8 ms·

Fallacies about RISC-V

83 points by JoachimS 4y ago

olliej 4y ago
Fairly generic fluff piece, but it’s fairly disingenuous to claim that armv1-v7 vs armv8 represents ISA fragmentation. More legitimate (and fairly self inflicted) is the intel and the semi-random deployment of avx, etc.
vodou 4y ago
"For example, the once-popular DEC VAX, DEC Alpha, and Sun SPARC ISAs are extinct." Extinct is a bit harsh. SPARC V8 is alive and kicking in the space industry as the LEON processor models. Not the most performant CPU, but good enough for a lot of applications.
- rwmj 4y ago
  It would imprudent to base a whole new business on the SPARC ISA (if that's even possible - would Oracle license it?), unless you were in some niche like the space industry where interoperability is important. And David is quite right that the Digital architectures are really extinct. [Now I wonder if Compaq would open source the IP of Alpha. I think they licensed a lot of it to Intel so sadly they probably still make a bit of money from patents there.]
  - opencl 4y ago
    SPARC has been "open" for several decades at this point. All of the ISA specifications are freely available[1] and you can buy a license to manufacture SPARC hardware for $99. [1] https://sparc.org/technical-documents/ https://sparc.org/technical-documents/
  - kragen 4y ago
    https://en.wikipedia.org/wiki/DEC_Alpha https://en.wikipedia.org/wiki/DEC_Alpha says compaq sold "all Alpha intellectual property" to intel in 02001, so probably they haven't made any money from it since then also though any us patent filed since 01995 expires 20 years after filing, so any us αxp patents filed in the period 01995–02001 have expired already, and us patents filed before 01995 expire 17 years after issuance, so we only have to worry about pre-01995 us patents if first dec and then compaq and intel managed to delay their issuance, lemelson-style, until 02005 or later, which seems extremely unlikely so αxp is almost certainly already open-source as an isa dec's implementations are i suppose not open-source though, and casual glancing around on opencores and the like doesn't turn up any obvious candidates
- jabl 4y ago
  The people behind LEON have a next generation family of cores NOEL which are RISV-V based. For the time being they are still selling LEON but I'd expect it to be obsoleted at some point.
  - sitkack 4y ago
    This is really cool! https://www.gaisler.com/index.php/products/processors/noel-v https://www.gaisler.com/index.php/products/processors/noel-v
naasking 4y ago
These aren't fallacies. I guess the usual "Top five myths" was too overused as a title?
- willnonya 4y ago
  well, the answers tend to embrace a few fallacies so maybes thats what they meant.
Laaas 4y ago
No. 6: How can an ISA "be closed"? It seems to me like all ISAs are open by definition, since they are simply an interface between the CPU and human (essentially). See e.g. https://en.wikipedia.org/wiki/List_of_x86_manufacturers https://en.wikipedia.org/wiki/List_of_x86_manufacturers. Many of these haven't acquired any "license" from Intel to implement a processor that complies to the x86 interface. If things are as I assume, I find it disingenuous to portray RISC-V as an "open" ISA since it implies people can't build on top of other ISAs. If there's something more to it, it would be helpful if someone could explain. I still think RISC-V is superior to x86(-64) FWIW.
- afiori 4y ago
  You cannot sell a ARM processor that implements custom instructions. Or more properly to sell* ARM processors (or devices with ARM processors) you need a license from ARM where they give you permissions and (unless you are Apple) ARM is not going to give you a license where custom instructions are not forbidden. RISC-V is open in the sense that you do not need permission from the IP owner to use it. * I believe that you probably can manufacture custom ARM processors without a if you never plan to sell them.
  - klelatti 4y ago
    I think its possible that you could sell this processor but that you can't sell it as an Arm processor - in the recent Arm / Qualcomm case Arm seemed to rely heavily on trademark infringement rather than Qualcomm infringing their IP wrt the ISA. Happy to be corrected though if I've got this wrong.
  - CalChris 4y ago
    > You cannot sell a ARM processor that implements custom instructions. That was the case if you are an ARM architecture licensee. And of late, that's not even the case anymore. https://www.arm.com/technologies/custom-instructions https://www.arm.com/technologies/custom-instructions
- ndsipa_pomu 4y ago
  It's complicated by patents for some of the low level implementations. Typically, you'd license x86 from Intel if you want to build an x86 chip, whereas there are no licenses/patents around RISC-V as far as I know. The huge advantage of RISC-V is the lowered barrier to entry for chip manufacturers. It's now becoming possible for tech companies to make their own processors for their specific hardware and requirements. In my mind, the even bigger advantage is that the door is now open for open source RISC-V implementations, so that people/companies can work together on high performance/low energy CPUs. It seems such a waste of talent for different companies all spending time and effort to implement similar features in their chips, when they could be building and extending open source chip designs.
brookst 4y ago
