"After all, Arm, is British..."
No, it's Japanese.
FYI, America is no longer British nor is Canada.
Don't live in the past.
China is reportedly leaning on domestic tech giants Alibaba and Tencent to design RISC-V chips as part of the country's efforts to insulate itself from the effects of the United States' growing silicon sanctions. This is according to a Wednesday report by The Financial Times, which said China has "enlisted" the two companies. …
> Arm is a British semiconductor and software design company based in Cambridge, England. ... Since 2016, it has been owned by Japanese conglomerate SoftBank Group.
Good enough for me.
All together now: Land of Hope and Glory, Mother of the Free
So, if these two succeed in ousting even a noticeable chunk of ARM devices from made-in-China products isn't that going to make them the Number 1 in terms of installed RISC-V base, putting them in prime position to drive the adoption of processor extensions and hence reduce the fragmentation? As in, everyone elses extension choices become bit players as far as application developers are concerned.
So rather than being concerned about the fragmentation problem and waffling over it, if they just plough on as fast and hard as possible, they simply get rid of the problem and can push towards whatever is needed to supplant their remaining ARM devices.
Which gives us all a better RISC-V landscape for running our applications over multiple products and gives Arm some healthy competition.
When you look at what RISC-V is currently mainly used for, which is embedded devices, and when you compare those to the ARM equivalents, you see that ARM is just as fragmented. It doesn't matter though, as they are single purpose devices.
When it could be an issue is when RISC-V starts being used in phones, servers, PCs, etc., where the user acquires software separately and wants to run it. There, ARM has only recently started sorting itself out after a good deal of effort put in by various Linux trade groups.
What they can do with RISC-V is learn from the mistakes that ARM made and design chips that adopt similar solutions to ARM for things like device discovery in servers.
I hope by the way that nobody is under the impression that there isn't fragmentation in the x86 or ARM markets because both of those require a programmer to jump through a lot of hoops and do a lot of testing if you want to use anything other than the lowest common denominator features.
> ... embedded devices ... It doesn't matter though, as they are single purpose devices
For j.random IoT thing, true.
But some less fashion-driven embedded devices do have long production lives and/or long contractual support lives, where you do have to worry about being able to source equivalent parts after the first choice processors went EOL. If the new part isn't software compatible then your costs just went up (painfully, as even if the original devs are still hanging around they've probably forgotten all about that code).
I've been using this type of RISC for decades in various guises. For the things I do its a better choice than ARM; ARM has marketing momentum and like the x86 one of the strongest arguments for using it is 'everyone else does' but like the choice of x86 this might be a good business reason but its is a poor engineering one. The only problem with RISC-V type architecture has been a relatively limited choice of parts which looks as if its going to be fixed thanks to the Chinese.
There have been some excellent RISC processors offered in the past but they've all failed due to the dominance of the x86 -- they were good but the money was Intel. ARM snuck in not because it was particularly good -- early versions of the part were performance dogs -- but because they used very small amounts of power and silicon, they were exactly what everyone needed when you were making some kind of ASIC that needed a small processor. ARM has grown and grown but as a user I've never had much luck using it to move a lot of data (it needs coprocessors for that) whereas something like a MIPS hauls. (One relatively weird RISC was the AMD 29K series; these got dropped 'for business reasons' but they also hauled; I suspect they got revived for AMD's graphics products.)
(For those who can't envisage like without the x86 remember that its microcoded -- I wouldn't be surprised to find a 860 or 960 type architecture under the hood but Intel keeps that really close to their chest.)
You like RISC-V. Great, but what to you like about it? What's in the ISA that has made it a better engineering fit for you?
Just refrain from mentioning any particular implementations because the implementations are not what characterise the RISC-V architecture - the only thing that characterises the architecture is the ISA. All the rest is microarchitecture (i.e. chip circuitry) and you can find good and bad microarchitectures for all the dominant ISAs.
I predicted this would happen and I also predicted that the US government will, at some time in the future, brand the RISC-V ISA as a risk to the technological and military superiority of the US, branding it a antagonistic platform which must be throttled.
Behind the scenes Big Tech CEO's will be prodded to limit their investment in RISC-V and to support proprietary architectures which the US government can use in sanctions.
The XiangShan processors are a product of the Research Center for Advanced Computer Systems at the Chinese Academy of Sciences.
I emailed Yungang Bao, the Director there, last June, and he said:
"Thank you for paying attention to XiangShan.
"The first generation Yanqihu was back and brought up successfully.
"The second generation Nanhu will tape out in 2-3 months (delayed due to the lockdown of Shanghai).
"The third generation Kunminhu is ongoing."
Alibaba's T-Head has had a family of processors since 2019, of course, including the C906 and C910, whose designs have been published. Their 1520 SoC is coming out soon I believe, including in the RISC-V ROMA laptop. There's been the suggestion that Alibaba have already started incorporating their processors in their datacentres.