Slashdot Mirror

← Back to Stories (view on slashdot.org)

RISC-V and Linux Foundations Partner to Promote Open Source CPU (techrepublic.com)

Posted by EditorDavid on from the RISC-y-business dept.

"The Linux Foundation and RISC-V Foundation announced yesterday a joint collaboration project to promote open source development and commercial adoption of the RISC-V instruction set architecture (ISA)," reports TechRepublic: Though some devices that integrate RISC-V will use real-time operating systems rather than Linux, the use of Linux in development will be instrumental as existing tools are being extended to support the RISC-V ISA when developing software on traditional computers. "This joint collaboration with the Linux Foundation will enable the RISC-V Foundation to offer more robust support and educational tools for the active RISC-V community, and enable operating systems, hardware implementations and development tools to scale faster," said Rick O'Connor, executive director of the RISC-V Foundation, in a press release.

In many ways, RISC-V is a hardware equivalent to the open source principles that guide the Linux project, as the ISA is open source, is not subject to patent encumbrances, and is available under the BSD license. [L]icensing fees for Arm or MIPS ISAs -- both of which are fundamentally RISC in principle -- can be avoided by using RISC-V.... As alternatives like Alpha, SuperH, MIPS, and even Intel's own Itanium processors have fallen by the wayside, organizations using those ISAs in their products have had difficult adjustment periods transitioning away, while patent encumbrances largely prevent third parties from continuing development or providing drop-in replacements for those technologies. RISC-V's open nature prevents these issues, as it is possible for any organization to extend or customize their own implementation, and any organization can produce their own RISC-V processors.
Manufacturers like how RISC-V CPUs aren't restricted to a single manufacturer, according to the article, which points out that NVIDIA and Western Digital have both announced plans to use RISC-V in some upcoming products.

RISC-V is also "gaining popularity in Internet of Things, low-power, and embedded applications," and Western Digital even plans to ultimately transition its annual consumption of processors -- one billion cores per yer -- to RISC-V.

43 of 92 comments (clear)

  1. So, do tell by Anonymous Coward · · Score: 1

    Who's fabbing these things and where can I buy them? How about an ATX board I can plug these things into?

    1. Re:So, do tell by 110010001000 · · Score: 1

      TSMC will fab them for you.

    2. Re:So, do tell by ShanghaiBill · · Score: 2

      TSMC will fab them for you.

      Western Digital plans to use RISC-V processors in their storage products starting next year. Most likely they will be fabbed by TSMC.

      It is not clear if these RISC-V chips will be available to other companies or to the general public.

    3. Re:So, do tell by 110010001000 · · Score: 1

      Why wouldn't it be available? It is available to anyone. You need to come up with the money for the fab.

    4. Re:So, do tell by ShanghaiBill · · Score: 1

      Why wouldn't it be available?

      Because Western Digital is in the storage business, not the chip business.

      It is available to anyone.

      No it isn't. Open Source doesn't work that way. For instance, if I install Linux on my computer, that doesn't mean I have to give you my computer.

      Western Digital is under no obligation to distribute their chips to anyone, nor do they have any obligation to share the innovations done to turn a Verilog design into working silicon.

    5. Re:So, do tell by alvinrod · · Score: 1

      If you need a lot of ARM chips, or want to be able to do some customization to the architecture (which I believe requires a separate license) eventually the volume cost of using RISC-V is going to be cheaper. This is even a bigger driver if you're trying to produce some component yourself that's designed to be extremely low cost as the additional fees to ARM will become a substantial portion of the cost. It's the same as it was with Windows where no one really noticed the extra cost when computers were hideously expensive themselves, but once companies were able to delivery extremely cheap entry level computers Microsoft had to adjust because the Windows tax became obvious.

    6. Re:So, do tell by mikael · · Score: 1

      That would be insane - an SSD disk drive with a built in GPU / compute engine. That would get close to the "take the CPU to the data" approach for big data processing.

