Ubuntu Kylin 20.04 OS works on RISC-V hardware

China-developed Ubuntu Kylin 20.04 is now supporting RISC-V architecture with an image for HiFive Unmatched mini-ITX motherboard, and work will be done on an unnamed Starfive SBC that should be the VisionFive board with a GPUless JH7100 dual-core RISC-V SoC or an upgraded version with JH7110 SoC featuring an Imagination IMG BXE-4-32 GPU.

You may have read recent reports about China asking government entities, including state-owned enterprises (SOE), to replace foreign hardware and software within a two-year period. So that means avoiding systems based on Intel and AMD processors, so working on RISC-V open architecture makes perfect sense, since over time, Chinese manufacturers should be able to make RISC-V SoCs and PCs based on those processors, albeit probably not within the next two years at any significant scale.

Ubuntu Kylin 20.04 RISC-V, as well as the newly released Ubuntu Kylin 22.04 x86, can be found on the English download page on the Ubuntu Kylin website. The RISC-V image is now only available for the HiFive Unmatched mini-ITX board with SiFive FU740 quad-core 64-bit U74 RISC-V processor, 16GB RAM, NVMe storage, and supporting discrete graphics cards through a PCIe slot. They did not do that from scratch since Ubuntu RISC-V images for HiFive boards have been available since last year.

Some installation instructions can be found in a PDF in English. The announcement explains there’s still much work to be done to use RISC-V as an x86 replacement, and there are currently over 20 self-developed packages. The image relies on a Linux 5.11 kernel and features the Ephiphany 3.36 browser and Calligra 3.10 office suite, as well as UKUI 3.0 desktop environment.

Initial Ubuntu Kylin support for the StarFive SBC will be added as part of the “Summer of Open Source (OSPP)” activity for college students with the project description, translated from Chinese to English, reading:

This project plans to make a RISC-V version of the Ubuntu Kylin system image based on the StarFive development board, and complete a certain degree of startup on the basis of the basic image. Optimization and adaptation of power management.

So there are not putting huge resources behind this development, but at least somebody will be working on it between July 1st and September 30th.

Via The Register

Share this:

Support CNX Software! Donate via cryptocurrencies or become a Patron on Patreon

10 Replies to “Ubuntu Kylin 20.04 OS works on RISC-V hardware”

  1. Is this why Chinese Rockchip designed the RK3588, not RISC V but desktop capabilities 🤔

  2. It’s a pity they didn’t choose to use UBPorts’ Ubuntu Touch – a lightweight morphing OS suitable for touchscreens as well as keyboard and mouse.

    1. Seriously ? We’ve yet to see any *real* progress on this front. Even the recently announced VisionFive was advertised as a *dual-core* 1.5 GHz and ended up being at 1 GHz and sold with a *heatsink+fan*! Even the Snowball that I got 12 years ago had an out-of-order dual-core A9 at 1 GHz which was totally cold to the touch, and I have not used that board for about that long because it was already considered outdated by then!

      For several years we’ve been told “look, we’re coming soon, next year you will see real hardware with 1990-era performance but we’re catching up, you’ll see”. Still not seen anything real that advertised better than the level of performance of an Allwinner A20 (dual cortex A7) and the consumption of a 486DX-50.

      The problem is always a chicken and egg one: it takes a lot of time and experience to design serious chips that can match competition on the performance and efficiency levels. By the time you release them, competition has progressed a lot and nobody’s interested in your antique chips. There’s always a niche market for such chips, with those who fear that the chip vendor has placed backdoors, GPS trackers, and 5G in their CPUs. But such supporters tend to stay away from the big companies that are needed to mass-product the chips, so that’s not easy.

      Where RISC-V can succeed is the MCU market. A wide range of chips are neither interesting for their performance nor their power efficiency because they’re used in dishwashers, fridges, microwave ovens etc. What matters there is solely the BOM cost. Being able to lower it thanks to chips designed with little or no license is appealing, given that PIC, AVR and STM32 likely cost way more in licensing than silicon. And I do really hope that we’ll see more of them there.

      1. RISC-V has been succeeding in the in the microcontroller market for a while now, to the extent that Arm have had to change their licensing approach. This will only continue, since MCUs are low level fruit in design terms.

        The next sector for competition won’t be APU, it will be higher spec custom processing : AI, ML, Edge, Auto, where RISC-V can go head-to-head on new silicon design. We’re already seeing seeing silicon for that too – notably Esperanto’s chip, which is being favourably received at the moment.

        Consumer electronics is a more complicated issue, since a wealth of apps are needed too. Work is ongoing with that. There’s promise of a laptop, and two phones this year, or just into next. Even if they’re just developer editions : that’s progress.

        Personally – I don’t think competition is the factor : we’ve been the victims of “buy the latest / fastest” for so long now, just to keep OEMs in business. For years I was happy enough with a Nexus 5 (running UT) and a 2006 MacBook (running Ubuntu)… Anything more than Sufficiency seems like a waste ( – I realise that that’s a minority opinion.)

        1. > Personally – I don’t think competition is the factor : we’ve been the victims of “buy the latest / fastest” for so long now, just to keep OEMs in business. For years I was happy enough with a Nexus 5 (running UT) and a 2006 MacBook (running Ubuntu)… Anything more than Sufficiency seems like a waste ( – I realise that that’s a minority opinion.)

          I totally agree on this. The laptop I’m using at home to read my emails and cnx while drinking my morning coffee as an old eeepc 1005HA with a dual-core Atom N2600. It’s 12 years old now and sufficient for my usage. Even firefox works OK (except when I restart it, it takes a while to reload the 100+ tabs). I know that the pain of occasional sluggishness is nothing compared to the one I’d experience with a newer model where they make extreme savings by removing the touchpad buttons, and using an even tighter keyboard. So I’m keeping it in case the fashion changes back to usable devices again.

        2. I agree with your points. Especially considering we slowley hit a dead end when it comes to performance improvements. They are nearly not as great anymore. At the end of the day: price matters. Price per performance will be the key indicator for a lot of industries. We are running into an high inflation economy. If you can drive prices down, you can come out as a huge winner.

      2. I doubt AVR and PIC cost any license fees as they are microchip proprietary designs, so just development costs to factor in.

        1. It’s the same, the customer at the end of the chain pays for their proprietary features that are not in other ones. A license fee is nothing but a compensation for a possible occasional loss of sales that goes to a competitor that uses your techno and forces you to be more competitive. The license fee is implicitly included in products sold by the technology owner as a higher price.

          1. If they compete with other companies as well? I would expect once development costs were written off you have much more flexibility than an arm licensee…

Leave a Reply

Your email address will not be published. Required fields are marked *