GNU/blobcat

RISC-V fucking when

And for fuck's sake you lazy ass SoC designers would you please write some goddamn specifications for booting and device enumeration before trying to ship a revolutionary new ISA with fucking u-Boot and device trees

@sir Or for that matter do something like adopt Open Firmware? I mean, it already exists, has a *very* decent track record over the years, I don't see a particular reason why it isn't more widespread beyond NIH/wanting to lock down systems for no good reason
replies
1
announces
3
likes
4

@sir Designing a SoC is hard and expensive (including the tools, often proprietary for the lack of choice, and the hardware designers are way scarcer than web-developers for example). Making it produced is even harder. Not only because this is objectively expensive, but also there is quite a lot of monopolization, and every company tries to secure the status quo. So you can't even buy what you need without going through the hell of negotiations.

@allison @sir yes, that's the reason

@sir RISC-V is great and all, but it's (intentionally) not a complete spec. It was designed so you could make some real innovative stuff.

To make RISC-V processors that can go toe to toe with x86 and ARM, there needs to be an additional spec that standardizes RISC-V for general purpose processors.

Maybe that will happen as someone starts making successful RISC-V processors, becoming the de-facto standard, maybe it'll be set by a standards body. Either way it still needs to be done.

@urusan @sir what about PowerPC? IBM just released their POWER10 architecture, but their POWER9 architecture isn't a slouch either.

@RyuKurisu @urusan the PPC ABI gives me nightmares