Stories
Slash Boxes
Comments

SoylentNews is people

posted by janrinok on Friday May 20 2022, @09:56AM   Printer-friendly
from the dont-put-some-of-your-eggs-in-too-many-baskets dept.

Tech war: China bets on open-source RISC-V for chip design to minimise potential damage from 'being cut off' by US sanctions

A growing number of Chinese chip design firms have adopted open-source RISC-V in their chip designs as an alternative to Intel's proprietary X86 and Arm's architecture, in a bid to minimise potential damage from US sanctions and to save on licensing fees.

[....] "[This] gives Chinese companies access to a global open standard instruction set architecture (ISA) ecosystem," said Stewart Randall, head of electronics and embedded software at consultancy Intralink. "So Chinese companies can have access to, and create, their own cores or chips based on it."

However, some industry experts said China's adoption of open-source RISC-V architecture would not shield them from all US sanction risks, as America still holds the trump card when it comes to electronic design automation (EDA) tools, the key software needed for chip design, as well as chip manufacturing technologies.

If you really want to create your own cores from scratch, without licensing anyone else's IP, is it truly possible to do so with RISC-V?

See Also:

Tech war: China bets on open-source RISC-V for chip design to minimise potential damage from 'being cut off' by US sanctions


Original Submission

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 4, Insightful) by bradley13 on Friday May 20 2022, @10:32AM (10 children)

    by bradley13 (3053) Subscriber Badge on Friday May 20 2022, @10:32AM (#1246532) Homepage Journal

    What is really astounding, is that the Intel architecture has maintained dominance for so long. Back in the dark ages, there were three contenders: National Semiconductor, Motorola, and Intel. NS had a clean architecture and well designed instruction set. Motorola was a bit messier, but OK. Intel's architecture was a total kludge.

    Somehow,Intel won. Better marketing? Back room deals? No idea, bit the x86/x64 architecture has only gotten kludgier over the decades.

    --
    Everyone is somebody else's weirdo.
    Starting Score:    1  point
    Moderation   +2  
       Insightful=2, Total=2
    Extra 'Insightful' Modifier   0  
    Karma-Bonus Modifier   +1  

    Total Score:   4  
  • (Score: 5, Interesting) by maxwell demon on Friday May 20 2022, @11:34AM (2 children)

    by maxwell demon (1608) Subscriber Badge on Friday May 20 2022, @11:34AM (#1246541) Journal

    IMHO it's actually a combination of several factors:

    • IBM chose it for their PC.
    • IBM required a second source for the processors, thus we got a competition of different manufacturers selling the same ISA, driving the price down.
    • IBM built their PC entirely from standard parts. IBM also did not require an exclusive licence from Microsoft for DOS. Those two facts allowing competition also on IBM compatible systems (though in this case, that definitely wasn't IBM's intent). Of course that include the system to be x86.

    In short, had IBM chosen another CPU back then (and it almost doesn't matter which one), that other CPU would likely now be the industry standard.

    Note that IBM did not, at that time, take the PC seriously (thus the open design and lack of IP protections), thus they probably chose the CPU mainly based on cost.

    --
    The Tao of math: The numbers you can count are not the real numbers.
    • (Score: 1, Informative) by Anonymous Coward on Friday May 20 2022, @05:20PM (1 child)

      by Anonymous Coward on Friday May 20 2022, @05:20PM (#1246629)

      In short, had IBM chosen another CPU back then (and it almost doesn't matter which one), that other CPU would likely now be the industry standard.

      I think that is correct, but completely unjustified by everything else you said. The question is not why a particular architecture won the battle in the early eighties. The question is why we are still using a descendant of that architecture 40 years later. And the answer is that in each subsequent technology generation consumers valued compatibility with their existing install base over any other factor.

      • (Score: 2) by DannyB on Saturday May 21 2022, @12:22AM

        by DannyB (5839) Subscriber Badge on Saturday May 21 2022, @12:22AM (#1246738) Journal

        Software lock-in is a powerful phenomena. IBM had figured this out by the 1960s.

        Backward compatibility with an existing expensive collection of software is also a big draw.

        Can I interest you in a new processor that will require you to re-purchase and re-write all of your expensive software?

        Now with the rise of Linux, this problem began to disappear. Once the OS and compiler(s) are ported for other architectures, the huge software base tends to come along for a fairly low cost. But that was not true for much of computing history. Especially in the 1980s and prior.

        Once IBM chose Intel x86 for their PC -- we were stuck with it for decades.