I like RISC-V and hope it succeeds. But this article is disingenuous to the point of willful misrepresentation. It asserts that the only business risk to consider when choosing an ISA is whether the owner of the ISA might be sold to a new owner. Which I think might not even make the top 10 risks. Certainly it’s far less important than whether the ISA will reach critical mass for toolchain support, whether it will scale from small devices to datacenter applications, whether it will evolve and innovate, whether there is a pool of programmers with expertise, and so on. It’s bizarre to not even mention risks in these areas. I’m not saying RISC-V falls short in these things; I don’t know enough to have an opinion on that. But I do know they’re all more important than whether e.g. a hedge fund takes Intel private. Lots of other similar errors and omissions. This is evangelism from a True Believer, not a legit piece about RISC-V’s place in the industry.
- tenebrisalietum 4y ago
  A lot of those risks would be risks before there was Linux, GCC, and the plethora of open source tools. There's working RISC-V emulators running Linux with toolchains now, and there isn't even any real RISC-V hardware you can get yet beyond some demo boards, etc. > whether it will evolve and innovate CPU evolution and innovation is driven by the transistors beneath the ISA, not the ISA. For Intel, MMX/SEE and all the SIMD stuff was an innovation and something new, but the instructions that use it are really not that big of a deal in and of themselves. When some other capability arises in CPUs, any ISA will simply get new instructions.
  - brookst 4y ago
    Fair enough. For a startup considering committing to an ISA for low level development, would you say RISC-V is no riskier (har) than ARM? Despite, as you say, very little HW being available?
    - Gordonjcp 4y ago
      Realistically how many products are going to be reliant on the development team writing bare assembly?
PAPPPmAc 4y ago
I went and listened to the RISC V BoF at SC22 a month or so ago, mostly because I wanted to hear what their plans for making the proliferation of extensions manageable for toolchains and libraries looked like... The presented plan included quality choices like "vendor toolchains forked from upstream that support their secret sauce features, but suck at everything else compared to mainline, and are only maintained until until the vendor runs out of VC money," "You can always use the nearest base profile and ignore the special hardware you theoretically paid this vendor a fortune for" and special attention to "Most of your workloads are torch or something anyway, just use the vendor's binaries." I was not impressed.
- zozbot234 4y ago
  If a "secret sauce feature" turns out to be generally useful with interest from multiple implementers, it can be standardized as an official ISA extension. If it doesn't, you're no worse off than if you used literally any other bespoke ISA - except that your vendor toolchains will likely be higher quality if they're built on a shared RISC-V base.
klelatti 4y ago
> RISC-V cannot become the dominant ISA I absolutely want RISC-V to succeed but there is a difference between wanting it to succeed and wanting it to become 'dominant' or a 'lingua franca'. I find it a bit unnerving that to 'dominate' other ISAs is something that is being discussed.
- nwah1 4y ago
  If you believe in the ideal of open specifications, it is something that you root for. Probably at least half the desire for dominance is coming from that motivation. The other half from normal profit motive. I guarantee that Intel and Qualcomm have the same aspirations of dominance.
  - klelatti 4y ago
    I have to admit that I was a bit taken aback by that section. I don't think I've heard an open source software project (even linux) articulate a desire to dominate other projects. The principle of open source yes, but not a particular implementation. Also an ISA is quite different to a networking protocol where inter-operability is key.
    - nwah1 4y ago
      I have seen a desire for dominance from Linux and Android users. From Rust users. From Typescript users. And from Gnome and KDE users. The developers generally don't feed into the "flame wars" but it is a long tradition in the open source world to pick a piece of software and viciously attack the merits and users of competing software. This also happens with users of proprietary software, as with Apple cultists and so on. It is just human nature.
- legulere 4y ago
  I think the ethernet example from the article showcases pretty well why you would want RISC-V to become a lingua franca.
  - 4y ago
Dwedit 4y ago
The creators of Zig did something interesting to make it independent of a particular ISA: Use WebAssembly as an intermediate language form. (For bootstrapping of the compiler itself only) But it's not just that, it's the use of an intermediate language in general. When you use an intermediate language, you are less tied to one specific architecture, and are instead tied to the existence of a JIT, Recompiler, or Interpreter for that intermediate language being available on the target architecture.
- puffoflogic 4y ago
  What does this comment have to do with the article? And what does it have to do with Zig? ILs are standard for all compiled languages; they are taught in compilers 101.
  - Dwedit 4y ago
    Point #2 in the article is all about having your code become too dependent on a particular architecture that was popular at the time, but later goes out of favor. This has to do with shipping an executable in native code vs shipping something that isn't directly executable, because it's in an intermediate language instead of native code.
xiphias2 4y ago
RISC-V wins just like Linux won in number of installations: it is already the default for new developments where x86 and ARM is too expensive / hard to use (which is most embedded processors).