      --
      Vintage computer adverts: http://www.vintageadbrowser.com/computers-and-software-ads
    7. Re:So, do tell by ShanghaiBill · · Score: 2

      These chips are available to anyone who wants a fab to make them

      It is not that simple. You can't just take a pile of Verilog source, even code that runs fine on an FPGA, and slap it onto real silicon. There is a huge amount of work fitting the design to the process. Of course, you could just make a metalized gate array, but that will give you little advantage over just sticking with a core on an FPGA.

      there's no secret sauce there.

      Actually, going from a raw design to a working state of the art chip involves plenty of "secret sauce".

    8. Re:So, do tell by spth · · Score: 2

      Low-end boards are not that expensive. The HiFive1 are just 59 $ in single-unit quantitities. Philipp

    9. Re:So, do tell by ShanghaiBill · · Score: 2

      That would be insane - an SSD disk drive with a built in GPU / compute engine.

      RISC-V does not have an integrated GPU, nor does it even have SIMD vector instructions. Things may change in the future, but for now it is in no way a "compute engine".

      It is ideal for low power embedded systems ... like drive controllers.

    10. Re:So, do tell by Aighearach · · Score: 2

      Yes, it is very clear.

      No, they will not be available to other companies, or to the general public.

      This article, about the Linux Foundation partnering with RISC-V to encourage creation of chips available to the general public. Western Digital was mentioned because they're an example of what is being done already.

      If those things were the same, they would not have even mentioned them both.

    11. Re:So, do tell by Aighearach · · Score: 2

      an SSD drive mostly only needs a fancy multiplexer. The computational part could be done on a SOT-26 package micro otherwise.

      That's why the economics lean so heavily on them doing it themselves; what they need is about the cheapest possible embedded processor with that pin count. Not having to license a design means they push the costs down to the theoretical minimum.

    12. Re:So, do tell by gtall · · Score: 1

      Ditto. If anyone wants to see part of the difference, contrast ASIC designs with FPGA designs. You have to do a lot more work for ASIC and you need chip design experience. There is also a lot that goes into any SoC that isn't part of the ISA. RISC-V is just the ISA, that's it.

    13. Re: So, do tell by zemanel · · Score: 1

      Many componies have closed source riscV implementations that they sell. Because, although the ISA is open source, the guys that write verilog, do the verification, make the physical layout still eat regularly, pay mortgages and have kids studying. RiscV, the ISA, is open source which is awsome because people have a community maintained compiler but they'll still need to do the hardwork or buy closed source solutions. Any optimized cores will never be freely open because that involves a considerable investment. This is not software.

    14. Re:So, do tell by Anonymous Coward · · Score: 1

      I don't think there are any consumer grade ones that are being manufactured yet. I've only seen fpga based dev boards. It's expensive to get such a low volume chip manufactured.

      The cheapest dev board available is $60 (currently OOS): https://www.sparkfun.com/products/15026
      Here's a faster $119 one: https://dev.sifive.com/freedom-soc/evaluate/fpga/

      Despite the hype surrounding RISC-V, it's basically just a free/open processor that's comparable to a MIPS design, without the IP licensing costs. It's likely going to be used for embedded products first (like WD's announcement), and may move up to displace MIPS cores in something like China's homegrown CPU over the next few years. Or it could just peter out.

    15. Re:So, do tell by ChrisMaple · · Score: 2

      If you have the Verilog and a working FPGA, a standard cell ASIC is pretty straightforward, assuming that good design practice was used for the source. For a given level of performance and a given process node, standard cell will have better timing margins.

      AMI used to offer FPGA to standard cell conversion as a service.

      --
      Contribute to civilization: ari.aynrand.org/donate
    16. Re:So, do tell by WhoBeDaPlaya · · Score: 1

      It is not that simple. You can't just take a pile of Verilog source, even code that runs fine on an FPGA, and slap it onto real silicon. There is a huge amount of work fitting the design to the process. Of course, you could just make a metalized gate array, but that will give you little advantage over just sticking with a core on an FPGA.

