5 ms·
Loongson Technology develops its own CPU instruction set architecture
- nayuki 6y agoSounds like another China-only proprietary derivative technology that will go nowhere, kind of like https://en.wikipedia.org/wiki/Super_Video_CD#CVD https://en.wikipedia.org/wiki/Super_Video_CD#CVD , https://en.wikipedia.org/wiki/Enhanced_Versatile_Disc https://en.wikipedia.org/wiki/Enhanced_Versatile_Disc , https://en.wikipedia.org/wiki/China_Blue_High-definition_Disc https://en.wikipedia.org/wiki/China_Blue_High-definition_Dis... , https://en.wikipedia.org/wiki/Audio_Video_Standard https://en.wikipedia.org/wiki/Audio_Video_Standard
- mjevans 6y agoIf they had picked RISC-V then there's a strong chance of international support from ecosystems outside of China, such as open source compilers and operating systems. IIRC most of the advantages this architecture claims to have should also be addressed by the RISC-V specification (at least in baseline profile) and they're also free to add vendor specific extensions? Other than not-invented-here, a project already in the pipeline, or possible national security (for China) applications; I fail to see the advantage of yet another standard.
- snvzz 6y agoFortunately, RISC-V is really strong in China. This decision by one CPU vendor is unlikely to change this.
- rurounijones 6y ago> possible national security applications I wonder how much of a benefit that would be. To have an entirely separate instruction set and all the hardware / software that builds upon it that is only really used in your country _sounds_ like it would be a good way to frustrate espionage / hacking to a layman since adversaries lack the expertise. I am sure adversarial agencies would appoint a person or a small team to become experts in this specific technology though so maybe not so much although the costs would be more than for a "common" tech. Interesting theoretical situation to play with though.
- kube-system 6y agoA national security advantage could be a simple as: a foreign nation can’t cut off the supply.
- XorNot 6y agoBut it also means you can't seek extra supply from outside your borders. In reality this is going to be a hugely limited benefit: to be useful it has to be widely dispersed, probably internationally. Which means everyone is going to wind up with samples in hand, and realistically it closes no exploits (which usually operate higher then architecture level). It's in fact far more likely that some devastating architectural flaws are discovered.
- ClumsyPilot 6y ago"hugely limited benefit" Erm, no, independance of supply is the entire point. Multiple nations, such as Iran and Cuba, have seen their economies rekt by US sansctions or embargoes. This is literally priority N1 for any country on bad terms with US, whatever the reasons.
- XorNot 6y agoI meant in regards to going your own way on architecture. In-countrying your chip making just makes sense, but there's no reason not to use an existing architecture that experiences global cooperative development for as long as possible.
- geokon 6y agoPer this comment https://www.cnx-software.com/2021/04/17/loongson-loongarch-cpu-instruction-set-architecture/#comment-582068 https://www.cnx-software.com/2021/04/17/loongson-loongarch-c... It seems risc-v relies on a patent pool dominated by us companies. So the us can arbitrarily extrajudicially stomp on chinese chip manufacturer as they did with Huawei
- rwmj 6y agoCan you point to a "RISC-V patent pool"? I've never heard of such a thing. The ISA was shown to be patent free. Microarchitectures have heavily patented features, but microarchitectures are generally independent of the ISA, so changing ISA doesn't help.
- zibzab 6y ago> The ISA was shown to be patent free. Please stop repeating this. First of all, before winning a few court cases we can't really be sure. Also, probably well above 99% of patents in this area are not about the ISA but various implementation and optimization details. As fun as riscv is, the patent free statement is (at this point) just marketing talk by Berkley.
- snvzz 6y agoI knew FUD was going to show up sooner than latter.
- ahepp 6y agoHow is the law extrajudicial?
- rwmj 6y agoRISC-V is being widely adopted in China, just not by this one company. I'm sure they have their reasons but the decision seems to make little sense for Loongson. They'll have to port the toolchains and kernels (which was a multi-year effort for RISC-V). If they needed extended instructions for specific features, then RISC-V is ideal for that since it's built around the concept of extending the ISA.
- vinsci 6y agoRISC-V was OK before the 1.1 spec. Version 1.1 is self-sabotaged by the RISC-V team, in such a way that you can't create a cycle-accurate software emulator running at reasonable speed. The dirty deed is in changes in the instruction encodings, multiple bit shuffling instructions in the emulator are now required to decode some 1.1 insns. To me, this indicates that the RISC-V team has bad intentions, and I dropped RISC-V for this reason.
- doublepg23 6y agoI actually got my hands on a boxed Lemote Yeeloong after years of curiosity. They're mentioned in old Free Hardware discussions as they had a fully FOSS "BIOS" of sorts. Been meaning to write a blog post or video on it.
- walrus01 6y agoFrom the article: > However, only time will tell whether the company can actually develop a competitive ecosystem for its LoongArch. It may not need to be open market competitive if it's heavily subsidized by the CCP and government, for reasons of having a wholly domestic cpu industry.
