Comments by "Scott Franco" (@scottfranco1962) on "X86 Needs To Die" video.

  1. The 8086 series was a looser from the get-go. Software professionals preferred the cleaner architecture of the 68000 series, which did away with segmentation (a disease more than a technology). Indeed, the 8086 was in fact failing in the marketplace until IBM "rescued" it by using it in the IBM PC. That was the first crisis for the x86 family. The thing about segmentation is it is impossible to hide from higher level software. Even the language has to adapt to that hack, and indeed, C of the day had special features just to adapt to it. The result was that software was divided into x86 specific software and non-x86 software. Intel was happy about this because their users were non-portable. They doubled down on segmentation long after the size of chips enabled better architectures with the 80286, which could be explained as "you will like eating sh*t if we make it a standard". IBM again propped up the x86 family by releasing the IBM-PC/AT, which, even though it used the 80286, never saw wide use of a 80286 enabled operating system (cough OS/2). This carried x86 into the age of RISC. The x86 family entered its second crisis as improved 68000 processors, the Sparc and other RISC processors nipped at its heels. The introduction of the 80386 saved Intel, got rid of segmented mode, and allowed a Unix implementation of x86 for the first time. The next crisis for the x86 was when the series fell behind RISC processors in performance. Intel pulled off a genuine coup de gras by rearchitecting the x86 as an internal RISC that translated the crappy x86 instruction set to "ROPs" or internal RISC operations and made the CPU superscalar with Pentium. The final crisis for the x86 was when Intel tried to dump the dog of a CPU with Itanium, only to 180 again and join AMD with the "hammer" AMD64 bit arch. Now we are in the age of RISC-V. The x86 has become like a case of herpes, bad, but livable, and seemingly never to go away. We'll see.
    1
  2. I'm very impressed that the presenter can write backwards.
    1
  3. PS Von Neuman didn't invent sh*t. He took credit for the stored computer architecture because he collected notes from the group that came up with it and published them.
    1