this post was submitted on 28 Jul 2024
46 points (100.0% liked)
Technology
933 readers
56 users here now
A tech news sub for communists
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
On the other hand, there's value in exploring different approaches as well.
I don't think there are any benefits compared to RISC-V with custom instructions, maintaining your own toolchain is also extra work which could benefit everyone with RISC-V
The benefit is that you can make fairly drastic changes to the architecture, which they already have done incidentally, while with RISC-V the architecture is already mature. I agree there's a downside that there's a duplication of effort as a result. The way I look at it is that RISC-V is a good mature platform, but Loongson is more of an experimental architecture they can try different ideas with.
RISC-V only gives you the instruction set and standard, it does not tell you how to actually do it. The way you handle the microarchitecture internally is up to you.
It could just be that they're already invested in the architecture they built, and don't see much value in switching.
Fair, but I still think they should switch. I don't think it would be too difficult anyway. It should mostly amount to rewrite of the decoder as can be inferred from the fact that it is mostly a superset of MIPS64.
The perceived benefits, are being independent of the west on technology. Which RISC-V already provides.
There are chinese companies making RISC-V chips and machines: https://milkv.io/pioneer
I get the impression that LoongArch has more priority because they fully own that ISA which means foreign entities such as the US Empire can have 0 impact on it. If something happens to RISC-V that could put China at a disadvantage they still have LoongArch. If they go all in on RISC-V they could easily get fucked or bullied into submission, that's my view anyways.
There's no danger in using RISC-V -- if the official standard is compromised somehow, China can just make their own "fork". I think it's more likely just a business decision by Loongson. Unless it's replaced, it'll likely become an open standard eventually
Good point, I didn't think of that! Either way, I'm quite happy to see more general purpose RISCs emerge. At this point I'm just really sick of dealing with Intel's CISC crap lol.
I'd like to see more people contributing to RISC-V as well. I'm just saying they probably have their reasons for pursuing the current approach. I tend to give people benefit of the doubt because a lot of the time the devil is in the details. It'll be interesting to watch how this develops one way or the other I suspect.