Follow Slashdot blog updates by subscribing to our blog RSS feed

 


Forgot your password?
Close
typodupeerror
×
Iphone Apple

Apple Shows Interest in RISC-V Chips, a Competitor To iPhones' Arm Tech (cnet.com) 109

Posted by msmash from the shape-of-things-to-come dept.
Apple wants to hire a programmer who knows about RISC-V, a processor technology that competes with the Arm designs that power iPhones, iPads and newer Macs. The company's interest emerged in a job posting for a "RISC-V high performance programmer" that Apple published Thursday. From a report: It's not clear exactly what Apple's plans are for the technology. Landing even a supporting role in an Apple product would be a major victory for RISC-V allies seeking to establish their technology as an alternative to older chip families like Arm or Intel's x86.

One of the RISC-V's creators is seminal processor designer David Patterson, and startups like SiFive and Esperanto Technologies are commercializing RISC-V designs. The job description offers some details about Apple's plans. The programmer will work on a team that's "implementing innovative RISC-V solutions and state of the art routines. This is to support the necessary computation for such things as machine learning, vision algorithms, signal and video processing," the job description says.
This discussion has been archived. No new comments can be posted.

Apple Shows Interest in RISC-V Chips, a Competitor To iPhones' Arm Tech More

Apple Shows Interest in RISC-V Chips, a Competitor To iPhones' Arm Tech

Comments Filter:

  • experience required (Score:4, Funny)

    by algaeman ( 600564 ) on Friday September 03, 2021 @04:10PM (#61760505)
    Does the job posting require 10 years of experience with RISC-V?
    • Probably just RISC cpus in general. Power PC cpus are RISC chips. They were designed and made by the AIM alliance. Surprising they donâ(TM)t have a one still on payroll who could fill that need.

      • Re: (Score:1)

        by Anonymous Coward

        Probably just RISC cpus in general. Power PC cpus are RISC chips. They were designed and made by the AIM alliance. Surprising they donâ(TM)t have a one still on payroll who could fill that need.

        OP was making the point that there are almost always (near) impossible requirements in a job listing like this. RISC-V being "created" in 2010 (not sure when the first actual system came online), it would be virtually impossible to have that 10 years requirement, yet most postings would still say that.

        • My point was that while RISC-V may only be 11 years old, RISC CPUs are legacy and Apple used to make some, and used them in Macs. I was surprised no one inside Apple fit the bill.
          • Comment removed based on user account deletion
          • 1) You missed the joke.
            2) I've been writing low level software for "RISC" and "CISC" CPUs for a few decades, and in those decades I have learned this:
            There are 2 kinds of people that use the term RISC.
            Marketing dumbfucks, and poser dumbfucks.

            There's no magical or special architectural knowledge that comes along with "RISC"
            PowerPC, ARM, and MIPS are all "RISC" (whatever the fuck that even means anymore)
            Writing machine code for PowerPC and going to ARM will cause your head to explode when you first run

      • Does the job posting require 10 years of experience with RISC-V?

        Probably just RISC cpus in general. Power PC cpus are RISC chips. They were designed and made by the AIM alliance. Surprising they donâ(TM)t have a one still on payroll who could fill that need.

        Whoosh!

      • And cisco routers, and SPARCstations, etc. plenty of people with prior RISC experience.

  • My guess would be Apple is planning on adding some kind of support chip and probably to its iOS family devices. Something that can be very low power that can either remain active while most of the ARM cores are sleeping or maybe something to farm out certain chunks of algorithms to where a big fancy ISA does not off much advantage and a simpler core can get more computation done with less power consumption.

    • Hedge (Score:5, Insightful)

      by JBMcB ( 73720 ) on Friday September 03, 2021 @04:26PM (#61760549)

      It's probably a hedge, in case RISC-V ends up being more efficient than ARM, or if nVidia ties up ARM in overly restrictive licensing.

      I doubt that chips based on an ISA that's a few years old is going to be anywhere near as efficient as ARM, who has been optimizing their architecture for almost 30 years.

      • It's probably a hedge, in case RISC-V ends up being more efficient than ARM, or if nVidia ties up ARM in overly restrictive licensing.

