Announcement

Collapse
No announcement yet.

LLVM Patches Confirm Google Has Its Own In-House Processor

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

  • LLVM Patches Confirm Google Has Its Own In-House Processor

    Phoronix: LLVM Patches Confirm Google Has Its Own In-House Processor

    Patches published by Google developers today for LLVM/Clang confirm that the company has at least one in-house processor of its own...

    http://www.phoronix.com/scan.php?pag...i-Architecture

  • #2
    I don't think a backend like that should be merged into mainstream, as it has absolutely no use to anyone besides Google. I'd say let them do the maintaining.

    Comment


    • #3
      More and more distros are considering dropping 32-bit support, the "best" time to come out with a 32-bit processor. It may be dead on arrival.

      Comment


      • #4
        Whatever...as long as distros like gentoo, funtoo, void and alpine support 32bit. Everybody else is on the feature creep bandwagon anyway so they are pretty much garbage in my eyes.

        Most applications don't even remotely need 64bit support... or need it because they are written badly. In fact 32bit applications have a potential to run faster due less wasted instruction cache and dram bandwidth.

        Comment


        • #5
          Originally posted by eydee View Post
          More and more distros are considering dropping 32-bit support, the "best" time to come out with a 32-bit processor. It may be dead on arrival.
          I think it is highly irrelevant if distros are dropping 32-bit support or not. It's very unlikely that Google is designing this with "Yeah, let's run 32-bit len0x, w00t!" in mind.

          The only thing relevant is WHY they are doing this?

          Speed? Not likely. Quite hard to break any computational records unless you spend like a gazillion engineering hours combined with a will to sink a large chunk of money into fabs.

          Research? Not relevant. There is not much you can't do with simulating both caches, memory backends, pipeline etc. Don't know what Google think they can contribute to the CPU designing community with an simple in-order machine with does not seem to present any cases for high level research.

          Security? Quite possibly. If they don't trust the innards of a non in-house developed CPU, then possibly so. Combining an unknown architecture with an unknown codebase can be pretty tough to beat from the outside. Crashes.. sure. Exploits? Less likely so. In that case I think it is a mistake to publish the backend for LLVM. They should have kept it maintained offline.
          Last edited by milkylainen; 02-09-2016, 03:56 PM.

          Comment


          • #6
            There is a Myricom Lanai processor used in Myricom Network Adapters.

            Gnu Compiler Collection with Myricom Lanai processor support https://github.com/myri?tab=repositories

            Comment


            • #7
              No out-of-order execution, no floating point instructions, no 64-bit.
              This sounds very simple and sounds pretty useless for things like a desktop, laptop, or smartphone or such.

              Sounds more like something massively parallel explicitly designed for some cluster to do one workload.
              Maybe just do video encoding for YouTube, or break crypto or something, I don't know.

              Comment


              • #8
                Originally posted by M1kkko View Post
                I don't think a backend like that should be merged into mainstream, as it has absolutely no use to anyone besides Google. I'd say let them do the maintaining.
                Agreed. And if the point of the CPU is security and network-centric work... then this would be considered a mistake in my book.

                Comment


                • #9
                  Originally posted by log0 View Post
                  There is a Myricom Lanai processor used in Myricom Network Adapters.

                  Gnu Compiler Collection with Myricom Lanai processor support https://github.com/myri?tab=repositories
                  OK. So not a secret CPU by a longshot then. So this is just Google porting LANAI to LLVM/Clang and nothing more?

                  Comment


                  • #10
                    Originally posted by uid313 View Post
                    No out-of-order execution, no floating point instructions, no 64-bit.
                    This sounds very simple and sounds pretty useless for things like a desktop, laptop, or smartphone or such.

                    Sounds more like something massively parallel explicitly designed for some cluster to do one workload.
                    Maybe just do video encoding for YouTube, or break crypto or something, I don't know.
                    Network-centric stuff with network security would be my guess. If you had to exploit an unknown arch running unknown code... that would be a pretty tough target. Especially if the network is monitored at the same time.

                    Comment

                    Working...
                    X