4 ms·

Except that doesn't quite explain why soo many developers use macbooks (OS/X), yet deploy on Linux. Personally I think ARM could be leapfrogged by RISC-V, thou

Except that doesn't quite explain why soo many developers use macbooks (OS/X), yet deploy on Linux. Personally I think ARM could be leapfrogged by RISC-V, though it will take longer than people predict.

lugg 8y ago
The ops guys use Linux.
- rbanffy 8y ago
  After SSH was invented, I can use anything (my corporate laptop is a Mac, my personal ones are Linux-running x86's). To be fair, I was using anything when telnet was not considered suicidal.
AnIdiotOnTheNet 8y ago
I'm curious why you believe that. Aside from the hype and the association with the word "open", I honestly can't think of anything interesting or important about RISC-V.
- rbanffy 8y ago
  It'll be big in embedded, where IP licensing is a significant part of the end price.
- jacknews 8y ago
  There seem to be a few indications that there is opportunity for change, and RISC-V could make gains, eg https://www.nextplatform.com/2019/02/05/the-era-of-general-purpose-computers-is-ending/ https://www.nextplatform.com/2019/02/05/the-era-of-general-p... https://www.sigarch.org/simd-instructions-considered-harmful/ https://www.sigarch.org/simd-instructions-considered-harmful...
yaantc 8y ago
> ARM could be leapfrogged by RISC-V For the low-end deeply embedded space, I could see RISC V becoming a significant player in the next 5 years. ARM is relatively expensive there (it's cheap compared to Intel for servers, but at the low end it's more the IBM of Microsoft of old), and there's less need for their ecosystem (most 3rd party software is open source). The Cortex M are less flexible then some competitors in their configuration, and designing good small CPU IP is doable at the low end. So all the small custom designers (Andes, BA semi, Cortus...) can rally around RISC V and have cheap and flexible design, with a good shared software ecosystem. Today it's still one or the other. At the high-end it's very different. To be competitive one has to take advantage of the latest process nodes, and work well ahead of time with fabs and EDA vendors on the next node(s). This is highly labor intensive, and it takes deep pockets and a lot of resources. ARM is already there, and well entrenched. Getting enough money to replicate this on the RISC V side will take a loooooong while, if it ever happens. Unless a deep pocketed company decides to use RISC V and go their own way, but it seems very unlikely. Look at how even Qualcomm reduced its work on custom ARM cores for their high end, and now do tweaks on ARM designs. But personally I'm fine with this. It's mostly at the low end / embedded that I feel there's a need for more competition.
- petra 8y ago
  I agree that at the low-end , at least in low-end ip-cores, RISC-V could do well. But in microcontrollers ? To be meaningful there, they will have to convince one of the big western guys to bet on them. But why would any of them care , enough to hurt their future relationship with ARM ,a significant risk ?
gvb 8y ago
Except... 1. OS/X is Unix[tm] and linux is a "Unix clone" so the operating system works the same and (for purposes of this debate) both run on the X86 ISA. 2. The first thing all the macbook developers do is install Homebrew or equivalent and install all the same packages that are installed in their linux deployment environment. At that point, their OS/X development environment is effectively indistinguishable from their linux deployment environment.
- bunnycorn 8y ago
  What's that? The 2000's. The first thing we install is docker and run everything containerized. Homebrew is for tools that we actually want to run locally, like ffmpeg or python, or node...

lugg 8y ago

The ops guys use Linux.

rbanffy 8y ago
After SSH was invented, I can use anything (my corporate laptop is a Mac, my personal ones are Linux-running x86's). To be fair, I was using anything when telnet was not considered suicidal.

AnIdiotOnTheNet 8y ago

I'm curious why you believe that. Aside from the hype and the association with the word "open", I honestly can't think of anything interesting or important about RISC-V.

rbanffy 8y ago
It'll be big in embedded, where IP licensing is a significant part of the end price.
jacknews 8y ago
There seem to be a few indications that there is opportunity for change, and RISC-V could make gains, eg https://www.nextplatform.com/2019/02/05/the-era-of-general-purpose-computers-is-ending/ https://www.nextplatform.com/2019/02/05/the-era-of-general-p... https://www.sigarch.org/simd-instructions-considered-harmful/ https://www.sigarch.org/simd-instructions-considered-harmful...

yaantc 8y ago

> ARM could be leapfrogged by RISC-V For the low-end deeply embedded space, I could see RISC V becoming a significant player in the next 5 years. ARM is relatively expensive there (it's cheap compared to Intel for servers, but at the low end it's more the IBM of Microsoft of old), and there's less need for their ecosystem (most 3rd party software is open source). The Cortex M are less flexible then some competitors in their configuration, and designing good small CPU IP is doable at the low end. So all the small custom designers (Andes, BA semi, Cortus...) can rally around RISC V and have cheap and flexible design, with a good shared software ecosystem. Today it's still one or the other. At the high-end it's very different. To be competitive one has to take advantage of the latest process nodes, and work well ahead of time with fabs and EDA vendors on the next node(s). This is highly labor intensive, and it takes deep pockets and a lot of resources. ARM is already there, and well entrenched. Getting enough money to replicate this on the RISC V side will take a loooooong while, if it ever happens. Unless a deep pocketed company decides to use RISC V and go their own way, but it seems very unlikely. Look at how even Qualcomm reduced its work on custom ARM cores for their high end, and now do tweaks on ARM designs. But personally I'm fine with this. It's mostly at the low end / embedded that I feel there's a need for more competition.

petra 8y ago
I agree that at the low-end , at least in low-end ip-cores, RISC-V could do well. But in microcontrollers ? To be meaningful there, they will have to convince one of the big western guys to bet on them. But why would any of them care , enough to hurt their future relationship with ARM ,a significant risk ?

gvb 8y ago

Except... 1. OS/X is Unix[tm] and linux is a "Unix clone" so the operating system works the same and (for purposes of this debate) both run on the X86 ISA. 2. The first thing all the macbook developers do is install Homebrew or equivalent and install all the same packages that are installed in their linux deployment environment. At that point, their OS/X development environment is effectively indistinguishable from their linux deployment environment.

bunnycorn 8y ago
What's that? The 2000's. The first thing we install is docker and run everything containerized. Homebrew is for tools that we actually want to run locally, like ffmpeg or python, or node...