back to article Intel, AMD team with tech titans for x86 ISA overhaul

The shape of the x86 instruction set architecture (ISA) is evolving. On Tuesday, Intel and AMD announced the formation of an ecosystem advisory group intended to drive greater consistency between the brands' x86 implementations. Intel and AMD have been co-developing the x86-64 instruction for decades. But while end user …

  1. abend0c4 Silver badge

    Don't cast me in the role of Cassandra, but...

    ... the usual point in history in which bitter rivals agree to cooperate is the moment they realise the jig is up.

    1. Baird34

      Re: Don't cast me in the role of Cassandra, but...

      Hi Cassie.

      1. abend0c4 Silver badge

        Re: Don't cast me in the role of Cassandra, but...

        I prefer Sandra... Oh, what a giveaway!

        1. MyffyW Silver badge

          Re: Don't cast me in the role of Cassandra, but...

          It's ok, if we follow the classical example your powers of prophecy, however correct, will be ignored.

          Alternatively (and I quote one of my favourite NZ actresses)

          Sandy!

          Tell me about it ... stud

    2. Groo The Wanderer

      Re: Don't cast me in the role of Cassandra, but...

      No doubt the encroachment of Arm on their server chip markets is a big factor, but I think RISCV is the greater long-term threat to their business.

    3. Zoopy

      Re: Don't cast me in the role of Cassandra, but...

      If the new ISA name includes the word "Open", it's time to start a new area in the local computer museum.

  2. Anonymous Coward
    Anonymous Coward

    Progress!

    Going 64-bit only (dropping 16-bit and 32-bit legacy silicon) as discussed in the "x86S spec" link, and as mulled last year, makes good sense to me, especially now that those chips and systems aren't booting through some 16-bit PC BIOS anymore but through UEFI instead, straight into 64-bit operation.

    ARM had the same idea 4 years ago with all new Cortex-A in its Tocal Compute Solutions (TCS) being 64-bit only from last year (approx.), like the Cortex-X925, Cortex-A725, and Cortex-A520 (AFAIK).

    8-bit, 16-bit, and 32-bit chips might still find uses in the tiniest, or least expensive, of computational devices though, just not as main CPUs in phones, tablets, laptops, desktops, workstations, servers, etc ..., IMHO.

    Hopefully they also start considering 128-bit addressing in their reshaped ISAs, as needed for the Zettascale (and beyond)!

    1. Someone Else Silver badge

      Re: Progress!

      Hopefully they also start considering 128-bit addressing in their reshaped ISAs, as needed for the Zettascale (and beyond)!

      So the return of the Intel segment registers, then? Oh-comma-goodie!

      1. Herring` Silver badge

        Re: Progress!

        I miss far pointers

        1. Someone Else Silver badge

          Re: Progress!

          No, you don't Not really. Nostalgia for them, maybe, but you really, really don't.

          1. Herring` Silver badge

            Re: Progress!

            I also miss zero page addressing on the 6502. Those Z80 folks going on about how many registers they had when effectively the 6502 could use those 256 bytes at speed.

            I don't miss 16-bit windows where using more than one data segment was a pain in the arse.

            1. CowHorseFrog Silver badge

              Re: Progress!

              Zero page and similar concepts dont really work in the modern world. Back in those days memory access was fast enough, and the cpu was the slow part so ther ewas little advanage in having lots of registers.

              We can see basically all cpus from that time had the same philospophy.

            2. CowHorseFrog Silver badge

              Re: Progress!

              Zero page is effectively a data register (even if its implied to be the zero page).

    2. Anonymous Coward
      Anonymous Coward

      Re: Progress!

      Aren't 16 and 32 bit modes implemented in microcode these days, using barely any silicon?

      1. Groo The Wanderer

        Re: Progress!

        Not entirely - you still need a level of hardware support. But I predict a huge outcry from the gaming community as their favourite older games wouldn't be too likely to run on such stripped processors.

        Well maybe not a huge outcry, but definitely some whimpering and whining.

        1. dharmOS

          Re: Progress! 16 and 32-bit x86 can be emulated on a potato

          I am certain that even demanding 32-bit games of yesteryear could be emulated on the WOW32 built into Windows emulated on the 64-bit processor. For everything else, there is 86Box, DOSBox-X and its ilk.