83 points by JoachimS 4y ago

Fairly generic fluff piece, but it’s fairly disingenuous to claim that armv1-v7 vs armv8 represents ISA fragmentation. More legitimate (and fairly self inflicted) is the intel and the semi-random deployment of avx, etc.

vodou 4y ago

"For example, the once-popular DEC VAX, DEC Alpha, and Sun SPARC ISAs are extinct." Extinct is a bit harsh. SPARC V8 is alive and kicking in the space industry as the LEON processor models. Not the most performant CPU, but good enough for a lot of applications.

rwmj 4y ago
It would imprudent to base a whole new business on the SPARC ISA (if that's even possible - would Oracle license it?), unless you were in some niche like the space industry where interoperability is important. And David is quite right that the Digital architectures are really extinct. [Now I wonder if Compaq would open source the IP of Alpha. I think they licensed a lot of it to Intel so sadly they probably still make a bit of money from patents there.]
- opencl 4y ago
  SPARC has been "open" for several decades at this point. All of the ISA specifications are freely available[1] and you can buy a license to manufacture SPARC hardware for $99. [1] https://sparc.org/technical-documents/ https://sparc.org/technical-documents/
- kragen 4y ago
  https://en.wikipedia.org/wiki/DEC_Alpha https://en.wikipedia.org/wiki/DEC_Alpha says compaq sold "all Alpha intellectual property" to intel in 02001, so probably they haven't made any money from it since then also though any us patent filed since 01995 expires 20 years after filing, so any us αxp patents filed in the period 01995–02001 have expired already, and us patents filed before 01995 expire 17 years after issuance, so we only have to worry about pre-01995 us patents if first dec and then compaq and intel managed to delay their issuance, lemelson-style, until 02005 or later, which seems extremely unlikely so αxp is almost certainly already open-source as an isa dec's implementations are i suppose not open-source though, and casual glancing around on opencores and the like doesn't turn up any obvious candidates
jabl 4y ago
The people behind LEON have a next generation family of cores NOEL which are RISV-V based. For the time being they are still selling LEON but I'd expect it to be obsoleted at some point.
- sitkack 4y ago
  This is really cool! https://www.gaisler.com/index.php/products/processors/noel-v https://www.gaisler.com/index.php/products/processors/noel-v

naasking 4y ago

These aren't fallacies. I guess the usual "Top five myths" was too overused as a title?

willnonya 4y ago
well, the answers tend to embrace a few fallacies so maybes thats what they meant.

Laaas 4y ago

No. 6: How can an ISA "be closed"? It seems to me like all ISAs are open by definition, since they are simply an interface between the CPU and human (essentially). See e.g. https://en.wikipedia.org/wiki/List_of_x86_manufacturers https://en.wikipedia.org/wiki/List_of_x86_manufacturers. Many of these haven't acquired any "license" from Intel to implement a processor that complies to the x86 interface. If things are as I assume, I find it disingenuous to portray RISC-V as an "open" ISA since it implies people can't build on top of other ISAs. If there's something more to it, it would be helpful if someone could explain. I still think RISC-V is superior to x86(-64) FWIW.

afiori 4y ago
You cannot sell a ARM processor that implements custom instructions. Or more properly to sell* ARM processors (or devices with ARM processors) you need a license from ARM where they give you permissions and (unless you are Apple) ARM is not going to give you a license where custom instructions are not forbidden. RISC-V is open in the sense that you do not need permission from the IP owner to use it. * I believe that you probably can manufacture custom ARM processors without a if you never plan to sell them.
- klelatti 4y ago
  I think its possible that you could sell this processor but that you can't sell it as an Arm processor - in the recent Arm / Qualcomm case Arm seemed to rely heavily on trademark infringement rather than Qualcomm infringing their IP wrt the ISA. Happy to be corrected though if I've got this wrong.
- CalChris 4y ago
  > You cannot sell a ARM processor that implements custom instructions. That was the case if you are an ARM architecture licensee. And of late, that's not even the case anymore. https://www.arm.com/technologies/custom-instructions https://www.arm.com/technologies/custom-instructions
ndsipa_pomu 4y ago
It's complicated by patents for some of the low level implementations. Typically, you'd license x86 from Intel if you want to build an x86 chip, whereas there are no licenses/patents around RISC-V as far as I know. The huge advantage of RISC-V is the lowered barrier to entry for chip manufacturers. It's now becoming possible for tech companies to make their own processors for their specific hardware and requirements. In my mind, the even bigger advantage is that the door is now open for open source RISC-V implementations, so that people/companies can work together on high performance/low energy CPUs. It seems such a waste of talent for different companies all spending time and effort to implement similar features in their chips, when they could be building and extending open source chip designs.

