Optimizing Linux MD Bitmap Code Yields 89% Throughput Boost For Quad SSDs

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts
  • jeisom
    Senior Member
    • Mar 2013
    • 270

    #11
    I wonder if this improves a dual raid nvme system at all and not just quad or higher groups? I am using a pair of gen 4 drives in raid 0 and when I have tried games it appears that most of the load time is decompression/processing the data, not loading off a disk.

    Comment

    • pWe00Iri3e7Z9lHOX2Qx
      Senior Member
      • Jul 2020
      • 1578

      #12
      For home use if I wanted some fault tolerance with 4x NVMe I'd choose RAID 10 instead, even at 50% usable space. But really I'd just RAID 0 those suckers and snapshot / backup anything important to another pool of drives that has some fault tolerance. 4x 4TB WD SN850X in RAID 0 FTW!

      Comment

      • sophisticles
        Senior Member
        • Dec 2015
        • 2590

        #13
        Fine I will be the douche that says it: this comes from Chinese researchers working at a Chinese university, last time we trusted anything from a Chinese developer it took 2 years to realize that a backdoor existed.

        I don't care who reviews this code, I can see the articles in a few years already about how this compromised systems that used it and no one noticed until a Microsoft engineer was having problems with his system.

        Comment

        • npwx
          Senior Member
          • Mar 2022
          • 134

          #14
          Originally posted by sophisticles View Post
          Fine I will be the douche that says it: this comes from Chinese researchers working at a Chinese university, last time we trusted anything from a Chinese developer it took 2 years to realize that a backdoor existed.

          I don't care who reviews this code, I can see the articles in a few years already about how this compromised systems that used it and no one noticed until a Microsoft engineer was having problems with his system.
          Assuming you're not just trolling, that is something exceptionally stupid to say, even for your standards. There are plenty of chinese developers regulary contributing code to the Linux kernel and elsewhere.

          Comment

          • sophisticles
            Senior Member
            • Dec 2015
            • 2590

            #15
            Originally posted by npwx View Post

            Assuming you're not just trolling, that is something exceptionally stupid to say, even for your standards. There are plenty of chinese developers regulary contributing code to the Linux kernel and elsewhere.
            Show me a single contributor from China that "regularly" contributes code to the Linux kernel.

            I don't need "plenty", just one, and he or she needs to be contributing code at regular intervals and must be physically located in China and working for a Chinese university.

            Just one.

            Comment

            • lowflyer
              Senior Member
              • Aug 2013
              • 911

              #16
              Originally posted by npwx View Post

              Assuming you're not just trolling, that is something exceptionally stupid to say, even for your standards. There are plenty of chinese developers regulary contributing code to the Linux kernel and elsewhere.
              No. that's NOT stupid. Not even remotely exceptionally. It's just the kind of prudence that should be enforced by now. The signs of changing time's are getting obvious by now. You must be living under a rock if you don't see them.
              Last edited by lowflyer; 12 December 2024, 03:01 AM. Reason: fix a typo. better wording

              Comment

              • schmidtbag
                Senior Member
                • Dec 2010
                • 6614

                #17
                Originally posted by lowflyer View Post
                No. that's NOT stupid. Not even remotely exceptionally. It's just the kind of prudence that should be enforced by now. The signs of changing time's are getting obvious by now. You must be living under a rock if you don't see them.
                It's stupid if you think other governments aren't just as capable of doing the same thing. At least if the Chinese government is spying on me, I can probably extrapolate what it is they're trying to find and what they intend to do with it.
                That's not to say we shouldn't concern with what's being submitted to the kernel, but rather, being sourced from China isn't a compelling reason to care more; all code from all sources should be vetted.

                Comment

                • lowflyer
                  Senior Member
                  • Aug 2013
                  • 911

                  #18
                  Originally posted by schmidtbag View Post
                  It's stupid if you think other governments aren't just as capable of doing the same thing. At least if the Chinese government is spying on me, I can probably extrapolate what it is they're trying to find and what they intend to do with it.
                  That's not to say we shouldn't concern with what's being submitted to the kernel, but rather, being sourced from China isn't a compelling reason to care more; all code from all sources should be vetted.
                  I did not mention "China". But having had an issue with a Chinese developer recently in our company I'd say it is foolish to assume that similar issues do not happen in open source projects. This is not against China specifically. There are numerous "forces" attempting to exercise power over their adversaries. Open source projects are a prime target for them because they are ... exactly: "open".

                  Thinking that "you can extrapolate the intentions" of a malicious actor is blue-eyed. You would not even know that you have been set-up.

                  Commonplaces like "all code from all sources should be vetted" do not help because that's like saying "if the others would do like I do ...". Saying "being sourced from China isn't a compelling reason" is being blue-eyed.

                  Let me repeat: This is not about China specifically. A few years back I had another incident that was caused by a German developer.

                  Comment

                  • schmidtbag
                    Senior Member
                    • Dec 2010
                    • 6614

                    #19
                    Originally posted by lowflyer View Post
                    I did not mention "China". But having had an issue with a Chinese developer recently in our company I'd say it is foolish to assume that similar issues do not happen in open source projects. This is not against China specifically. There are numerous "forces" attempting to exercise power over their adversaries. Open source projects are a prime target for them because they are ... exactly: "open".
                    Uh... you don't have to mention China; they're kinda the subject of this discussion, so it's implied.
                    Open source's weakness is also its strength: while just about anyone can make malicious/devious changes without much restriction, just about anyone can identify such changes too. So really, if large open-source projects are to be safe, all it really takes is having enough eyes (ideally anonymous) inspect the commits for suspicious activity. Personally, I think this is necessary beyond security reasons - it's also good to identify potential conflicts or inefficient code.
                    Thinking that "you can extrapolate the intentions" of a malicious actor is blue-eyed. You would not even know that you have been set-up.
                    That's not really the point I was making and your own response seems aware of this. I think it's quite likely that a government could be spying on me and I'd be oblivious, but that doesn't change the fact that, depending on what the software is and which organization it is, I can extrapolate what it is they're trying to collect from me. When it comes to China specifically, there is very little they can collect from me that would matter. There's virtually nothing incriminating about me (at least in my home country) that they could collet and use against me, I have no military or political influence, I very rarely buy anything directly from China, and I have no intentions on ever going to China. So, pretty much all data they could ever collect from me is powerless; if they want to hurt me, it will be from falsified data.
                    I would be more worried about a NATO country spying on me, since they have a lot more power over my life.
                    Commonplaces like "all code from all sources should be vetted" do not help because that's like saying "if the others would do like I do ...". Saying "being sourced from China isn't a compelling reason" is being blue-eyed.
                    I don't really get what you're saying here, but I think you'd understand that obviously, you don't have people of the same demographic doing the vetting. That's like those papers written on social behaviors where only college students were tested/surveyed - that paper is effectively moot because all they did was target a single narrow demographic, and therefore says nothing about social behavior as a whole. Vetting code for safety and stability is the same - you need to diversify who is checking, and they should be picked at random. Besides, with proper "influence", anybody can turn a blind eye or even become corrupt themselves, so that's why you need anonymity and randomization.
                    Let me repeat: This is not about China specifically. A few years back I had another incident that was caused by a German developer.
                    That was kinda my point all along - fearing a developer's contributions because of their country of origin is a stupid notion, because just about every country has reasons for doing something malicious or devious. Again: China was the subject here, hence my response. Otherwise, why reply?

                    Comment

                    Working...
                    X