        I doubt that chips based on an ISA that's a few years old is going to be anywhere near as efficient as ARM, who has been optimizing their architecture for almost 30 years.

        As an ARM founder, they have a perpetual license. I think you're right about hedging, though it's possible they just want to keep abreast of what's happening in other segments.

        • I see this claim that Apple has a "perpetual license" bandied about but I don't see any verifiable source for it.

          • ARM appears to have said they are "long term"

            https://appleinsider.com/artic... [appleinsider.com]

            Term is one step down from perpetual

            https://www.anandtech.com/show... [anandtech.com]

            I rate the claim "believed false" until a better citation comes along :D

            • No.
              The Perpetual license licenses a specific core perpetually.
              The Term license licenses a specific core for a specific amount of time.
              Above all of those, is the architectural license, which is also perpetual.
              The architectural license allows you to make an ARM-compatible core, instead of using an ARM core.
              That is what Apple has.

              The catch, is that an architectural license is only perpetual for the architecture you license it for.
              If you had one for ARMv7, you need a new one for ARMv8, etc.
              Any core you
          • ARM has architectural licenses, and they are perpetual.
            They're specific to the generation (ARMv7, ARMv8, etc) that you purchase them for.

            We can deduce that Apple does have one of these, because it's the only ARM license that allows the level of change hat Apple has made.
            Qualcomm, Samsung, Intel, and AMD (among others) also have them.

            If you don't have an architectural license, you can't make ARM compatible cores, you have to make actual ARM cores (though you're free to put as much other stuff on the So

            • Deductions are not evidence.

              • No, of course they aren't.
                Deductions are what are made from evidence.
                In this instance, the evidence is that Apple's cores are not ARM cores, so they have an architectural license. These are a standard part of ARM's licensing portfolio, and several big companies (Samsung, Intel, Qualcomm, AMD) all have them too.
                The evidence for them is similarly good: They produce (or produced) ARM compatible cores (as opposed to ARM cores)

                Architectural licenses are perpetual (for that architecture), though you still pa

                • You have no idea what the terms of Apple's architectural license are, so you are speculating and twisting around that with weasel words. Oh I know, you Apple people: shut up you're analyzing it wrong!

                  • Lol. I had forgotten how much of a fucking idiot you are.

                    I have worked for an ARM architectural licensee. I have even gone through the hell of running their compatibility suite to validate engineering samples (Architectural licensee cores must pass this suite, or the product is not licensed)
                    I've even been legally threatened by another ARM architectural licensee for a published CVE in my name. Also a fun story. There's an article about the exploits that led to it on this very site, if you'd like to read i

                    • Methinks you doth protest too much. In way too many words you just confirmed you know nothing about the terms of Apple's license just as I said, and you're speculating. What a self absorbed dipshit.

                    • You're right, I don't know anything about "Apple's license"
                      I do have first-hand knowledge of 2 other organizations with architectural licenses.
                      You seem to be arguing that Apple has less favorable license terms than them.
                      This is nonsensical on its face.

                      Deduction of Apple's license is possible based on known facts and other inside information from organizations who have published the terms of their architectural licenses.

                      Self-absorbed? Perhaps. Dipshit? Not remotely.

                      So, again, what the fuck do you do

        • As an ARM founder, they have a perpetual license.

          Their architectural license actually has nothing to do with their once-involvement in the most recent iteration of Arm.
          Many companies have architectural licenses.

      • It's probably a hedge, in case RISC-V ends up being more efficient than ARM, or if nVidia ties up ARM in overly restrictive licensing.

        Nvidia would have to rewrite contract law to do that.

        • Re: (Score:2)

          by Junta ( 36770 )

          I would presume that nVidia would be able to decline to extend a licensee's rights to ARMv10 and up, if they succeed in acquiring ARM. Maybe they can't revoke entitlement to ARMv9, but they certainly could control access to new IP. As to the likelihood that they would want to, that's more debatable.

          Even if Apple continued and made amazingly awesome processor based on ARMv9 that was technically superior to ARMv14, the optics would still awkward about seemingly being left behind by the platform they are locke

          • Maybe they can't revoke entitlement to ARMv9, but they certainly could control access to new IP.
            Only if it is their own IP. And why would Apple be interested in that? They are making their own chips/socs.