- jabl 6y agoA commenter on another forum had looked into this, and it seems this is more or less a renamed MIPS64r6. The LoongArch Vector Extension (LSX) is the MIPS SIMD instructions (MSA), except with a new name, and so on.
- justicezyx 6y agoLink? Very interested in learning the details.
- throwaway4good 6y agoI saw this link in another discussion: https://github.com/loongson-community/docs/tree/master/unofficial/loongarch https://github.com/loongson-community/docs/tree/master/unoff... If you can make head or tail of it ...
- jabl 6y agoNot very much to go by, but here's what I was referring to: https://www.realworldtech.com/forum/?threadid=201588&curpostid=201593 https://www.realworldtech.com/forum/?threadid=201588&curpost... Another post in that thread says that instruction encoding is different: https://www.realworldtech.com/forum/?threadid=201588&curpostid=201599 https://www.realworldtech.com/forum/?threadid=201588&curpost...
- TwoBit 6y agoIt's just MIPS64 with some additions. They should have used RISC-V instead but they probably had too much technical investment in their MIPS tweak.
- throwaway4good 6y agoHow much does the ISA actually matter in a modern CPU? Is there any advantage at all in creating a proprietary ISA? For political reasons: The threat of US sanctions and China's drive to become indepedent of Western technology it makes sense not to go for ARM, X86 or RISCV. But from a pure technical perspective (ignoring software compability issues), can you actually make a modern general purpose CPU better by designing the ISA from scratch?
- sumanthvepa 6y agoYes. If you design the ISA carefully, you can create a microarchitecture for it that is much more performant and simultaneously more energy efficient. This is one of the reasons that Intel struggles to compete with the Apple M1. The latter does not have baggage of backward-compatibility that Intel is forced to support. For specific use cases, you can get even more performance by tailoring the ISA to the application. E.g. many neural processors fall into this category, their SIMD instruction sets are carefully chosen to be most effective for the sort of calculations you will need to do when you build a neural network.
- limoce 6y agoNot everyone agrees that it's a wise decision due to the popularity of RISC-V in China. I guess looongarch is just something like a variant/extension of MIPS. It's not easy to move from one microarchitecture to another. MIPS is dead, and Loongson was not commercially successful in the past ten years. But I do admire their work on developing CPUs with good performance independently. It seems to be necessary for China, especially after US sanctions. Hope they'll finally get on the right way.
- naisshooz 6y ago>Hope they'll finally get on the right way. Yes, it'll take a lot of compute power to train all the various Uighur-tracking models. The thought of supporting a regime that commits genocide, systematic rape and hosts concentration camps that houses several millions of people, just so that I can maybe buy a CPU that's cheaper than Intel/AMD is absolutely repulsive to me. No! China (the CCP and it's supporters) can go F*ck itself, and I pray to God that any material "progress" that they've accumulated by exploiting the poor and vulnerable gets reversed. I know this will get down-voted, because I know you CCP-bots out there hate when people point out how disgustingly evil you are. Technology has no place in that hands of people without any ethics!
- arthur2e5 6y agoLoongson will probably stay fed with their military contracts. But just don't expect anything good to come out of the commercial front: they are known to never provide any useful documentation, with the Chinese Loongson user/hobbyist community itself relying on leaks. With inexistent documentation comes an awful open-source toolchain. Their official community "Loongnix" cgit is a collection of extremely outdated, long-winded forks of gcc and the kernel. They do provide patches to GCC and Binutils, but only barely enough for the errata (and the old MMX copycat). LLVM is overlooked here, so good luck with that ll/sc lockup when you try Rust. [The "grassroots" @loongson-community on GitHub does try to figure out some more instructions.] And that brings me to the topic of their fancy ISA extensions. Accelerated X86/ARM emulation, 256-bit SIMD (with a mnemonic of "VBCMULADDPD"), everything! But to hobbists who bought their commercial samples, all that silicon area is just expensive padding with no way to use them. If you can tolerate Google Translate, I recommend reading https://zh.wikipedia.org/wiki/LoongISA https://zh.wikipedia.org/wiki/LoongISA and some of the refs. Especially the comment article by "windows1089" on Zhihu.
- saagarjha 6y agoSomewhat related: I've had a passing interest in Chinese ISAs but have found it really difficult to find details on them. For example, consider the Sunway architecture, which powered a former world's fastest supercomputer: https://en.wikipedia.org/wiki/Sunway_(processor) https://en.wikipedia.org/wiki/Sunway_(processor). One would think that this would be enough to get basic details of the ISA, perhaps a compiler toolchain? But the best I've gotten is that that it seems to be sort of like the Alpha and that there apparently exists a GCC-based toolchain for it that I have been unable to find. Are there export restrictions on these kinds of things, or do I need to learn Chinese to find more information?
- pabs3 6y agoI note they are porting musl to LoongArch: https://www.openwall.com/lists/musl/2021/04/19/1 https://www.openwall.com/lists/musl/2021/04/19/1
- zibzab 6y agoIt seems we have absolutely no numbers to go by. This could be the next best thing or yet another also-ran. Given that, I'm confused why most posters think RISC-V would be a better solution. Don't you need to see both sides for a comparison?