        Now we are in the situation where the entire value proposition of both Windows and Intel x86 is nothing more than the compatibility with the existing software base. There are both better processors and better operating systems. Yet the WinTel duopoly will continue for some time just because of the tremendous need of backward compatibility with bass ackward systems.

        --
        I get constant rejection even though the compiler is supposed to accept constants.
  • (Score: 0) by Anonymous Coward on Friday May 20 2022, @11:46AM (5 children)

    by Anonymous Coward on Friday May 20 2022, @11:46AM (#1246542)

    "Somehow,Intel won."
        Being code translatable-ish from the 8080 and the design win for the IBM pc didn't hurt.
        Sometimes just having good enough tech makes other things the deciding factor.
        Like code base compatibility supported by some of the kludges.

    Risc-V could end up that way.
      How good are the open implementations. Do they include all the tricks that a modern processor uses to go fast?
      It takes a lot of tricks to get a classic single instruction stream cpu clocking multiGhz with multiple instructions per clock.

    IF SOapbox,present()
    The can you go it alone question probably applies to everybody. There are critical pieces spread over the world. EDA tools for design, sim, and process modeling, chip architects in many different directions, fabrication tools and knowhow, raw materials, packaging, and systems integration to make the stuff do something useful. I'd bet there is no single country that can totally go it alone. Humanity did amazing things when we all worked together. Too bad we can't agree on a common set of rules. Attempts by one cuontry or another to get ahead are lowering the living standard for all. The get aheaders included.
    ENDIF

    • (Score: 3, Informative) by driverless on Friday May 20 2022, @12:55PM (4 children)

      by driverless (4770) on Friday May 20 2022, @12:55PM (#1246555)

      Risc-V could end up that way.
          How good are the open implementations.

      If by "open" you mean "Chinese implementations of the open Risc-V architecture", they're pretty bad. Actually, really bad: Buggy hardware, buggy software, almost nonexistent documentation, and poor performance. Imagine buying some random x86 box off Aliexpress which outside of the CPU is full of semi-unknown support chips and partly-debugged circuitry, and that's the level of the hardware. We have a couple of Risc-V devices here and each one has been a massive headache just to get going, with a lot of the capabilities unexplored for lack of documentation and huge amounts of time wasted discovering and working around hardware bugs. The icing on the cake is that each one only seems to last 6-12 months before it's abandoned and replaced with something else with a new set of bugs and problems.

      It's really frustrating to have this nice new architecture available but the implementations are so totally underwhelming in terms of functionality, documentation, and performance. So at the moment I think it's more a bargaining chip than anything else, and I haven't caught wind of anything new coming down the track that would change that.

      • (Score: 1, Insightful) by Anonymous Coward on Friday May 20 2022, @08:54PM

        by Anonymous Coward on Friday May 20 2022, @08:54PM (#1246694)

        So, just like no-name ARM SoCs but with RISC-V cores. Hmmm...

      • (Score: 2) by DannyB on Saturday May 21 2022, @12:25AM

        by DannyB (5839) Subscriber Badge on Saturday May 21 2022, @12:25AM (#1246740) Journal

        It is a mistake to assume that bad implementations of an idea at present mean it will always be bad.

        These new fangled automobile thingies will never be a success. They are noisy, smelly and unreliable. They make too much noise. You can actually break your arm if it backfires while you are crank starting it. And worst of all -- they frighten the horses! I can confidently predict that the beautiful horse and buggy will be with us forever and these new automobile thingies are just a temporary fad.

        --
        I get constant rejection even though the compiler is supposed to accept constants.
      • (Score: 0) by Anonymous Coward on Sunday May 22 2022, @05:21PM (1 child)

        by Anonymous Coward on Sunday May 22 2022, @05:21PM (#1247054)

        i find it amazing that you guys go thru all the trouble.
        i hope, at least, you're getting this garba... err.. stuff for free!
        and thx for the heads-up. somebody's got to look to see...

        • (Score: 0) by Anonymous Coward on Sunday May 22 2022, @05:51PM

          by Anonymous Coward on Sunday May 22 2022, @05:51PM (#1247059)

          ah, also, if the butcher starts eating his/her own sausages then ... well ... i guess i'll eat them too.

  • (Score: 2) by fliptop on Friday May 20 2022, @02:24PM

    by fliptop (1666) on Friday May 20 2022, @02:24PM (#1246578) Journal

    there were three contenders: National Semiconductor, Motorola, and Intel

    Texas Instruments? They bought National Semiconductor about 10 years ago, but they were a player in the "dark ages."

    --
    To be oneself, and unafraid whether right or wrong, is more admirable than the easy cowardice of surrender to conformity