            • Re: (Score:2)

              by Junta ( 36770 )

              The same can be said of all ARM licensees, but the licensees all want to stay abreast of any new IP from the platform. If some very nice capabilities come along, then Apple may be stuck without access and trying to do a 'knock-off' may earn a lawsuit from nVidia.

              Sure they could decide that ARMv9 is the end of the line for them, and de-emphasize the fact that is even ARM at all (like how Apple of early 2000s went to the trouble and expense to get OSX officially certified as Unix, then stopped bothering whil

              • RISC-V is structured in a way that even if the copyright of the name is sold, the last iteration before sale is not only still available, but those using it are entitled to re-share and effectively fork it into a continuing open ecosystem, even if they have to change the name, they can still share the IP.

                This is the critical aspect.
                ARM is highly patented, and ARM architectural licenses require passing of a compatibility suite. Meaning Apple can't make any modifications to improve their processor if it breaks compatibility.
                RISC-V is free to use. If you break compatibility, you just can't call it RISC-V anymore. But you can break compatibility. That's something you cannot do with an ARM architectural license.

                • require passing of a compatibility suite. Meaning Apple can't make any modifications to improve their processor if it breaks compatibility.
                  Apple is not selling it's chips under an ARM label. AFAIK they are not selling them at all.
                  Hence: no one cares about any compatibility.

                  That's something you cannot do with an ARM architectural license.
                  Of course you can. The ARM chips are deliberately designed so that a licensee can make his own variations from the design. No one cares if a TI ARM or an IBM ARM or an nVIDI

                  • Apple is not selling it's chips under an ARM label. AFAIK they are not selling them at all.

                    That isn't relevant.
                    ARM owns the IP, not Apple. Apple merely licenses it.

                    Hence: no one cares about any compatibility.

                    What. The. Fuck. Are. You. Talking. About.

                    Of course you can. The ARM chips are deliberately designed so that a licensee can make his own variations from the design. No one cares if a TI ARM or an IBM ARM or an nVIDIA ARM or an Samsung ARM are compatible. Because that is not the point of having a license. The point is having your own SOC which does exactly what you want it to do.

                    No.
                    ARM offers several tiers of license.
                    The bottom tier, you license a straight netlist for a specific core for integration into your own SoC. You are not allowed to modify it at all.
                    The middle tier, you license the VERILOG source for a specific core and are allowed to make non-core related changes to it (caching, core arrangement, etc)
                    The top tier, you license a specific architecture (ARMv

                    • That isn't relevant.
                      ARM owns the IP, not Apple. Apple merely licenses it.
                      It is relevant.

                      As Apple obviously owns its own IP.

                      Regardless if it is based on old ARM stuff or not ... sorry: you are out of your league (fuckface?)

                      The bottom tier, you license a straight netlist for a specific core for integration into your own SoC. You are not allowed to modify it at all.
                      That is plain wrong, as ARM instruction set and Verilog designs were always explicitly designed so that licensees can vary form it as much as they

                    • As Apple obviously owns its own IP.

                      How the fuck is this relevant?
                      The M1 is made with IP licensed from ARM.
                      It is not something made with nothing but Apple's IP.
                      Without license, the M1 cannot be produced. What are you trying to say?

                      Regardless if it is based on old ARM stuff or not ... sorry: you are out of your league (fuckface?)

                      What?
                      You're making no fucking sense.

                      That is plain wrong, as ARM instruction set and Verilog designs were always explicitly designed so that licensees can vary form it as much as they wish.

                      Incorrect.
                      Per the usual, you have no idea what the fuck you're talking about. [semiaccurate.com]

                      What the fuck is miswired in your brain that you feel compelled to offer misinformation in the form of facts on seemingly every fucking topic?
                      I have lost fucking count of how many times you've be

                    • Why do you not simply go to the ARM website, and figure what you need to do to build your own ARM bases processor, with your own processor extensions like everyone else who is exactly doing that: is doing?

                      Hu?

                      You are talking about stuff you have no clue about. And insult me constantly.