      Just place-and-route alone can be a pain in the rear, let alone functional verification, design for test, synthesis, etc.

    17. Re:So, do tell by 110010001000 · · Score: 1

      What does this have to do with Western Digital? They don't own the chip! NVIDIA is also using the design.

    18. Re:So, do tell by 110010001000 · · Score: 1

      I think I misunderstood. You are talking about Western Digital's implementation. We are talking about RISC-V in general.

    19. Re:So, do tell by _merlin · · Score: 1

      Only the CPU core itself is open source. You'll need a memory controller, cache controller, PCI-e bridge, and a lot of other stuff to make a working CPU. The RISC-V people will be happy to sell you this stuff under a proprietary license.

    20. Re:So, do tell by pezezin · · Score: 1

      AMD already sells a GPU with an integrated SSD: https://www.amd.com/en/product...

  2. Nvidia had RISC-V on their GPUs for years. by Anonymous Coward · · Score: 1

    The controllers on their graphics cards are RISC-V. Now they're considering implementing their compute cores as well since the compilers are good enough.

    Assuming this trajectory keeps up for the next couple of years, nothing short of a Mill Computing level breakthrough will stop RISC-V from replacing ARM and x86. There's just little to no value in paying for ISA IP when the fabs are doing all the real hard work anyhow.

    1. Re:Nvidia had RISC-V on their GPUs for years. by gtall · · Score: 1

      So with RISC-V, I need to create all the other stuff that goes into an ARM SoC. I presume that grows on trees? For one thing, you'd have to get the parties pushing RISC-V to settle on I/O standards between all those other widgets on an SoC, as well as the widgets themselves. So keep dreaming of pink unicorns, they're out there somewhere.

    2. Re:Nvidia had RISC-V on their GPUs for years. by KonoWatakushi · · Score: 1

      The controllers on their graphics cards are RISC-V. Now they're considering implementing their compute cores as well since the compilers are good enough.

      Assuming this trajectory keeps up for the next couple of years, nothing short of a Mill Computing level breakthrough will stop RISC-V from replacing ARM and x86. There's just little to no value in paying for ISA IP when the fabs are doing all the real hard work anyhow.

      I’m just curious how long we will remain in the dark ages because of Imaginary Property. Many are skeptical of the Mill, but suppose it pans out; how would that innovation benefit people in this lifetime? We’d now have a wonderful new proprietary architecture that no one will touch, because there isn’t a second source. So it will remain confined to niches until the patents run out and someone implements the ideas anew, which may only repeat the cycle with a minor variation. Without an open model, I fear the Mill will be doomed to obscurity.

      The RISC-V ecosystem has demonstrated rapid progress with community efforts, and while the architecture is more attractive than ARM and x86, it is basically a nice yet open conventional ISA which suffers from the same fundamental drawbacks. Even before entering the nightmare of speculative execution exploits, the hardware security mechanisms have long been lacking. Current in-order RISC-V cores are extremely compact, making them attractive for embedded and many-core applications. However, OoO will reduce security even further, and the performance it offers comes at great expense in complexity, area, and power, sacrificing much of the benefit.

      The Mill aims to deliver DSP efficiency and cost on general purpose workloads, and is invulnerable to those exploits and many others. One might argue that the Mill performance advantage is a luxury, but the greatly enhanced security characteristics of a Mill are not; they are basic functionality which is desperately needed by general purpose CPUs. While the Mill is a fascinating novel architecture, the most compelling aspect may be the security model, which will enable efficient microkernels and much greater isolation in applications.

    3. Re:Nvidia had RISC-V on their GPUs for years. by drinkypoo · · Score: 1

      So with RISC-V, I need to create all the other stuff that goes into an ARM SoC. I presume that grows on trees? For one thing, you'd have to get the parties pushing RISC-V to settle on I/O standards between all those other widgets on an SoC,

      They already have done so. There are typical interconnects used between cores and other logic blocks, because this is already how chips are designed by the people who design them. They create blocks which they can glue together to form complete designs.