brookst 4y ago

I like RISC-V and hope it succeeds. But this article is disingenuous to the point of willful misrepresentation. It asserts that the only business risk to consider when choosing an ISA is whether the owner of the ISA might be sold to a new owner. Which I think might not even make the top 10 risks. Certainly it’s far less important than whether the ISA will reach critical mass for toolchain support, whether it will scale from small devices to datacenter applications, whether it will evolve and innovate, whether there is a pool of programmers with expertise, and so on. It’s bizarre to not even mention risks in these areas. I’m not saying RISC-V falls short in these things; I don’t know enough to have an opinion on that. But I do know they’re all more important than whether e.g. a hedge fund takes Intel private. Lots of other similar errors and omissions. This is evangelism from a True Believer, not a legit piece about RISC-V’s place in the industry.

tenebrisalietum 4y ago
A lot of those risks would be risks before there was Linux, GCC, and the plethora of open source tools. There's working RISC-V emulators running Linux with toolchains now, and there isn't even any real RISC-V hardware you can get yet beyond some demo boards, etc. > whether it will evolve and innovate CPU evolution and innovation is driven by the transistors beneath the ISA, not the ISA. For Intel, MMX/SEE and all the SIMD stuff was an innovation and something new, but the instructions that use it are really not that big of a deal in and of themselves. When some other capability arises in CPUs, any ISA will simply get new instructions.
- brookst 4y ago
  Fair enough. For a startup considering committing to an ISA for low level development, would you say RISC-V is no riskier (har) than ARM? Despite, as you say, very little HW being available?
  - Gordonjcp 4y ago
    Realistically how many products are going to be reliant on the development team writing bare assembly?

PAPPPmAc 4y ago

I went and listened to the RISC V BoF at SC22 a month or so ago, mostly because I wanted to hear what their plans for making the proliferation of extensions manageable for toolchains and libraries looked like... The presented plan included quality choices like "vendor toolchains forked from upstream that support their secret sauce features, but suck at everything else compared to mainline, and are only maintained until until the vendor runs out of VC money," "You can always use the nearest base profile and ignore the special hardware you theoretically paid this vendor a fortune for" and special attention to "Most of your workloads are torch or something anyway, just use the vendor's binaries." I was not impressed.

zozbot234 4y ago
If a "secret sauce feature" turns out to be generally useful with interest from multiple implementers, it can be standardized as an official ISA extension. If it doesn't, you're no worse off than if you used literally any other bespoke ISA - except that your vendor toolchains will likely be higher quality if they're built on a shared RISC-V base.

klelatti 4y ago

> RISC-V cannot become the dominant ISA I absolutely want RISC-V to succeed but there is a difference between wanting it to succeed and wanting it to become 'dominant' or a 'lingua franca'. I find it a bit unnerving that to 'dominate' other ISAs is something that is being discussed.

nwah1 4y ago
If you believe in the ideal of open specifications, it is something that you root for. Probably at least half the desire for dominance is coming from that motivation. The other half from normal profit motive. I guarantee that Intel and Qualcomm have the same aspirations of dominance.
- klelatti 4y ago
  I have to admit that I was a bit taken aback by that section. I don't think I've heard an open source software project (even linux) articulate a desire to dominate other projects. The principle of open source yes, but not a particular implementation. Also an ISA is quite different to a networking protocol where inter-operability is key.
  - nwah1 4y ago
    I have seen a desire for dominance from Linux and Android users. From Rust users. From Typescript users. And from Gnome and KDE users. The developers generally don't feed into the "flame wars" but it is a long tradition in the open source world to pick a piece of software and viciously attack the merits and users of competing software. This also happens with users of proprietary software, as with Apple cultists and so on. It is just human nature.
legulere 4y ago
I think the ethernet example from the article showcases pretty well why you would want RISC-V to become a lingua franca.
- 4y ago

Dwedit 4y ago

The creators of Zig did something interesting to make it independent of a particular ISA: Use WebAssembly as an intermediate language form. (For bootstrapping of the compiler itself only) But it's not just that, it's the use of an intermediate language in general. When you use an intermediate language, you are less tied to one specific architecture, and are instead tied to the existence of a JIT, Recompiler, or Interpreter for that intermediate language being available on the target architecture.

puffoflogic 4y ago
What does this comment have to do with the article? And what does it have to do with Zig? ILs are standard for all compiled languages; they are taught in compilers 101.
- Dwedit 4y ago
  Point #2 in the article is all about having your code become too dependent on a particular architecture that was popular at the time, but later goes out of favor. This has to do with shipping an executable in native code vs shipping something that isn't directly executable, because it's in an intermediate language instead of native code.

xiphias2 4y ago

RISC-V wins just like Linux won in number of installations: it is already the default for new developments where x86 and ARM is too expensive / hard to use (which is most embedded processors).