                    • Why do you not simply go to the ARM website, and figure what you need to do to build your own ARM bases processor, with your own processor extensions like everyone else who is exactly doing that: is doing?

                      What the fuck are you talking about?
                      Feel free to send any link from any ARM website that says you can do that. You won't be able to of course, because such a thing does not exist.
                      As I said, I have worked for an ARM licensee. I'm familiar with the details.

                      I'll give you some examples.
                      The MSM7201A is a SoC built before Qualcomm had an architectural license.
                      The SoC is their design, of course, but lacking the rights to build their own core, it includes a bone stock ARM1136J-S, and an ARM940T. Qualcomm is n

                    • As I said, I have worked for an ARM licensee. I'm familiar with the details.
                      Then post Apple's ARM license?

                      https://www.arm.com/why-arm/cu... [arm.com]

                      Idiot?

                    • Oh, you poor fucking dolt.
                      The ARM Flexible Access License doesn't let you modify the core
                      It licenses cores for a custom SoC

                      Do you really not know the difference? ARM licenses many function blocks. This allows you to pull in the ones you want in a "custom" fashion.

                      Fucking hell. Don't you ever feel stupid?

              • Sure they could decide that ARMv9 is the end of the line for them, and de-emphasize the fact that is even ARM at all
                Apple actually is not calling it ARM.
                They call it "Apple Silicon".

            • Simple. Because the ARM architectural license requires compatibility with the architecture you licensed.
              This means that if Apple wants to continue to advance their ISA along with the rest of the ARM ecosystem, they have to continue purchasing new architectural licenses for each new architecture.
              They could of course just stay stuck in time with what they have, but ultimately, it means they cannot make compatibility breaking improvements with being in breach of the license.

              • That is complete nonsense.

                Most ARM architectures are not compatible with older ones.

                And Apple is not selling "ARM (TM)" chips. They are simply making their own chips based on their old ARM IP.

                No idea in what fantasy wold you live that you think you know anything about the founder's of ARM "license". They probably do not even have a "license" in the classical sense. But simply own the IP, and licensed it out to ARM.

                • That is complete nonsense.

                  Oh boy. Here we go again.

                  Most ARM architectures are not compatible with older ones.

                  Complete fucking bollocks.
                  The ISA hasn't changed significantly from the ARM7TDMI.
                  Sure, there are extensions to it. aarch64 being a notable example. But 32-bit and 16-bit modes are still there, and the instructions are the same, and the behavior is the same.

                  And Apple is not selling "ARM (TM)" chips. They are simply making their own chips based on their old ARM IP.

                  They are making ARM-compatible chips.
                  It's a requirement of their license. Any core they make has to pass the ARMv9 compatibility suite in order to be licensed.

                  No idea in what fantasy wold you live that you think you know anything about the founder's of ARM "license". They probably do not even have a "license" in the classical sense. But simply own the IP, and licensed it out to ARM.

                  Apple is not an owner of ARM. What they were a long time ago isn't r

                  • Complete fucking bollocks.
                    Non of a modern ARM can run my ARM 3 code.

                    So no idea about what you want to nitpick.

                    If they had special magical terms, then they would not both with ARM compatibility whatsoever; and yet, a generic linux ARMv9 target boots just fine on the M1.
                    Because it is not a generic ARMv9? Perhaps?
                    Or they thought it makes sense to be actually compatible?

                    You are nitpicking about stuff that has nothing to do with the "license", which neither you nor me ever have seen or read.

                    You're out of your le

                    • Non of a modern ARM can run my ARM 3 code.

                      You don't have any ARM 3 code.
                      I know this because you're an ignorant toolshed. Every claim you ever make turns out to be easily provably false.

                      "ARM 3" is pre-ARM7TDMI. Those were MMU-less chips. They're like an 8086 vs an 80286.
                      They're related, but they're not remotely the same.
                      ARMv4 (ARM7TDMI) 32-bit and 16-bit code will run on ARMv9. There are some errata involving unaligned memory access (which is OK in ARMv8+) that some people abused the undefined behavior of, but every general processor change is

                    • You don't have any ARM 3 code.
                      As I'm one of the oldest ARM 3 owners, yes I have.

