Forgot your password?
Close
typodupeerror
Linux

Ubuntu Working To Provide Good Support For The VisionFive Low-Cost RISC-V Board (phoronix.com) 39

Posted by EditorDavid from the taking-a-RISC dept.
"Ubuntu developers have been working on bringing up and improving support for the Starfive VisionFive," writes Phoronix, calling the $179 device "one of the most promising 'low-cost' RISC-V single board computers to date... intended to run full-blown RISC-V Linux distributions." The board comes in two varieties with 4GB or 8GB of system memory, powered by a dual-core SiFive U74 RV64 SoC @ 1.0GHz, an NVDLA deep learning accelerator engine, a Tensilica-VP6 Vision DSP, and a neural network engine.... Considering the performance of the much more capable HiFive Unmatched development board (that is also multiple times more expensive) and even that sometimes being outpaced by the Raspberry Pi, don't get too excited for the dual-core 1.0GHz RISC-V 64-bit SoC for general purpose workloads. But this VisionFive board may be interesting for some machine learning and other use-cases thanks to its additional accelerators.

It's also one of the few RISC-V boards capable of running a full Linux distribution in the sub-$200 space.

Since the upstream Linux 5.17 kernel there has been mainline support for the VisionFive v1 board. Ubuntu developers are working on enabling the StarFive VisionFive board for their RISC-V kernel build.
This discussion has been archived. No new comments can be posted.

Ubuntu Working To Provide Good Support For The VisionFive Low-Cost RISC-V Board More

Ubuntu Working To Provide Good Support For The VisionFive Low-Cost RISC-V Board

Comments Filter:
  • I feel that RISC-V really has the power to become what ARM is today and will so in a while

    • Re: (Score:2)

      by caseih ( 160668 )

      SoCs are interesting, but I hope RISC-V goes beyond that and defines a computer platform, unlike ARM. Outside of Apple, ARM has never seen success in general computing (laptops and desktops), nor will it ever until there's some hardware standardization. ARM is a wasteland of non-standard, incompatible boot systems, hardware trees, GPU integrations, etc. although there is Linux for Arm64, there's no such thing as a generic ARM distribution that can boot off a usb stick on any ARM device. Every device is

  • Not saying its a bad thing, but I still hate the economic model. Someone "up there" made the decision that this is a good project and needs the resources. That also means those resources aren't going to other projects.

    How about asking who wants to use this RISC-V card seriously enough to donate a ten spot to the project? If enough people agree its a good idea, then the money could flow. If not, then it was a bad idea or the proposal explaining why it deserves support should be improved to attract more donors.

    (Yeah, it's the CSB idea again, but I'm still trying to figure out what's wrong with it. Also what's wrong with 37 other ideas I've been toying with for the last few decades. (Obviously excluding the ones that have been adopted and succeeded on someone else's initiative. Doesn't really matter who thought of it first. (Apparently the only thing that matters these days is who dies with the most toys.)))

    • Re:Please define "good" support and for how long (Score:5, Insightful)

      by bill_mcgonigle ( 4333 ) * on Sunday June 12, 2022 @01:54PM (#62613496) Homepage Journal

      > How about asking who wants to use this RISC-V card seriously enough to donate a ten spot to the project? If enough people agree its a good idea, then the money could flow. If not, then it was a bad idea or the proposal explaining why it deserves support should be improved to attract more donors.

      WTAF.

      Somebody at Ubuntu figured out RISC-V is important and decided this was the best board to support, for whatever matrix of reasons that went into that decision.

      This is fantastic.

      Why are you trying to apply your weird form of anarcho-syndicalist commune to Ubuntu? That's not how they are governed. If you're an owner you don't need to complain here and I am sure Debian will gladly take any support your collective wants to provide to support any other RISC board.

      This is so weird.

      Go, Ubuntu!

      • For some people, their own particular purity test matters far more than actual results.

      • Re: (Score:2)

        by shanen ( 462549 )

        NAK

      • Why are you trying to apply your weird form of anarcho-syndicalist commune to Ubuntu? That's not how they are governed. If you're an owner you don't need to complain here and I am sure Debian will gladly take any support your collective wants to provide to support any other RISC board.

        This is so weird.

        Go, Ubuntu!

        I have a simpler capitalist explanation. The makers of the board have default support for Fedora [github.com]. If little boards like these take off like the RPi, Ubuntu would be missing out big time. The alt-fruit Pi's suffer from some proprietary blobs and what not (ironically just like the Pi itself), but if the VisionFive board is truly "open" hardware, then this board should be good enough as another. Support one and you own the whole clone army.

        • Re: (Score:2)

          by tlhIngan ( 30335 )

          I do ARM, but RISC-V looks interesting and we can always have more competition in the architecture game - after all, it used to be we had x86/x64, ARM, MIPS, SPARC and PowerPC. SPARC and PowerPC are still around, but aren't common anymore (SPARC is barely around. while PowerPC seems to be limited to embedded systems for the military and maybe networking gear). Same goes with MIPS - though I don't know what's up with it.

          So we're limited to x86/x64 and ARM as common architectures. Thus having RISC-V is needed

          • I do ARM, but RISC-V looks interesting and we can always have more competition in the architecture game - after all, it used to be we had x86/x64, ARM, MIPS, SPARC and PowerPC. SPARC and PowerPC are still around, but aren't common anymore (SPARC is barely around. while PowerPC seems to be limited to embedded systems for the military and maybe networking gear). Same goes with MIPS - though I don't know what's up with it.

            I have what appears to be a MIPS-based router [openwrt.org], the TP-Link Archer C5 v1 [openwrt.org] It's good enough for low-speed Wireless AC 1200 (Wifi 5). Maybe RISC-V can be a replacement for such low-cost CPUs for non-processor intensive IoT tasks.