      --
      "You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
  3. IoT? by TechyImmigrant · · Score: 1

    >"gaining popularity in Internet of Things, low-power, and embedded applications,"

    Because it's a slow CPU architecture.

    --
    I should use this sig to advertise my book ISBN-13 : 978-1501515132.
    1. Re:IoT? by AmiMoJo · · Score: 1

      Relatively slow but also low transistor count and therefore decently low power. Also free to embed in your ASIC as Western Digital is doing.

      --
      const int one = 65536; (Silvermoon, Texture.cs)
      SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
  4. FPGA? (Re:Now we need an open source FAB.) by blindseer · · Score: 1

    Might not need a fab but it would help to have FPGAs that are cheap enough and fast enough to make usable computers for people to use.

    There's some pretty powerful development boards out there that can be turned into a usable general purpose PC if given the right programming, software, and maybe some help with off the shelf on GPUs or such. Just being able to drive a few USB ports for display (DisplayLink USB to HDMI chip comes to mind), storage, keyboard, mouse, etc. can go a long way. Compile a Linux kernel and some other open source software for it, and there's a lot that can be done.

    There would have to be an instruction set built to fit on FPGAs that enough people can afford to play with. Get it this far and perhaps there would be enough interest in time for a fab to produce a lower cost processor as a drop in replacement.

    --
    I am armed because I am free. I am free because I am armed.
  5. Re:Now we need an open source FAB. by drinkypoo · · Score: 1

    Does anyone know of anybody working towards that goal?

    Here's an even better question, is it actually a realistic goal? If a modern fab fails, people around it will die. What happens if your desktop fab fails?

    --
    "You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
  6. I want this in my laptop. by michaelcole · · Score: 1

    It is so time for this.

  7. Re:Now we need an open source FAB. by mermeid007 · · Score: 1

    That might suck but no worse to be less bad than you not having tried one at all. Are these to be underclocked?

  8. Re:Now we need an open source FAB. by ShanghaiBill · · Score: 1

    If a modern fab fails, people around it will die.

    What? What possible fab failure could cause people to die? Sure, a stepper could be a nanometer out of alignment, or a half-step UV photolithography lens could be slightly out of focus, but neither of these is lethal.

    What happens if your desktop fab fails?

    Get a new wafer and another bottle of etchant? Fabbing can be done at home, and there are hobbyists that do it. If a 100 micrometer step size is good enough, and you have time and money to burn, then why not?

  9. Re: The fundamental principle isn't cost-avoidance by mikael · · Score: 4, Interesting

    Unfortunately, many of those patents for performance enhancing features using out-of-order execution were based on a single research paper. That was implemented in one CPU vendor design, then cross-patented to other CPU vendors. RISC-V has the advantage that it doesn't have those vulnerabilities baked in and built upon.

    --
    Vintage computer adverts: http://www.vintageadbrowser.com/computers-and-software-ads
  10. open source ISA, not CPU. by Gravis+Zero · · Score: 1

    It should be noted that RISC-V is an Instruction Set Architecture and not a specific CPU. There are both open and closed implementations of it but they all match a specification. It's like UNIX, there is no single implementation of UNIX but there is a specification of what a UNIX provides.

    --
    Anons need not reply. Questions end with a question mark.
  11. Re:Now we need an open source FAB. by ChrisMaple · · Score: 1

    MOSIS. https://www.mosis.com/

    --
    Contribute to civilization: ari.aynrand.org/donate
  12. Re:The fundamental principle isn't cost-avoidance by Anonymous Coward · · Score: 1

    The most congent observation one can make is that the BSD license killed BSD.

    Linux owes it's success to the GPL. Technologies being equal, if the licenses had been reversed, it would be BSD crowing about its success. Instead BSD cries in its beer.