                      No idea what your rant is about.

                      Apple doesn't get special treatment from ARM. They're just another company licensing ARMs work.

                      Yeah, and what has that to do with the topic?

                      Apple is doing its own ARMs - and those have nothing to do with the company ARM. No idea about what you want to rant. Apple can invent as many ARM based SoCs as they want and ARM "the company" can't do anything about it.

                      I spared myself to read the rest of your

                    • As I'm one of the oldest ARM 3 owners, yes I have.

                      Now that's a pretty laughable claim.
                      I believe you're an owner of one. I don't believe you ever wrote a single line of code on it, judging by the fact that you don't seem to know anything at all about its ISA.
                      Or perhaps you wrote code in a higher level language? But if that's the case... why the fuck are you commenting?

                      Yeah, and what has that to do with the topic?

                      It was fundamental to your insinuation that Apple somehow owned the right to use ARM IP.

                      Apple is doing its own ARMs - and those have nothing to do with the company ARM. No idea about what you want to rant. Apple can invent as many ARM based SoCs as they want and ARM "the company" can't do anything about it.

                      Apple is building ARM-compatible cores contained within SoCs, licensed by ARM. ARM owns the ARM ISA. A

                    • I believe you're an owner of one. I don't believe you ever wrote a single line of code on it
                      That is up to you.

                      Hint: because of my /. name: I'm easy to google.

                      Idiot.

                      See above, where I said Apple is not getting special treatment from ARM.
                      Apple and ARM have no relationship with each other since Apple sold its stock in it, decades ago.

                      You are simply talking about stuff that has nothing to do with reality.

                    • Hint: because of my /. name: I'm easy to google.

                      I have. And that's another reason I know you're an idiot.

                      Apple and ARM have no relationship with each other since Apple sold its stock in it, decades ago.

                      No shit. What the fuck is wrong with you. That was my point to begin with.
                      You implied that Apple, as a "Founder of ARM" had some kind of special privileges, or that ARM cores somehow belonged to them, or that since they were once an owner, everything that is ARMs today is somehow theirs.

                      Apple licenses the rights to build ARM compatible cores in their SoCs.
                      That is not a right they have without ARMs consent. Apple's IP is not fucking relevant.

      • It's probably a hedge, in case RISC-V ends up being more efficient than ARM, or if nVidia ties up ARM in overly restrictive licensing.

        I doubt that chips based on an ISA that's a few years old is going to be anywhere near as efficient as ARM, who has been optimizing their architecture for almost 30 years.

        Since Apple is also apparently trying to bring pretty much everything it touches in-house, could it also possibly be baby steps towards building their own data centers with custom hardware? Right now they farm a great deal of their cloud ops out to hosts like Azure and AWS.

        If anyone could afford this approach, and have the management and foresight to make it profitable long run, it's certainly Cupertino.

    • Re: (Score:2)

      by MikeMo ( 521697 )
      Maybe it's for a different product altogether - maybe one that isn't power constrained. Like a car.
  • That's essentially what you need for self-driving algorithms. Sounds like a job in the car division.

    • Genuine question: Why is choosing RISC-V superior to ARM for self-driving algorithms?

      • or nevermind, you meant the job posting requirements .. not the architectural choice?

        • Re: (Score:2)

          by mccalli ( 323026 )
          Yes, meant the job requirements seem to match a vision training algorithm job. As to why RISC-V....not sure, but could well be that they're working on an existing SoC setup.

          I could well be off and this is for iCloud server-based video processing for some future thing, but they do seem active in car recruiting right now.

  • seems familiar (Score:3)

    by celeb8 ( 682138 ) <celeb8&gmail,com> on Friday September 03, 2021 @04:36PM (#61760567)
    What Apple is going to try RISC chips instead of Intel? Hmmm this seems somehow familiar...
  • Apple does know that programming in RISCV is the same as programming in any language right? It's gcc at the end of the day

    • Re: (Score:3)

      by jythie ( 914043 )
      Unless you are using assembly, or processor specific features. For that matter, using GCC with embedded processors rather than desktop ones always takes a bit more knowledge and tinkering.. and if they are not developing to a preexisting board it might require writing the mapper files that GCC uses to cross compile.
      • All true. However, it takes any embedded developer worth their salt about 8 minutes to build the linker maps.
        Now, getting your first baremetal bootstrapper running on the proc, that part is a real bitch.

