5 ms·

What's the reason behind companies adopting ARM instead of the open standard RISC-V? E.g. Apple designs its own chips, wouldn't it make more sense to use RISC-V

What's the reason behind companies adopting ARM instead of the open standard RISC-V? E.g. Apple designs its own chips, wouldn't it make more sense to use RISC-V instead of paying fees to ARM?

ArgyleSound 6y ago
Presumably because ARM’s ISA is more advanced and costs Apple basically nothing (I’ve heard it might be as low as six or seven figures for their architectural license).
- raghava 6y ago
  Not judging, IMO but MAGAF* can really afford to pump money into enhancing RISC-V - individually or via a consortium, and stand to benefit together, while developing the whole ecosystem too. The early days of Apple and MS were sorta on these lines anyway - which caused an explosion of growth and wealth. Not able to reason why this isn't happening now! *Microsoft, Apple, Google, Amazon, FB [Edit 1]: A quick search throws up that Google, IBM, SONY etc are already members of RISC-V foundation. Also, https://riscv.org/2018/08/eet-asia-article-microsoft-and-google-planning-silicon-level-security/ https://riscv.org/2018/08/eet-asia-article-microsoft-and-goo...
  - saagarjha 6y ago
    But why should they?
    - raghava 6y ago
      Cause, they ... can? Why would a nation choose to go to the moon, or build something otherworldy?
mumblemumble 6y ago
Presumably it's because ARM is a fully-baked technology that's been around for a while, with good compilers and all that, right now. RISC-V is promising, but it's not there yet. A small company like Pine64 isn't necessarily going to have the resources to, for example, get the Linux kernel support for RISC-V into a production-ready state. Apple, for its part, adopted ARM years before the RISC-V project even started.
simonh 6y ago
Apple helped found ARM back in 1990 and is a major shareholder, they have a massive investment in the architecture, including enormous experience with it. They can't just drop the massive experience, knowledge, investment and hardware and software ecosystem they have built up around ARM and switch to a completely different architecture. Also Apple already hold a perpetual license for the ARM architecture, the major costs of their investment in it were spent years or even decades ago. The obvious counter is the sunk cost fallacy, but what does RISC-V actually have to offer that Apple doesn't already have with ARM? Those sunk costs have bought Apple actual tangible benefits on ARM that RISC-V doesn't have, such as architectural features like the secure enclave, better device integration, better GPU options, more mature device driver support, a stronger development and testing tools ecosystem. The opportunity for RISC-V is new use cases not already strongly addressed by ARM, or where ARMs more mature ecosystem offers weaker benefits, for companies not already strongly invested in ARM.
- mumblemumble 6y ago
  In this case, I think that the "sunk cost fallacy" counter is as specious as it is obvious. It completely misplaces the burden of evidence. Apple has a platform with hundreds of millions of active users, perhaps over a billion actual devices in use. In a situation like that, conservatism is warranted. The guiding principle is simple: If it ain't broke, don't fix it. Meaning that the argument doesn't start with any sort of discussion about technical advantages that RISC-V might have over ARM, or challenging reasons to stay on ARM as if, having dismissed them, the obvious default plan of migrating to RISC-V would kick into action. It starts with trying to demonstrate that something about ARM is broken. Because, if it ain't broke, don't fix it. So. . . is it hindering Apple's strategic or operational priorities in any way?
microcolonel 6y ago
I think a big part of the issue with these things is that porting popular applications is not a high priority in the industry. I pitched porting v8 to RISC-V to a few companies a couple years back, and it seems like they were not interested in that. Getting ATX (or other standardized) form-factor boards with chips big enough to test applications on, in my opinion, should be a high priority. As much as people talk about "post-PC" or whatever, PC-form-factor computers are the center of development and understanding, especially for server applications. Being able to buy or use existing power supplies, chassis, disks, NICs, and GPUs (and maybe DIMMs, if that's not a massive engineering challenge) and put together a "PC" I think is really the key to development of the architecture, and they're a portal to applications like Windows laptops (see Qualcomm AArch64 laptops) and Chromebooks (big part of what I was thinking of when I begun and abandoned porting V8). If there were ATX form-factor AArch64 boards available for less than a thousand bucks early on, it would have hit Android and laptops possibly a year or two earlier, with significantly less specific investment from ARM Holdings.
repiret 6y ago
Apple didn't design its own cores when they started using Arm for iPhone, and RISC-V wasn't especially mature then either. There's a cost to switching, and you can bet that Arm designs its business relationship with Apple to make sure Apple continues to prefer Arm.
Followerer 6y ago
This is corporate mindset 101. Questions like "why don't you do X" make no sense to a corporation. The question is always "why should I do X". Every decision has a cost and a benefit, and things happen when benefits outweight the costs proportionally to the implicit risk. Companies have been successfully working with ARM for a loooong time. It's a massive, well supported, ecosystem that works quite well (no customer complains about delayed/cancelled products related to ARM's performance). What does RISC-V offer that's worth jumping into the void? That's startup territory, you won't see big companies betting big on RISC-V for a while, if ever.
coreblocks 6y ago
They have a lot of experience designing ARM chips that they're not just gonna throw out. And obviously the major advantage of going ARM is that they can finally unify IOS and Mac app development. In fact as far as I know, all IOS apps will be natively executable on Apple Silicon for Mac on day one. I think RISC-V will eventually come to consumer facing (rather than embedded) products, but it might take another 10 years.