  13. Need single board equiv, like Raspberry PI by Lady+Galadriel · · Score: 1

    One of the things holding RISC-V back, is a single board computer, usable as a simple desktop. Something like a Raspberry PI, with;

    - at least dual core CPU, potentially with a quad core option
    - reasonable amount of memory, 1GB or more, (or DIMM slots)
    - multiple USB ports
    - storage, (SATA, SDXC or fast USB port for external storage)
    - network, (WiFi, Ethernet, or a fast USB port for network dongles)
    - video, (or PCIe slot for video card)
    - Some expansion, (like PCIe, or more than 1 USB port that is a funnel of I/O bandwidth...)

    This would let people test out software, and run through debuging instructions. Ideally it would have both 32 bit and 64 bit instruction sets, but starting with a 32 bit only might be acceptable in the short term.

    I personally want to see a simple RISC-V 64 bit single board computer like above. Maybe it does not have to be a full blown motherboard with normal features, (like sound, and several PCIe slots), but if it's usable it would let me test out software. Then I can submit bug reports for compiler issues, and OS related quirks or out right bugs.

    --
    Lady Galadriel
  14. Re:Now we need an open source FAB. by pezezin · · Score: 1

    Do you realice how complex and expensive a modern semiconductor fab is? You are never going to see one in your local maker space.

  15. Re:Now we need an open source FAB. by drinkypoo · · Score: 1

    What? What possible fab failure could cause people to die?

    Silane leak. We've never had one because we use double-walled pipes and leak detectors. Peripherally-related rant: We could do the same thing for oil pipelines, and utterly prevent spills, but we don't because that would cut into oil executive yacht purchases.

    It's not safe to have normal people owning and operating chip fabs, because humans are insufficiently responsible. Just look at 3d printers, how many people are using a negative-pressure cabinet with carbon filters on the exhaust to capture their worst emissions? Statistically nobody.

    --
    "You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
  16. Re:They are BRED insufficiently responsible. by drinkypoo · · Score: 1

    Oh, and accept that death is not a bad thing per se.

    Death due to some lame's desktop chip fab is a bad thing by any reasonable measurement. It won't just kill the schmuck who fails at operating it, but also his neighbors. I don't care if people autoeuthanize, but I do care if my neighbor's fuckup kills me

    --
    "You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
  17. Re:"BSD" Copyright by JonMaddogHall · · Score: 1

    "BSD", as some idiot quoted it, is NOT a "License", it's a *Copyright*. There is a huge difference there. Get it right.

    The BSD style Copyright is now best represented by, and deployed as, the following...

    https://www.openbsd.org/policy.html https://cvsweb.openbsd.org/src... https://www.freebsd.org/copyri...

    With additional discussion here... http://landley.net/toybox/lice... https://urchin.earth.li/~twic/...

    You should also know that the next major release of FreeBSD 12.0 will be out in 1.5 weeks :-) You can liveboot the RC3 sampler from USB today.

    Copyright is a legal concept emblazoned in the United States Constitution and in many other countries around the world. To make a lot of legal stuff as simple as possible, it means "if you write the code, you own it and have the right to say what happens to it". Since you are the owner, you now have to do something to allow usage by someone else. That "something" is the license. BSD is a license, not a copyright. BSD always was more permissive than most other licenses, and over the years it removed the few requirements that it had, until now the "0BSD" license is almost indistinguishable from "Public Domain" software. However, it is best to put a license somewhere in your code stream, just to let the people who want to use your code that they can and under what circumstances.

  18. Re:BSD by cas2000 · · Score: 1

    in fact there are a lot of potential negatives for the end-user, including:

      * insecurity by obscurity

      * fragmentation and incompatibility - i.e. vendor lock-in

      * customised back doors

    these problems would be mitigated if the license required sharing via a GPL-like license (GPL v3, to prevent the patent loophole)

  19. standards by sad_ · · Score: 1

    don't know much about the risc-v ecosystem, but lets hope it doesn't turn into the same clusterfuck as ARM.

    --
    On a long enough timeline, the survival rate for everyone drops to zero.