    • Apple does know that programming in RISCV is the same as programming in any language right?

      Apple is specifically looking for someone who knows HPC on RISC-V.

      That is a lot more than just slapping together some C code.

      A qualified candidate will need to understand how RISC-V does vectorization, threading, caching, pipelining, SMP, etc.

      TFS describes many applications that require knowledge of optimizing low precision (FP16) matrix operations, and a thorough understanding of the interface to the GPU and/or neural engine. That is not something that a typical C coder will know.

    • Re: (Score:2)

      by tlhIngan ( 30335 )

      Apple does know that programming in RISCV is the same as programming in any language right? It's gcc at the end of the day

      Someone has to write the backend to that compiler, and compiler efficiency is a primary consideration. You can have a great architecture, but lousy compilers will hold it back.

      The itanium ia64 is an example of this - the architecture puts into software what used to be in hardware, but the compiler technology just never caught up with it and thus it got saddled with poor performance becau

    • No, they didn't know that. Thanks for letting 'em know. I'm sure Apple's head of software, UC Berkeley CS grad and former head of MacOS engineering Craig Federighi, will be happy to hear this news.

    • Apple does know that programming in RISCV is the same as programming in any language right? It's gcc at the end of the day

      No its clang at the end of the day.

      Also, contrary to popular mythology, even in C/C++ the programmer needs to understand the underlying architecture to write higher performing code. It is still the case that C/C++ programmers sometimes have to organize code to give hints to the compiler.

      And again, contrary to popular mythology, sometime the human can do a better job in assembly language. Its not as often as it used to be but it is still the case at times.

      • And again, contrary to popular mythology, sometime the human can do a better job in assembly language. Its not as often as it used to be but it is still the case at times.

        It's complicated. There are some things that compilers are notoriously good at optimizing. There are some things, they're not.
        Still, I do quite a bit of assembly in anything performant.

        Ultimately, there's just less need these days due to the fact that hardware is so damn fast.
        Can you squeeze another 30% out of a JPEG decoder written purely in C? Ya, you can. But is it really worth it?

        • Re: (Score:2)

          by drnb ( 2434720 )

          And again, contrary to popular mythology, sometime the human can do a better job in assembly language. Its not as often as it used to be but it is still the case at times.

          It's complicated. There are some things that compilers are notoriously good at optimizing.

          Yes, things found in benchmarks. :-)

          Ultimately, there's just less need these days due to the fact that hardware is so damn fast.

          Absolutely. Usually to justify the increased cost of development and maintenance there must be a perceptible improvement for the user. Not measurable with a stopwatch but perceptible with casual observation.

          Can you squeeze another 30% out of a JPEG decoder written purely in C? Ya, you can. But is it really worth it?

          If you are Apple and you are going to have your server farm decode billions of JPEG, probably.

  • Apple has to pay royalties to ARM, granted, comparatively less than other licences, but still a good chunk of dough.

    Apple also knows that their market share can not grow ethernally, so, the % of sales that the royalty represents will be heavier and heavier.

    While currently RISC-V is not fit for an iPhone (hell, probably is still not even ready for an iWatch), it will steadily improve. Apple can get an early start on the RISC-V, and

    • Re: (Score:2)

      by mbkennel ( 97636 )

      I believe they don't, uniquely. They were a major investor in ARM, essentially founding it in 1990 to work on the Newton, and out of that deal they have a perpetual license to the instruction set. I don't know what specifically it covers.

      https://appleinsider.com/artic... [appleinsider.com]

      • I sincerely doubt that's the case.
        Apple's status as one of the initial investors in modern day ARM doesn't really matter since they they sold their shares long ago.
        All ARM licensees pay royalties- even architectural licensees.
        If Apple had some kind of contract that gave them forever access to whatever ARM comes up with, then they wouldn't care who bought them, but they do. They weigh in every time someone buys them, which means they're concerned about the details of any future licenses with them they'll

