Now that we know the M1 chip is giving Apple a big advantage to the Intel/windows combo. When will Intel/windows start going the Arm route?
Fun but largely irrelevant fact - Intel used to make an ARM processor (it was originally developed by DEC, Intel "inherited" it when DEC split up:
https://www.intel.com/pressroom/archive/releases/1999/em033199.htm
However...
The 68000 was better than Intel in 1984
The ARM2 was better than Intel in 1987
The PPC was better than Intel in 1992
DEC Alpha, MIPS, SPARC were all more powerful than Intel in their prime.
...yet everybody is still using "wintel" PCs and even Mac users are whingeing over the demise of Boot Camp. if the industry went with the best technical solution we'd all have been using Acorn Archimedes or Commodore Amiga-descended systems today.
In the early 1980s, Intel had pretty much lost their original lead in the personal computer market to the Zilog Z80. The x86 was a so-so, stopgap, pseudo-16-bit kludge of a chip made by Intel in the early 80s
to tide them over their problematic 32 bit iAPX 432 chip. The personal computing world was pretty much waiting for cheap hard drives so they could dump stripped-down operating systems like CP/M and switch to Unix - and we'd have had an industry standard based on a culture of portable source code rather than Intel binary compatibility.
Then some 2-bit outfit called IBM chose the 8086 (technically 8088) for their staggeringly mediocre "me too" personal computer running a clone of CP/M which proceeded to take over the world because "nobody ever got fired for buying IBM" and management types preferred to deal with pinstripe suits selling machines with military-sounding acronyms rather than black turtlenecks selling computers with names from Star Trek or Lord of the Rings. Since then, backward compatibility with Intel binaries has been an albatross around the industry's neck.
Not that intel haven't, from time to time, had good chips - but they've survived their failures because all that legacy software makes them too important to fail, and they have a huge captive market to fund "brute force" solutions. (It's worth remembering that your modern x86 consists of an x86 instruction translator feeding a RISC-like core - and if ARM has a fundamental advantage it is that it can always leave off that translation hardware).
Intel
could produce a RISC chip by licensing ARM, using RISC-V... or they're big enough and ugly enough to create their own ISA... but as soon as they drop x86 compatibility then they'll find themselves competing on merit instead of backwards compatibility. Oops. Microsoft has the same issue, which might mean they're not trying
too hard with the Surface X.
What
may have changed this time is that Wintel have lost a lot of ground to mobile (Turns out that Windows and WIndows Apps
sucked on mobile devices, so the legacy card was out of play) and server-based web technologies (using open-source server software, which runs on linux, which runs on pretty much damn anything)... and while the Mac didn't win the format war with PC, it has won a respectable market share and is
very visible, so the M1 should do a lot to prove to the world that x86 isn't the only viable processor.
As for the x86 itself - of course Intel (and AMD who aren't so far behind) will play catch up. The legacy code drag-anchor and that big ugly translation unit on every chip will slow them down a bit, but if Apple wants to stay ahead, someone needs to read
The Tortoise and the Hare to Tim Cook every night before bed.
Meanwhile, we'll also get to find out what NVIDIA wants to do with ARM. They're now in a position to make ARM SoCs with NVIDIA graphics - and even the brand name is going to make that a force to be reckoned with.
arstechnica.com