What's the reason behind companies adopting ARM instead of the open standard RISC-V? E.g. Apple designs its own chips, wouldn't it make more sense to use RISC-V instead of paying fees to ARM?

ArgyleSound 6y ago

Presumably because ARM’s ISA is more advanced and costs Apple basically nothing (I’ve heard it might be as low as six or seven figures for their architectural license).

raghava 6y ago
Not judging, IMO but MAGAF* can really afford to pump money into enhancing RISC-V - individually or via a consortium, and stand to benefit together, while developing the whole ecosystem too. The early days of Apple and MS were sorta on these lines anyway - which caused an explosion of growth and wealth. Not able to reason why this isn't happening now! *Microsoft, Apple, Google, Amazon, FB [Edit 1]: A quick search throws up that Google, IBM, SONY etc are already members of RISC-V foundation. Also, https://riscv.org/2018/08/eet-asia-article-microsoft-and-google-planning-silicon-level-security/ https://riscv.org/2018/08/eet-asia-article-microsoft-and-goo...
- saagarjha 6y ago
  But why should they?
  - raghava 6y ago
    Cause, they ... can? Why would a nation choose to go to the moon, or build something otherworldy?

mumblemumble 6y ago

Presumably it's because ARM is a fully-baked technology that's been around for a while, with good compilers and all that, right now. RISC-V is promising, but it's not there yet. A small company like Pine64 isn't necessarily going to have the resources to, for example, get the Linux kernel support for RISC-V into a production-ready state. Apple, for its part, adopted ARM years before the RISC-V project even started.

simonh 6y ago

Apple helped found ARM back in 1990 and is a major shareholder, they have a massive investment in the architecture, including enormous experience with it. They can't just drop the massive experience, knowledge, investment and hardware and software ecosystem they have built up around ARM and switch to a completely different architecture. Also Apple already hold a perpetual license for the ARM architecture, the major costs of their investment in it were spent years or even decades ago. The obvious counter is the sunk cost fallacy, but what does RISC-V actually have to offer that Apple doesn't already have with ARM? Those sunk costs have bought Apple actual tangible benefits on ARM that RISC-V doesn't have, such as architectural features like the secure enclave, better device integration, better GPU options, more mature device driver support, a stronger development and testing tools ecosystem. The opportunity for RISC-V is new use cases not already strongly addressed by ARM, or where ARMs more mature ecosystem offers weaker benefits, for companies not already strongly invested in ARM.

mumblemumble 6y ago
In this case, I think that the "sunk cost fallacy" counter is as specious as it is obvious. It completely misplaces the burden of evidence. Apple has a platform with hundreds of millions of active users, perhaps over a billion actual devices in use. In a situation like that, conservatism is warranted. The guiding principle is simple: If it ain't broke, don't fix it. Meaning that the argument doesn't start with any sort of discussion about technical advantages that RISC-V might have over ARM, or challenging reasons to stay on ARM as if, having dismissed them, the obvious default plan of migrating to RISC-V would kick into action. It starts with trying to demonstrate that something about ARM is broken. Because, if it ain't broke, don't fix it. So. . . is it hindering Apple's strategic or operational priorities in any way?

microcolonel 6y ago

I think a big part of the issue with these things is that porting popular applications is not a high priority in the industry. I pitched porting v8 to RISC-V to a few companies a couple years back, and it seems like they were not interested in that. Getting ATX (or other standardized) form-factor boards with chips big enough to test applications on, in my opinion, should be a high priority. As much as people talk about "post-PC" or whatever, PC-form-factor computers are the center of development and understanding, especially for server applications. Being able to buy or use existing power supplies, chassis, disks, NICs, and GPUs (and maybe DIMMs, if that's not a massive engineering challenge) and put together a "PC" I think is really the key to development of the architecture, and they're a portal to applications like Windows laptops (see Qualcomm AArch64 laptops) and Chromebooks (big part of what I was thinking of when I begun and abandoned porting V8). If there were ATX form-factor AArch64 boards available for less than a thousand bucks early on, it would have hit Android and laptops possibly a year or two earlier, with significantly less specific investment from ARM Holdings.

repiret 6y ago

Apple didn't design its own cores when they started using Arm for iPhone, and RISC-V wasn't especially mature then either. There's a cost to switching, and you can bet that Arm designs its business relationship with Apple to make sure Apple continues to prefer Arm.

Followerer 6y ago

This is corporate mindset 101. Questions like "why don't you do X" make no sense to a corporation. The question is always "why should I do X". Every decision has a cost and a benefit, and things happen when benefits outweight the costs proportionally to the implicit risk. Companies have been successfully working with ARM for a loooong time. It's a massive, well supported, ecosystem that works quite well (no customer complains about delayed/cancelled products related to ARM's performance). What does RISC-V offer that's worth jumping into the void? That's startup territory, you won't see big companies betting big on RISC-V for a while, if ever.

coreblocks 6y ago

They have a lot of experience designing ARM chips that they're not just gonna throw out. And obviously the major advantage of going ARM is that they can finally unify IOS and Mac app development. In fact as far as I know, all IOS apps will be natively executable on Apple Silicon for Mac on day one. I think RISC-V will eventually come to consumer facing (rather than embedded) products, but it might take another 10 years.