1. 12
  1. 13

    TL;DR: He prefers x86 because of the more mature ecosystem built around it and it’s extremely well tested compatibility between implementations. He doesn’t like how on many ARM systems you have lots of random hardware hung off custom busses and the like.

    (As for core CPU architecture….x86_64 has certainly made it better, but I still have nightmares from years ago dealing with segmented memory…thankfully most of my early low-level stuff was on the M68k, God’s own architecture.)

    1. 2

      Segments can be powerful if you use them for POLA-enforcement throughout a system. Originals that got good reviews in pentesting were XTS-400’s STOP OS and Aesec GEMSOS kernel. Google’s Native Client later tried to use it in a medium-assurance design. Recently, Code Pointer Integrity is most clever with quite low overhead vs full memory safety or most SFI. Finally, I noted reading an Intel Atom paper that it’s still a faster way to isolate custom systems given accesses went from 1 to 2 cycles if segmented and 8 cycles if paging. That’s 4x speed advantage due to simplicity of segment in hardware implementation.

      Most likely, you were using a low-level, imperative language that forced you to manage segments you didn’t need by hand. That would suck. It’s not inevitable way to handle segments but it is what mainstream OS’s and C left us with.



    2. -3

      I prefer x86 because it’s so much faster. My 2008 AMD desktop is faster than the 2016 Raspberry Pi 3.

      1. 10

        Not a fair comparison. A current iPhone, or an ARM server board, is probably faster than your 2008 AMD desktop.

        1. -1

          8 year advantage though

          1. 1

            Consider pricing. It’s 35$ vs. whatever an Athlon 64 X2 cost in 2008.

            1. 0

              People are reporting me as a troll. Seems ARM vs x86 performance is a touchy subject on here.

              1. 5

                Yes. Clearly, that is the only possible reason the community are calling you a troll.

                1. 1

                  Your comment may have been made in good faith, but it sounds trolly. Obviously desktop chips plugged into a wall outlet will have a huge advantage over phone chips.

                  ARM vs Atom chips would be a more fair comparison.

                  And of course, ARM wins on performance/power.

                  1. 1

                    As you scale up chip sizes and performance, power consumption is a draw. Decoding logic is a minuscule part of the chip.