RISC-V must get its messaging right on open standard vs. open source | Hacker News Reader

RISC-V must get its messaging right on open standard vs. open source

theregister.com

42 pointsiamflimflam17 comments

Reminds me of the late 1990s export bans on Intel and Apple products, which they both used in ads demonstrating how powerful they were.

Too band the article didn’t include more justification from the politicians because this reeks of lobbying. I think the legal challenges would stop such a ban almost immediately. Perhaps this legal battle is sorely needed to set precedent.

I’m ambivalent about RV. I see why semi manufacturers like to use it for deeply embedded cores that replace M0.

> I see why semi manufacturers like to use it for deeply embedded cores that replace M0.

It's not just a matter of not having to pay for a C-M0 license.

- minimal RV32E cores (16 registers) are significantly smaller than an M0, but not slower

- if you want 32 registers (RV32I) you can have it. Arm makes you go to full on Aarch64, with MMU, FPU, NEON etc etc

- if you want no multiplier, you can have it

- if you want multiply AND divide (C-M0 has no divide), you can have that

- if you want CLZ or Popcount or anything like that, you can have it (Zbb standard extension). Arm makes you go to M3 for `clz` or Aarch64 for `cnt`

- if you want a 64 bit C-M0 equiv you can have it. Arm, obviously, makes you go to full on Aarch64 with MMU, FPU, NEON etc for that

In short, the huge advantage of RISC-V is flexibility. Arm tries to fit your needs into one of a very small number of buckets that are seldom Goldilocks for you.

Everything you said are facts.

In (my) reality, the level of detail you describe in the configuration is based on reading a spec, and not based on a design. Caveat: this is my experience evaluating RV v. Arm. In other words: no one is picking apart the requirements to that level. In fact, they are just shoveling NEW options into the ISA: look at T-Head, they've added like two dozen esoteric configs. That's a really outsized (and strange) influence IMHO.

I think the issues with RV adoption aren't the ISA, it's everything else a semi needs to build a product, specifically debug, validation, integration, and ecosystem : the standards are still in their infancy (go read the debug standard, it is severely underbaked), and I don't think there are any IP vendors selling RV IP, are there? Just those two vectors alone are a big deal.

The ecosystem is growing, its just not there yet. And I think Arm is aware that it IS only a matter of time. It is inevitable that the ecosystem vendors for RV will mature over time. I worry it is being polluted by special interest, because RV is too scared to say "no" right now.

RV looks leaderless right now. And that is by design. And I think that is bad news.

Good article. This is why RISC-V International is based in Switzerland.

I imagine the EU wouldn't be too happy about this either given how much they are backing RISC-V.

Earlier discussion: https://news.ycombinator.com/item?id=40512028

Non-copyleft open source licenses have the same property of being proprietarizable. The distinction here isn't quite the one the authors are claiming.

It's worth remembering that RISC-V International was founded in the USA and subsequently moved to Switzerland citing "concerns over U.S. trade regulations"[1].

Precedent won't be kind to arguments that RISC-V should not be subject to western sanctions.

[1]: https://en.wikipedia.org/wiki/RISC-V#RISC-V_Foundation_and_R...

Reply on news.ycombinator.com