Announcement

Collapse
No announcement yet.

The OpenPOWER ISA EULA Draft Published - Generous For Libre Hardware

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

  • The OpenPOWER ISA EULA Draft Published - Generous For Libre Hardware

    Phoronix: The OpenPOWER ISA EULA Draft Published - Generous For Libre Hardware

    Last summer it was announced that IBM's POWER ISA would be open-source and the OpenPOWER Foundation joining the Linux Foundation. Finally we're getting a look at how the end-user license agreement (EULA) is looking for those wishing to make use of the POWER CPU instruction set architecture...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    This is good news, in fact Power was the basis of xbox360 and ps3 cpu. riscv is going to get solid foundation.

    Comment


    • #3
      After a microcontroller proof-of-concept, they should start with an open SBC with no GPU/VPU, I would want hardware-accelerated crypto, GbE though. Use the profits to work on video for Phase 3.

      Comment


      • #4
        Originally posted by elatllat View Post
        After a microcontroller proof-of-concept...
        I don't think OpenPOWER is really suitable for microcontrollers. You'd be better off with RISC-V in this regard (simpler implementation & more compact machine code).

        Comment


        • #5
          I'm just betting the first thing they make will have barely enough to run Linux at which point I call it on microcontroller. V2 could be a mini server, v3 a mini desktop.

          Comment


          • #6
            This comes probably too late, RISC-V has taken off already. Had they become more liberal with their licensing 5 years ago, things might have turned out differently.
            (And btw.: PA-RISC would have been another great candidate, had they not unnecessarily sold out to Intel. It was a great and indeed very much precision engineered instruction set architecture. Not the VLIW-crap Intel turned it into.)

            Comment


            • #7
              Originally posted by dwagner View Post
              This comes probably too late, RISC-V has taken off already. Had they become more liberal with their licensing 5 years ago, things might have turned out differently.
              (And btw.: PA-RISC would have been another great candidate, had they not unnecessarily sold out to Intel. It was a great and indeed very much precision engineered instruction set architecture. Not the VLIW-crap Intel turned it into.)
              Bull. RISC-V hasn't gone anywhere. There's a few companies that are getting their toes wet and that's about it. It's not used in any consequential products.

              POWER on the other hand is used in real world systems in SMP servers, mainframes, supercomputers, and workstations and these systems exist right now. You can buy fully open workstations based on POWER9 right now from Raptor Computing. Where are the fully functional RISC-V desktops? Laptops? Servers? Phones? Oh right... there aren't any.

              Comment


              • #8
                I'd say RISC-V is still in its infancy, e.g. their vector extension is still not ratified (and still just at v0.8). And at the current speed it will still need at least another two years down the road. Also the software infrastracture is a very early WIP. OpenPOWER on the other hand still needs more momentum to gain more mindshare and traction, software wise it is also way behind ARM in terms of ISA optimizations. POWER could have been the ISA of choice for the European Processor Initiative if they had opened up their ISA two years earlier and it is late, but not too late. The Libre project is a good example that people do have a look on the POWER ISA now who originally considered the RISC-V route and it could make sense if the timeline and projects suits that ISA better.
                Last edited by ms178; 15 February 2020, 01:01 PM.

                Comment


                • #9
                  Originally posted by stormcrow View Post

                  Bull. RISC-V hasn't gone anywhere. There's a few companies that are getting their toes wet and that's about it. It's not used in any consequential products.

                  POWER on the other hand is used in real world systems in SMP servers, mainframes, supercomputers, and workstations and these systems exist right now. You can buy fully open workstations based on POWER9 right now from Raptor Computing. Where are the fully functional RISC-V desktops? Laptops? Servers? Phones? Oh right... there aren't any.
                  That's not really accurate. RISC-V ships today as the Falcon co-processor on all Nvidia GPUs, and as the microcontroller on Western Digital HDDs as confirmed products, and while I can't confirm it the RISC-V foundation at least throws around the idea that it's heavily used in In-Car-Infotainment and IoT devices. Regardless that's literally millions of units shipping with RISC-V CPUs within them just between Nvidia and WD, many millions more if the RISC-V foundation's website is accurate. It's true you can't really buy any of the products you've listed, but RISC-V seems to be well on its way in Microcontroller and embedded roles.

                  Comment


                  • #10
                    This is good for Power! On the other hand why would the RISC-V group be considering power for a graphics engine?? That makes no sense at all to me. Other attempts to take a similar path (Intel) didn’t go well. If they did go this route RISC-V’s GPU would be a better processor than the system processor.

                    by the way good for Power doesn’t mean an initial success as open hardware. People and companies will not instantly consider the difficulty in getting working silicon worth the effort. In a way you end up with the same problem the RISC-V group has. That is you really need working silicon to get mindshare. A couple of embedded design ins will not create a viable ecosystem.

                    ARM is a great example here, it took a very long time for them to get to the position they are in right now. That only happened after a few chip designers decided to come out with fully functional chips for the general market. In a nut shell you have success when you can go to DigiKey or another vendor and buy a solution.

                    Comment

                    Working...
                    X