          I take great delight in running Wing Commander 3 (which only ran on the fastest Pentiums of its 90s era, using 16-bit x86 code and MS DOS) on a Snapdragon ARM PC using DOSBox.

          That’s cross CPU arch and no native 16-bit x86 on my WinARM PC. Runs faster than ever.

        2. Spazturtle Silver badge

          Re: Progress!

          "But I predict a huge outcry from the gaming community as their favourite older games wouldn't be too likely to run on such stripped processors."

          This will not affect 32bit programs in anyway, 32bit programs already run on the 64bit part of the CPU.

          Currently the CPU boots to the 16bit mode, does initialisation, then boots to the 32bit mode, shuts down the 16bit mode, then boots to the 64bit mode, then shuts down the 32bit mode.

          If you are running a 64bit OS then the 16bit and 32bit parts of the CPU are already shut down.

          16bit mode supports 16bit.

          32bit mode supports 16bit and 32bit.

          64bit mode supports 32bit and 64bit.

    3. Anonymous Coward
      Anonymous Coward

      Re: Progress!

      There are multiple layers of stuff between the transistors and the ISA. So backward compatibility isn’t a problem (and that’s before the OS emulates stuff in code)

    4. martinusher Silver badge

      Re: Progress!

      One tiny snag with constantly expanding data and address buses is that they consume interface pins. Lots of them. These need routing and drivers on the silicon and routing off the silicon on whatever the part's attached to. If the memory can be locally attached -- integrated -- like a microcontroller then this will ease the problem somewhat but in general its a design, manufacturing and test headache.

      For many products this is justified, if for no other reason than apparent competitiveness ("they have it"). But for a lot of applications this is wasteful, consuming lots of real estate, power and other resources.

      (It would be a lot more useful to figure out how to write more efficient code than constantly expanding the resources needed to run it.)

  3. cjcox

    Yep, pure fear...

    Btw, this is 99.9999% beneficial to Intel only. AMD, as usual, is the sucker/loser that Intel carries along because of anti-trust.

    I might think differently if major computer manufacturers had even some parity with regards to Intel vs. AMD offerings. But things are still very skewed towards Intel.

    Again, it's just a sham show from Intel. Typical.

    I sort of hope that ARM eats Intel's lunch. Sorry AMD, but you're still a sucker/loser, but mainly because you're willing.

    1. RobHeffo
      WTF?

      Re: Yep, pure fear...

      You sir, need to step back and put down the keyboard for a while.

      AMD is the reason we have x86-64 (AMD64), without it Intel would still be shilling that IA64 platform that was so horrible.

      The ZEN processors have been fantastic and have had Intel sitting up taking notice for a while now.

  4. mattaw2001
    FAIL

    Why lead with AMDs consistent support for AVX 512 vs Intel's patchy madness?!

    Funny the article leads with AMDs 98% consistent support, double or single pumped Zen4 onwards runs the same machine code.

    I would lead with the hell of exactly which pieces of AVX an Intel core supports at any given time -it a mystery, ESPECIALLY WHEN THEY MIX SUPPORT IN THE SAME CHIP.

    Heck the Wikipedia article https://en.m.wikipedia.org/wiki/Advanced_Vector_Extensions is full of fun weirdness, e.g.

    'Intel does not officially support AVX-512 family of instructions on the Alder Lake microprocessors. In early 2022, Intel began disabling in silicon (fusing off) AVX-512 in Alder Lake microprocessors to prevent customers from enabling AVX-512.[29] In older Alder Lake family CPUs with some legacy combinations of BIOS and microcode revisions, it was possible to execute AVX-512 family instructions when disabling all the efficiency cores which do not contain the silicon for AVX-512.'

  5. Torben Mogensen

    Merger?

    In the past AMD and Intel were not allowed to merge due to monopoly concerns. But you can argue that x86 is no longer a monopoly, as ARM powers an increasing fraction of computers (and RISC-V is slowly rising too). So I don't see it as impossible that the two merge. Still selling under their original brands, but sharing more technology and making common business decisions to differentiate the brands more by targeting different segments.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like