SiFive HiFive Premier P550 RISC-V Price Lowered, Ubuntu 24.04 Support Ready

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts
  • treba
    Senior Member
    • Sep 2012
    • 716

    #11
    The datasheet mentions a hw video en- and decoder, but with very vague details - e.g. even the supported codecs are missing. Does anyone know more - and has anyone spotted kernel patches for drivers? Will it be a v4l2, and if so stateful or stateless? Or maybe something using the drm subsystem - with Vulkan or VA-API userspace drivers?

    Comment

    • coder
      Senior Member
      • Nov 2014
      • 8952

      #12
      Originally posted by treba View Post
      The datasheet mentions a hw video en- and decoder, but with very vague details - e.g. even the supported codecs are missing. Does anyone know more - and has anyone spotted kernel patches for drivers? Will it be a v4l2, and if so stateful or stateless? Or maybe something using the drm subsystem - with Vulkan or VA-API userspace drivers?
      If you want to use one any time soon, I'd suggest getting a RX 6400 dGPU for it.

      Comment

      • varikonniemi
        Senior Member
        • Jan 2012
        • 1102

        #13
        Originally posted by coder View Post
        It was not insane. The only reason you thought that is because you compared it to mass market SBCs, like the Raspberry Pi. You clearly don't understand the economics of mass market electronics.


        My guess is that they simply had an order volume large enough that they could afford to offer it at a lower price point.
        your guess is shit and trying to justify the insane initial cost. At some point price gougers realize that the initial estimate of how much can be milked was too much, and here they autocorrected

        Comment

        • coder
          Senior Member
          • Nov 2014
          • 8952

          #14
          Originally posted by varikonniemi View Post
          your guess is shit and trying to justify the insane initial cost. At some point price gougers realize that the initial estimate of how much can be milked was too much, and here they autocorrected
          And your evidence for that is?

          Just because you feel victimized because it's not selling at the price to which you think you're entitled doesn't make them predators. Posts like yours undermine my hope for humanity.

          I'm actually glad you think it's too expensive. Otherwise, I expect we also have to hear you whinge about how slow it is and what greedy bastards they are for including only four cores.
          Last edited by coder; 11 December 2024, 05:36 PM.

          Comment

          • NateHubbard
            Senior Member
            • Mar 2015
            • 588

            #15
            Originally posted by varikonniemi View Post

            your guess is shit
            There's the guy that's so much fun at parties.

            Comment

            • AndyChow
              Senior Member
              • Apr 2012
              • 771

              #16
              Unless it can run the mainline kernel, I'm not interested. Patched kernels that can't be updated make those platforms sort of jokes, and not the good kind.

              Comment

              • ayumu
                Senior Member
                • Oct 2008
                • 668

                #17
                Originally posted by AndyChow View Post
                Unless it can run the mainline kernel, I'm not interested. Patched kernels that can't be updated make those platforms sort of jokes, and not the good kind.
                This is at the level of slander, considering each and every generation of their development boards has upstreamed support.

                Comment

                • Shnatsel
                  Senior Member
                  • Jan 2012
                  • 143

                  #18
                  I managed to scavenge a few published benchmarks from the internet to get a sense for the performance.

                  Geekbench shows it significantly behind Raspberry Pi 4: https://browser.geekbench.com/v6/cpu...seline=8686267

                  For a native compilation workload (which is a big reason to get RISC-V hardware) it's ~25% faster than RPi 4: https://rwmj.wordpress.com/2024/11/1...60-and-others/

                  There's also a Milk-V board with the same CPU for only $200, that is for half the price, but it's also sold out right now: https://milkv.io/megrez
                  Last edited by Shnatsel; 12 December 2024, 02:06 AM.

                  Comment

                  • brucehoult
                    Phoronix Member
                    • Aug 2019
                    • 99

                    #19
                    Originally posted by Shnatsel View Post
                    I managed to scavenge a few published benchmarks from the internet to get a sense for the performance.
                    Geekbench shows it significantly behind Raspberry Pi 4: https://browser.geekbench.com/v6/cpu...seline=8686267
                    That's because most Geekbench tests use Vector/SIMD processing in some way. The P550 doesn't have any form of SIMD, so it's kind of silly to even run (scalar versions of) such tests. OF COURSE it will be slow. That's not useful information.

                    And the RISC-V version of Geekbench doesn't use the Vector extension even on CPUs that implement it e.g. The Banana Pi BPI-F3, Milk-V Jupiter, DC-Roma II laptop etc.

                    The Pi 4 has been around for 5 1/2 years and all the software for it caught up around 2021 or so.

                    For a native compilation workload (which is a big reason to get RISC-V hardware) it's ~25% faster than RPi 4
                    Yes. And fortunately that is EXACTLY what most people are using these boards for. Developing software so that it is ready when the fast boards arrive.

                    There's also a Milk-V board with the same CPU for only $200, that is for half the price, but it's also sold out right now:
                    I successfully ordered one on December 8th.

                    "It's too expensive"

                    "it's always sold out"
                    Both of those can't be true at the same time.

                    Comment

                    • SteamPunker
                      Senior Member
                      • Nov 2014
                      • 204

                      #20
                      Originally posted by brucehoult View Post
                      That's because most Geekbench tests use Vector/SIMD processing in some way. The P550 doesn't have any form of SIMD, so it's kind of silly to even run (scalar versions of) such tests. OF COURSE it will be slow. That's not useful information.
                      That's also my concern about this P550 platform.

                      Whether you're adventurous enough to start using a RISC-V platform as a daily driver, or would like to use it for furthering development of the RISC-V software ecosystem, at this point surely you'd want to go for a CPU that supports the recently adopted RVA23 Profile? That would make for a future-proof RISC-V platform, ensuring access to standardized vector and hypervisor extensions. The P550 apparently has neither.

                      It kind of sucks, since otherwise, this would indeed seem like a very affordable basis for a decent RISC-V development workstation.

                      Maybe this is part of the reason why SiFive is is discounting these boards, to clear inventory for a next-gen product that does support the RVA23 profile.

                      The downside of these boards being discounted and becoming popular is that this way, they might create the same problem as with the first version of the Raspberry Pi, which was still based on ARMv6, resulting in that older ARM platform becoming the "lowest common denominator" for 32-bit Raspberry Pi operating systems and software packages, even though from the Raspberry Pi 2 onwards, ARMv7 was adopted, including NEON, ARM's SIMD instructions.

                      If the P550 takes off too much, then common RISC-V distros and software packages might end up being compiled without support for any vector extensions, even though later RISC-V systems will all have support for them. And this time, we won't have a 32-bit/64-bit transition opportunity, which would allow us to start from a clean slate with determining a "base profile" for compatibility.

                      Comment

                      Working...
                      X