Announcement

Collapse
No announcement yet.

Ampere Is Designing Their Own Arm Server CPU Cores, Coming In 2022

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #11
    Originally posted by unic0rn View Post

    x86 is RISC under the hood since decades. compatibility layer is relatively small part of the die.
    will x86 survive soon ?

    microsoft wants to make an ARM server.
    apple wants to make an ARM server.
    nvidia says you don't need x86 cpu because you use our DPUs in the ocp. They also own arm cpus.
    intel will make x86 cpu and then who will sell it? what if microsoft says we just do our products for our ARM server and that's it.
    apple makes their os for their ARM server.
    you also have thunderx3 arm and fujitsu arm.
    amazon has it's own arm cpu.
    google will probably copy and make their own arm cpu?

    will intel drive linux x86 alone ?





    Comment


    • #12
      Originally posted by ferry View Post

      That doesn't make sense. There is a reason we're talking about servers here. And Linux servers running open source don't need a translation layer. They want a high performance/Watt because electricity costs are dominant.

      You likely want to run closed source games on your laptop (my advice get a tower with a decent video card and dual boot). There is no fundamental reason why ARM ISA would save power compared to x86 ISA (at the same performance).
      Agreed that Linux as-is, is not terribly dependent on any given ISA. Even for games, bespoke engines are few and far between, meaning the porting effort for many game would not be relatively minor should the underlying engine be ported.

      ARM though does have a front-end advantage and willingness to break binary compatibility every now and again. But the difference in decode gives ARM fundamental advantage. Intel moved mountains to get a 5-wide decode and 6 might be doable if they can ether get to their 7nm process. Meanwhile the M1 is 8-wide. on what will be a comparable process.

      The whole Dozen different SIMD extensions don't help x86 any either, meanwhile arm v.9 will likely only include Neon and SVE2.

      The other disadvantage for x86 is that there are only 2 and a half players, while ARM ISA licenses are available for purchase by anyone with the cash to do so.
      Last edited by WorBlux; 19 May 2021, 03:38 PM. Reason: small spelling fixes

      Comment


      • #13
        Originally posted by Mitch View Post
        One thing I'd love to see in Linux is some kind of X86 translation similar to what Apple has done, or even Microsoft if push comes to shove. I imagine these ARM processors are quite beefy and while Linux is great with ARM, a decent translation layer could close some gaps. When ARM chips have good Vulkan support, you can then toss Zink in and cross some big boundaries.
        Are you not aware of binfmt_misc? I know it's not a compatibility layer but it opens doors to seamlessly running x86 software on ARM.

        Comment


        • #14
          Originally posted by ferry View Post

          That doesn't make sense. There is a reason we're talking about servers here. And Linux servers running open source don't need a translation layer. They want a high performance/Watt because electricity costs are dominant.

          You likely want to run closed source games on your laptop (my advice get a tower with a decent video card and dual boot). There is no fundamental reason why ARM ISA would save power compared to x86 ISA (at the same performance).
          Maybe I'm mixing up non-related things. I was thinking servers as well as personal computers. So Qualcomm laptops too. If you could get performance, efficiency, and compatibility parity with Intel and AMD (like Apple M1), I imagine you'd be able to have competition between more CPU vendors in the server and cloud world. Maybe personal computers as well.

          Comment


          • #15
            Originally posted by schmidtbag View Post
            Are you not aware of binfmt_misc? I know it's not a compatibility layer but it opens doors to seamlessly running x86 software on ARM.
            Opening the door is one thing, getting your new fancy sofa inside and up the stairs is quite another.

            For historical reason general purpose desktop hardware need to support some degree of x86 binary for some fairly low-level stuff (certain printer peripheral drivers for example) . But because ARM and x86 have slightly incompatible memory models, near-native performance requires a degree of hardware support for a translation mode.

            Comment


            • #16
              Originally posted by Mitch View Post
              One thing I'd love to see in Linux is some kind of X86 translation similar to what Apple has done, or even Microsoft if push comes to shove. I imagine these ARM processors are quite beefy and while Linux is great with ARM, a decent translation layer could close some gaps. When ARM chips have good Vulkan support, you can then toss Zink in and cross some big boundaries.
              I would not worry too much. Nvidia will make running Linux (what else for high-performance servers?) on their CPU a piece of cake. I don't see what can possibly go wrong.

              Comment


              • #17
                Originally posted by Mitch View Post

                Maybe I'm mixing up non-related things. I was thinking servers as well as personal computers. So Qualcomm laptops too. If you could get performance, efficiency, and compatibility parity with Intel and AMD (like Apple M1), I imagine you'd be able to have competition between more CPU vendors in the server and cloud world. Maybe personal computers as well.
                Windows runs on Arm and has a translation layer that runs x86 and x86-64 applications. There is no point in doing this for Linux since it is used mostly in servers (plus Android mobile phones/tablets) where everything is compiled natively.

                Comment


                • #18
                  Ampere, if you're listening... partnering with another outfit to make a stripped-down 8-core version on some standard form factor with PCIe (mini-itx?) and UEFI would help drive adoption. Just being able to rack one up as a DEV box or get these into nerds' hands will drive adoption in ways that aren't immediately apparent.

                  Comment


                  • #19
                    Originally posted by gigi View Post
                    does it mean they are directly competing with amd zen4?
                    it's unclear how do you imagine competition based on process node. do you think any server customer goes like this "system x uses 6nm process, system y uses 5nm process, obviously 5nm is better, we should select it"? and then "oh, we have two 5nm alternatives, now it's time to compare price and performance"
                    Last edited by pal666; 19 May 2021, 09:36 PM.

                    Comment


                    • #20
                      Originally posted by Mitch View Post
                      One thing I'd love to see in Linux is some kind of X86 translation similar to what Apple has done, or even Microsoft if push comes to shove.
                      i'd love to see real hardware x86 instruction decoder, let's say 10% slower than native solution and taking additional 10% transistor budget. then i can switch to other isa cpu without severely handicapping my game collection. it requires x86 license, but several companies have one
                      Last edited by pal666; 19 May 2021, 09:18 PM.

                      Comment

                      Working...
                      X