser4-for-2.6

                Comment


                • #38
                  Originally posted by energyman View Post
                  so you used crap - one question, do you try to hurt reiser4 on purpose?

                  this is the place to get stable, up to date and WORKING patches. Straight from Edward:
                  ftp://ftp.kernel.org/pub/linux/kerne...eiser4-for-2.6
                  Reiser sucks. Get over it.

                  Comment


                  • #39
                    Originally posted by BlackStar View Post
                    Reiser sucks. Get over it.
                    and you are just a troll. A bad one. So go home and be nice to your mother.

                    Comment


                    • #40
                      Originally posted by bugmenot View Post
                      Get a clue. It's been said over and over in this thread that michael used broken reiser4 patches instead of "normal" ones provided by edward shishkin.
                      So the fact that Michael tested the wrong patches is an excuse for a summary that implies what he tested was fine for the most part?

                      There are two issues, one is the Reiser4 fans complaint about the patch set that was tested. I won't even pretend to be know or even care about this one as I am unlikely to ever use a Reiser filesystem.

                      The second issue is with the content of the report on what was tested. This is, hard as it may be to believe, a separate issue. Saying "it did well, but some tests would not run" implies it's a lot better performing than saying "While it passed some tests, the entire range of test applications could not complete on this filesystem" even if logically they are much the same.

                      I'm sorry, but my take of what was written was implying that the patchset he tested was fit for use, while in fact his own test results show it is clearly NOT fit for use. (This is a filesystem, test failure is NOT acceptable!)

                      Comment


                      • #41
                        Originally posted by energyman View Post
                        and you are just a troll. A bad one. So go home and be nice to your mother.
                        Oh, and you aren't one?! Your posts are nothing but the flaming and trolling.

                        Comment


                        • #42
                          pejakm, that from you?I am shocked, really, I am.

                          And RobbieAB, using broken patches and then mentioning that some tests did not run, is hurtting reiser4 and is bad for phoronix. You sound like a little, butthurt extX fanboy.

                          If Michael would have used the CORRECT patches, I am pretty sure all tests would have run. But so he hurt reiser4. Deliberately or not?

                          'We' complain that the test was not fair. And your own words support that. The test is BULLSHIT. Ok?

                          I have tried zen in the past and there were ALWAYS problems with that pile of crap. Always.

                          I can't remember the last time I had any problems with reiser4 using Edward's fine patches. Early 2007 AFAIR.

                          Why did Micheal use patches from some obscure project instead of the original ones? There is no sense in Micheal's choice

                          Comment


                          • #43
                            Originally posted by energyman View Post
                            pejakm, that from you?I am shocked, really, I am.
                            I'm sorry to hear that, but realize that being impolite in your posts is definitely not a good way to discuss something, and will get you nowhere. You keep flaming Michael for using broken patches: I'm using zen patches for about a year and a half, and the only issues I had were due to recently introduced BFS (which you don't have to use, by the way), but this also is fixed very quickly. As I understand, zen kernel uses Edwards reiser4 patches.

                            How about politely ask Michael to use vanilla kernel just with the reiser4 patchset? Maybe he would respond to that?

                            Comment


                            • #44
                              Originally posted by energyman View Post
                              And RobbieAB, using broken patches and then mentioning that some tests did not run, is hurtting reiser4 and is bad for phoronix. You sound like a little, butthurt extX fanboy.
                              energyman, I was specifically NOT commenting on the choice of patches. I was referring to the manner of reporting on what was tested, which to my mind was poor. If the wrong patch set was chosen, that makes it even worse, but to my mind the report implied that the patch set chosen is usable.

                              Yes, I use ext3 mostly, but that choice was based on specific reasons going back many years, and I have never had any reason to consider changing. For me, it does what I want. That is enough reason to use it, FOR ME. Your usage and desires may be different.

                              Originally posted by energyman View Post
                              If Michael would have used the CORRECT patches, I am pretty sure all tests would have run. But so he hurt reiser4. Deliberately or not?

                              'We' complain that the test was not fair. And your own words support that. The test is BULLSHIT. Ok?
                              You complain that he tested the wrong patches. I can't comment on that.

                              My complaint is that he misrepresents what he tested. By his own account, what he tested outright failed some tests. Is this acceptable in a file-system? I say not. Does he outright slam the patchset he tested? No. Why not?

                              The fact he should have tested a different patchset, or carried out tests on different hardware, don't change the fundamental complaint I have which is that he reports his results in a questionable manner.

                              Originally posted by energyman View Post
                              I have tried zen in the past and there were ALWAYS problems with that pile of crap. Always.

                              I can't remember the last time I had any problems with reiser4 using Edward's fine patches. Early 2007 AFAIR.

                              Why did Micheal use patches from some obscure project instead of the original ones? There is no sense in Micheal's choice
                              I can't comment on the zen patchset, as I have never used it. I mostly use either a vanilla kernel, gentoo-sources, or a lightly patched torvalds git pull.

                              Only Micheal can explain his choice of patchsets, I have never even attempted to defend that. Maybe if he used a different patchset it would have passed those tests and wowed us all with it's performance. Does that change my comments that he is mis-reporting what he DID test?

                              Comment


                              • #45
                                Originally posted by pejakm View Post
                                I'm sorry to hear that, but realize that being impolite in your posts is definitely not a good way to discuss something, and will get you nowhere. You keep flaming Michael for using broken patches: I'm using zen patches for about a year and a half, and the only issues I had were due to recently introduced BFS (which you don't have to use, by the way), but this also is fixed very quickly. As I understand, zen kernel uses Edwards reiser4 patches.

                                How about politely ask Michael to use vanilla kernel just with the reiser4 patchset? Maybe he would respond to that?
                                for the same reason he let ext3 without barriers run against reiserfs and xfs with barriers...

                                and zen patches are not Edward - at least not 'vanilla' - or there wouldn't have been that -rcX clusterfuck where it was so badly broken, they warned in their notes about it.
                                If you want reliable results, Edward'S patches are the only choice. How can you even can come up with the ida that extracting patches from a very experimental and very untested kernel is a good idea?

                                Comment

                                Working...
                                X