4 ms·

So RISC-V CPU designers should aim for the developer market then? First win the hearts and minds of developers, then everything else will come.

epx 8y ago
Absolutely yes!
dkersten 8y ago
I guess if nobody wants to develop for it, then it becomes a more difficult business proposition too.
amq 8y ago
But if there is nothing to develop for... Chicken and egg problem. However, I think that Windows on ARM will soon become a real thing, this could bring the developers, it will also help the ARM devices spread among Linux users. Regarding what Linus wrote, I do not buy it. There is so much stuff can run without cross-compilation: js and other scripts, java, c#. Also look at things like Android and iOS, you are already happily developing for ARM, you can develop for servers with the same approach.
- rjsw 8y ago
  I do buy Linus' argument, you need to be able to do native development. My ARM64 development machine is a Pinebook, ARM32 a Cubietruck. Both have only 2GB of RAM and could really do with more. At least the Cubietruck has SATA so mine has a SSD.
  - scarface74 8y ago
    iOS developers develop on x86 computers and deploy to ARM computers all of the time. But, most developers don’t need to do native development.
  - sam_lowry_ 8y ago
    I am pretty happy to develop on ODROID-C2 as well. But we are a minority.
  - pjmlp 8y ago
    With rich language runtimes, as app developer I don't care if my application instance runs on top of an OS, hypervisor or just plain bare metal. The only devs that need to care about the native part are the ones writing the glue between runtime and hardware.
    - rjsw 8y ago
      Somebody has to maintain the runtimes, they are not typically set up to be cross-compilable, I'm doing JVM development on my ARM systems right now.
      - pjmlp 8y ago
        Apparently you stop reading at the first sentence.
        rjsw 8y ago
        I'm not working on "glue between runtime and hardware".
        pjmlp 8y ago
        If you are maintaining a JVM runtime for ARM or writing JNI stuff then it looks to me you are doing just that. Otherwise you completely lost me there.
- deng 8y ago
  > However, I think that Windows on ARM will soon become a real thing, this could bring the developers, it will also help the ARM devices spread among Linux users. Windows an ARM has been a real thing for years, and it is horrible. Nobody in his right mind would use this for development.
  - mgamache 8y ago
    Windows ARM (Windows RT) has been a mess, but Microsoft is committing more resources to make proper Windows + ARM it a real thing. Will it work? I don't know, but they are trying, probably because of Apple & AWS moves toward ARM.
  - pjc50 8y ago
    Only because Microsoft deliberately chose to cripple it, a decision which they are gradually reversing.
  - WorldMaker 8y ago
    Windows 10 on ARM has come a long way in recent builds. The SKUs have many fewer differences from x86/x64 Windows 10. On top of that, x86 emulation on it is now stable, including full Win32 emulation, and reports are coming in of even doing things like running Steam and old PC games on ARM devices.
nostrademons 8y ago
No. It's unlikely that they'd be able to anyway - products (like, say, a Macbook Pro or ThinkPad) get targeted at developers. RISC-V is an instruction architecture, which gets used by a chip designer, which gets sold to a computer manufacturer, who does the actual marketing to end users, so they're about 4 links removed up the value chain. What they should do is make RISC-V useful for computing at the edge - applying computation to areas where it's previously been infeasible (eg. smartwatches, wearables, RFID, drones, vehicles, etc). Developers go where the end users are. If a new end-user market opens up that suddenly needs a lot of software, developers will buy the same ISA that their customers are using. And then there will be strong pressure to run the other software that developers write (eg. servers, as Linus notes) on the same ISA to prevent cross-development.
platform 8y ago
Yes, very much so. Having a C++, GO, Java, RUST, Python and F# IDEs running on Risc-V. Allowing to debug, experiment, without relying on centralized hosting providers -- would be a huge boon to Risc-V. Early adaptors will be developers, than their family/friends circle. Then there will be Risc-V-First movement (mirroring mobile-first movement), and there it will start... > So RISC-V CPU designers should aim for the developer market then? First win the hearts and minds of developers, then everything else will come.