  • Mixed Feelings (Score:3)

    by organgtool ( 966989 ) on Friday September 03, 2021 @05:25PM (#61760739)
    I should start by saying that Apple's job posting is a far cry from a large, dedicated commitment to RISC-V, but it is certainly interesting to see where it goes. I have mixed feelings about Apple using RISC-V. On the one hand, after the way Apple has bullied lots of other tech companies in lawsuits, including Nvidia, I kind of want to see them suffer a bit as a result of that bullying, and that is quite possible if Nvidia's deal to buy ARM goes through (yes, I know Apple has a perpetual license to use ARM technology, but Nvidia could play a shell game that puts new developments under a different shell that falls outside of Apple's agreement). On the other hand, I REALLY want RISC-V to succeed, and I can't think of another company that could do more to make that happen than Apple.

    With that said, I'm pretty sure that it's a long shot for Apple to switch architectures again so soon after switching to ARM, but who knows: they're all about vertical integration and independence and RISC-V offers way more independence than ARM.

    • Jesus H Christ, calling behavior bullying is something that pussies do. And anyway, one corporation cannot by definition "bully" another company. What, did Apple call Nvidia names? Did it kick sand in Nvidia's face? Did Apple give Nvidia a wedgie and throw them into a trash can?

      Come on, get real.

      • pussy is a dildo word

        Using the word pussy to disparage just shows you don't know jack about pussy

    • Re: (Score:2)

      by tlhIngan ( 30335 )

      It's likely less about "Apple is switching architectures" and more "Apple is researching RISC-V".

      Why? probably because Apple is keeping tabs on competition. Is it a threat? Is it any good? Can we do something with this? Etc.

      ARM is good, but it doesn't hurt to keep track of other architectures out there in case there may be a use for it that's disruptive. For example, Apple AirTags. Would using RISC-V make the battery last longer? Same goes for the other things they have that are battery powered, like the Ap

    • On the other hand, I REALLY want RISC-V to succeed, and I can't think of another company that could do more to make that happen than Apple.

      Holy crapola. You haven't been paying attention.
      Apple has zero incentive to maintain compatibility with RISC-V, and due to RISC-V's open ISA licensing, they don't have to.

      Ultimately, the fact that the computer I'm typing this on is an ARM instead of an Intel now isn't relevant ARM.
      Intel cares, because they lost money, but ARM sure as shit didn't succeed more or benefit by it (Short of the royalties for the chips, and the large sum for the architectural license)

  • So did VLIW just fall flat on its face? At one point in time Linus Torvalds worked for Transmeta writing their code morphing software on their 128bit VLIW processors.

    • So did VLIW just fall flat on its face? At one point in time Linus Torvalds worked for Transmeta writing their code morphing software on their 128bit VLIW processors.

      VLIW was an architecture, not a company.

      Linus worked for TRANSMETA, a company doing a VLIW x-86 compatible processor. That company went broke.

      Other high profile VLIW champions were Intel and HP with Itanium.

      But WLIB is still used in the embedded market, in signal processing chips

      • Re: (Score:2)

        by robi5 ( 1261542 )

        > VLIW was an architecture, not a company ... Linus worked for TRANSMETA ...

        I guess the GP knew all this already

    • It sucked for general purpose computing. Nobody ever figured out how to fix that and every attempt to do so failed.

  • This is just Apple hedging their bets. They already have the perpetual ARM license, but maybe there are some things about RISC-V they could incorporate or maybe they're starting to hit a wall with what they can do with ARM and are looking at RISC-V to carry them further. Maybe they also like the idea of RISC-V being an open spec so they can cherry pick ideas from other people for free.

    • They have an architectural license.
      It isn't what you think it is.
      There are many companies with architectural licenses.

      They only grant you the ability to make cores that pass compatibility with the architecture that you licensed (For Apple, that's ARMv9)
      They do not get automatic license to use ARMv10 if/when it comes out.
      Further, should someone hostile acquire ARM and refuse to license Apple for any new architectures, Apple can't do anything to their ARMv9 compatible cores that make them no longer comp

Slashdot Top Deals

I've noticed several design suggestions in your code.

Close