AMD, Intel, and a slew of tech companies are teaming up to fend off ARM chips

https://www.theverge.com/2024/10/15/24271080/amd-intel-x86-advisory-group-lenovo-arm

16 Comments

  1. x86 hit a wall a long time ago. Does it make sense to keep pushing vendors to stay on these chips with old architecture, or try something new?

  2. Personally, I think that the bigger but longer term threat is RISC-V. It is really hard to beat “no royalties”.

  3. Since half my apps don’t work on arm processors when I tried to use the new surface laptop I hope they team up to fix it instead of run away.

  4. and hopefully revise the ATX specification to add 24v or 48v … right?
    Maybe make a spec for routing the plugs to the back of the motherboard as well…. right?

  5. It’s been reported that Apple literally begged Intel to do more in the lower power space, but Intel’s emphisis 18 years ago was all-in on performance. The emerging cellphone/tablet/watch world went for what could meet low power and that was ARM.

    Having used ARM on iPhone, Apple appears to have realized that it’s easy to make ARM faster than x86 lower power.

    x86-64 has compact instructions (which could increase speed based on instructions-per-byte fetched from RAM), but in typical code, many instructions are wasted shuffling around the 16 register/two-operand limitation of the x86-64 instruction set. Rename registers really shine in 3-operand instruction sets. So code has to use the stack for scratch space more often. No matter how hard the CPU works to hide that overhead, the cache still has to be updated and communicated to other cores.

    The variable length instructions make branch prediction harder (though this stuff is often cracked in the cache, so it will often be pretty fast). But still, the hardware needs to be there to support these weird instruction lengths.

    Almost any trick to make x86 lower power will yield even bigger results in ARM.

    This is going to be tough for Intel to do. I suspect they will shed design wins, from the bottom up. ARM is already way cheaper and way lower power at the low end. As ARM gets better at raw performance, ARM will work it’s way up the stack taking more and more design wins. x86-64 isn’t dead, but it’s not going to be the massive market leader it once was.

  6. Ummm….a little late, don’t you think? Should’ve seen this coming about 15-20 years ago

  7. I’ve bought a few of the new Snapdragon Elite laptops, and the battery life is amazing and every day operating speed is on par with AMD and Intel. Plus it’s cheaper. They should be scared.

  8. Yes, fight the good fight. I teach x86 assembly and don’t want to have to build a whole new course based on RISC. That sounds like a lot of work.

    Jokes aside, the efficiency of ARM is awesome but there’s so much old x86 software out there that I think compatibility is going to be a massive problem for a very long time if people want to fully transition to ARM machines.

  9. DollarsAtStarNumber on

    Ah yes, an Alliance of

    AMD

    Intel

    Microsoft

    Or AIM for short.

    I feel like we’ve been down this hole before.

  10. Guess this means I’m not gonna see a “Ryzen 10800ARM” or something like that in the future, huh? I’d really hope the big giants would take this kind of tech and run with it, not try to bury it. Energy efficient Arm chips in desktops and laptops alike are what we’re gonna need to get away from these crappy 2-hour-battery portables and desktops that need liquid loops to run cool at the high end of the scale.