      • Somebody at Ubuntu figured out RISC-V is important and decided this was the best board to support, for whatever matrix of reasons that went into that decision.

        How about someone explain why RISC-V can even remotely be considered "important"? There's Intel and AMD for computing, and ARM for the mobile devices requiring power efficiency. There's even a slew of microprocessor companies putting out SoCs, which one can better understand their "importance" to the semiconductor industry. If I wasn't convinced Ubuntu was getting some sort of kickback from the French company fabbing their future designs, and them getting a kickback from the Chinese, who want their form

    • Re: (Score:3, Funny)

      by Dutch Gun ( 899105 )

      Found the Lisp programmer.

    • RISC-V is doing just fine in the embedded market. ARM is still dominating in the mid-power space (ie, cell phones), but if one of your "up there" benefactors were to decide to develop an "open source hardware" gpu to complement the cpu, perhaps with Vulkan acceleration, we could see larger system SoCs cheaper than ARM.

      • Re: (Score:2)

        by shanen ( 462549 )

        My main concern with the big-donor model is that the big donors will make the decisions and sometimes they will decide badly. Either that, or they discover their pockets aren't as large as they thought they were. (I haven't really been tracking Ubuntu, though I still use it, but I don't think Shuttleworth ever made much money on that side. I mostly credit him as a competent manager, and he obviously hasn't made any fatal decisions for Ubuntu, though some of the Japanese-related decisions were quite painful.

  • Vs ARM? (Score:3)

    by itzdandy ( 183397 ) on Sunday June 12, 2022 @01:33PM (#62613454) Homepage

    I would love to see RISC-V get a major backer in a high volume product. The problem as I see it is that ARM is a really smart company and they have priced licensing at just the right spot to make RISC-V look like an expensive arch to invest in even in the long term. RISC-V is *WAY* behind and would need at least a few years of development and refining to get up to the common ARM CPUs of today.

    I love my open source, but open source hardware has never fared well. It's hard to get top talent to spend the kind of time needed to make an Apple M1/M2 or Qualcomm 8cx gen3 CPU that can compete on performance/watt or /price etc to make it a viable option for a desktop computer. This is one of those 'needs a consortium' type tasks IMO.

    Frankly, I think for desktop use a RISC-V chip has to be basically at an Apple M1 level. That's something that only Apple has figured out. Qualcomm's 8cx gen3 is close, but if you've put hands on a Thinkpad x13s (with that chip) and compared with a Macbook Pro M1 they're not quite there.

    BIG ask to get RISC-V up to this level in the next few years IMO.

    • I just want an ARM or RISC board that uses hardware x265 decoding without Kodi. I hate the dummy interface of Kodi and Librelec. It’s near impossible to find a non x64 board that plays video with a standard Linux distro.

      • Raspberry Pi4 decodes HEVC10bit in VLC just fine, even stupidly high bitrate encodes. You can also get mplayer / MPV to play them with hardware decoding, but you have to compile FFMPEG with the HW decoder support, which can be finicky to get compiled. The only caveat with VLC is you lose the OSD in fullscreen.

        Well, unless they started shipping with the patched up FFMPEG. Then it should just work out of the box.

    • > RISC-V is *WAY* behind and would need at least a few years of development and refining to get up to the common ARM CPUs of today

      Maybe Ubuntu should support them to help build a groundswell of support and get that party started?

      Regardless, this board looks great for several applications and for those applications where trust is more important than maximum MIPS per Watt, ARM's current optimization is irrelevant.

      • Well, I can't completely argue that Ubuntu's support is interesting, but a dual core 1Ghz machine that is more or less early Pentium4 era performance isn't going to draw a lot of users in for 'desktop' use. maybe if there are other cores in the box like h.26x encode/decode etc then it could be interesting in that respect.

    • Comment removed based on user account deletion

    • Re: (Score:2)

      by Misagon ( 1135 )

      Everyone knows why Apple's SoCs are powerful. It is not only because the cores are efficient and wide.
      It is because Apple has taken technical choices that sacrifice flexibility for performance, and most customers who build systems with SoCs want the flexibility to build the systems they want.

      • 'flexibility for performance' as in RAM and primary storage integrated sure, but I'm putting all sorts of workloads through these and I've yet to find this flexibility trade off in compute. I run an Alienware m15r6 (11th gen intel i7 + nvidia 3070) in my hardware mix and it does exactly one thing better than my Macbook M1 Pro, that's play games. It's (Alienware) faster than my Mini M1 but not by very much and my MBP 16 M1 Pro feels faster. Heck, the M1 Pro feels faster running Windows 11 Arm in Parallels

    • Re: (Score:2)

      by AmiMoJo ( 196126 )

      The M1 is more a problem of vertical integration than of it being hard to develop a chip that runs that fast. It's the fact that it has massive caches and on-substrate memory that is tightly coupled that gives it an edge over competing ARM designs. Anyone else who wants to copy that will need to get very close to the manufacturers who integrate the SoC. It all has to be agreed down to what SKUs they produce, because the manufacturer can't choose how much RAM it comes with, for example.

      RISC-V might have some

      • I imagine RAM being another less interesting component of systems in the future. It's reasonably cheap and just getting cheaper. 99%, probably higher, are perfectly happy with 8GB of RAM and can't really notice any improvement above 16GB. I'm pretty sure that's where Apple is coming from.

        Indefinitely see products like RISC-V filling embedded roles. A dual core 1Ghz CPU is actually quite a lot for 'general compute' for a lot of embedded products and RISC-V's claim to fame here is really that it's easy to

  • because rpis are hard to come by [rpilocator.com] right now

Slashdot Top Deals

"Stupidity, like virtue, is its own reward" -- William E. Davidsen

Close