epx 8y ago

Absolutely yes!

dkersten 8y ago

I guess if nobody wants to develop for it, then it becomes a more difficult business proposition too.

amq 8y ago

But if there is nothing to develop for... Chicken and egg problem. However, I think that Windows on ARM will soon become a real thing, this could bring the developers, it will also help the ARM devices spread among Linux users. Regarding what Linus wrote, I do not buy it. There is so much stuff can run without cross-compilation: js and other scripts, java, c#. Also look at things like Android and iOS, you are already happily developing for ARM, you can develop for servers with the same approach.

rjsw 8y ago
I do buy Linus' argument, you need to be able to do native development. My ARM64 development machine is a Pinebook, ARM32 a Cubietruck. Both have only 2GB of RAM and could really do with more. At least the Cubietruck has SATA so mine has a SSD.
- scarface74 8y ago
  iOS developers develop on x86 computers and deploy to ARM computers all of the time. But, most developers don’t need to do native development.
- sam_lowry_ 8y ago
  I am pretty happy to develop on ODROID-C2 as well. But we are a minority.
- pjmlp 8y ago
  With rich language runtimes, as app developer I don't care if my application instance runs on top of an OS, hypervisor or just plain bare metal. The only devs that need to care about the native part are the ones writing the glue between runtime and hardware.
  - rjsw 8y ago
    Somebody has to maintain the runtimes, they are not typically set up to be cross-compilable, I'm doing JVM development on my ARM systems right now.
    - pjmlp 8y ago
      Apparently you stop reading at the first sentence.
      - rjsw 8y ago
        I'm not working on "glue between runtime and hardware".
        pjmlp 8y ago
        If you are maintaining a JVM runtime for ARM or writing JNI stuff then it looks to me you are doing just that. Otherwise you completely lost me there.
deng 8y ago
> However, I think that Windows on ARM will soon become a real thing, this could bring the developers, it will also help the ARM devices spread among Linux users. Windows an ARM has been a real thing for years, and it is horrible. Nobody in his right mind would use this for development.
- mgamache 8y ago
  Windows ARM (Windows RT) has been a mess, but Microsoft is committing more resources to make proper Windows + ARM it a real thing. Will it work? I don't know, but they are trying, probably because of Apple & AWS moves toward ARM.
- pjc50 8y ago
  Only because Microsoft deliberately chose to cripple it, a decision which they are gradually reversing.
- WorldMaker 8y ago
  Windows 10 on ARM has come a long way in recent builds. The SKUs have many fewer differences from x86/x64 Windows 10. On top of that, x86 emulation on it is now stable, including full Win32 emulation, and reports are coming in of even doing things like running Steam and old PC games on ARM devices.

nostrademons 8y ago

No. It's unlikely that they'd be able to anyway - products (like, say, a Macbook Pro or ThinkPad) get targeted at developers. RISC-V is an instruction architecture, which gets used by a chip designer, which gets sold to a computer manufacturer, who does the actual marketing to end users, so they're about 4 links removed up the value chain. What they should do is make RISC-V useful for computing at the edge - applying computation to areas where it's previously been infeasible (eg. smartwatches, wearables, RFID, drones, vehicles, etc). Developers go where the end users are. If a new end-user market opens up that suddenly needs a lot of software, developers will buy the same ISA that their customers are using. And then there will be strong pressure to run the other software that developers write (eg. servers, as Linus notes) on the same ISA to prevent cross-development.

platform 8y ago

Yes, very much so. Having a C++, GO, Java, RUST, Python and F# IDEs running on Risc-V. Allowing to debug, experiment, without relying on centralized hosting providers -- would be a huge boon to Risc-V. Early adaptors will be developers, than their family/friends circle. Then there will be Risc-V-First movement (mirroring mobile-first movement), and there it will start... > So RISC-V CPU designers should aim for the developer market then? First win the hearts and minds of developers, then everything else will come.