Monday, May 24th 2021

HiSilicon Develops RISC-V Processor to Move Away from Arm Restrictions

Huawei's HiSilicon subsidiary, which specialized in the design and development of semiconductor devices like processors, has made a big announcement today. A while back, the US government has blacklisted Huawei from using any US-made technology. This has rendered HiSilicon's efforts of building processors based on Arm architecture (ISA) practically useless, as the US sanctions applied to that as well. So, the company had to turn to alternative technologies. Today, HiSilicon has announced the new HiSilicon Hi3861 development board, based on RISC-V architecture. This represents an important step to Huawei's silicon independence, as RISC-V is a free and open-source ISA designed for all kinds of workloads.

While the HiSilicon Hi3861 development board features a low-power Hi3861 chip, it is the company's first attempt at building a RISC-V design. It features a "high-performance 32-bit microprocessor with a maximum operating frequency of 160 MHz". While this may sound very pale in comparison to the traditional HiSilicon products, this chip is used for IoT applications, which don't require much processing power. For tasks that need better processing, HiSilicon will surely develop more powerful designs. This just represents an important starting point, where Huawei's HiSilicon moves away from Arm ISA, and steps into another ISA design and development. This time, with RISC-V, the US government has no control over the ISA, as it is free to use by anyone who pleases, with added benefits of no licensing costs. It is interesting to see where this will lead HiSilicon and what products the company plans to release on the new ISA.
Sources: Huawei HarmonyOS Devices, via Tom's Hardware
Add your own comment

39 Comments on HiSilicon Develops RISC-V Processor to Move Away from Arm Restrictions

#26
Uroshi
Politically I am afraid Huawei and China are just one of the issues, the bigger issue could be that the rest of the world started looking at the US handling of this situation thinking "Could we be next?". This in order made a lot of nationalistic efforts start to grow.

If, for example, ASML starts very slowly pulling its US assets into Europe and searching for non US partners ... you know why. And most probably TSMC had a talk with them about this, as did EU officials in an unofficial way.

China is a bully but so is USA. Let me be clear, I do not think they are the same but both are international bullies.
Posted on Reply
#27
Valantar
watzupkenHuawei is unlikely to go down. The government won't let their opponents have their way for sure. The concept of face is too strong. But you can't deny that they are struggling since they pretty much lost access to Android and the ability to use cutting edge fabs for their chips.
Depends how you define "go down". They are already dead as a consumer phone manufacturer thanks to no access to the dominant OS and cutting-edge chip tech. Without those two the best you can hope for is a modicum of success in the low end. They've already spun off Honor as a consequence of this, allowing them to keep producing phones and laptops like before. We might see Huawei become more government/enterprise/infrastructure focused, but that would by default limit them to rather few or small markets - PRC, possibly Russia, NK, as well as various Asian and African countries (in which case most of the contracts are heavily subsidized by the PRC government). But as a global consumer-facing brand, they've already gone down.
Posted on Reply
#28
Caring1
ValantarDepends how you define "go down". They are already dead as a consumer phone manufacturer thanks to no access to the dominant OS and cutting-edge chip tech. Without those two the best you can hope for is a modicum of success in the low end. They've already spun off Honor as a consequence of this, allowing them to keep producing phones and laptops like before. We might see Huawei become more government/enterprise/infrastructure focused, but that would by default limit them to rather few or small markets - PRC, possibly Russia, NK, as well as various Asian and African countries (in which case most of the contracts are heavily subsidized by the PRC government). But as a global consumer-facing brand, they've already gone down.
I don't think they have yet, they are keeping their head above water and unlikely to go under.
They still have modem/ routers on the home front to retain mind share and their phones and laptops are fairly good quality still
Posted on Reply
#29
Valantar
Caring1I don't think they have yet, they are keeping their head above water and unlikely to go under.
They still have modem/ routers on the home front to retain mind share and their phones and laptops are fairly good quality still
Nah. Huawei on the consumer front was 99% a smartphone brand. That brand is now dead.

Going from 20% global smartphone marketshare to 4% in 3 quarters? That's dead. They're selling out existing stocks of decently specced phones (which they made a significant effort to build up when the ban was first put in place) and that is stopping them from dropping off the map entirely, but they are out of the running in every meaningful metric. Realme is the same size and Oppo is nearly 3x. And there's no reason to expect that trend to change either. Whether they have modems and routers (again, must be old stock, unless they're not ARM-based) and other peripheral products still available is largely irrelevant, as the size of those markets is nothing compared to the smartphone market. I agree that their laptops are good (and it seems like those aren't affected by the trade restrictions?), but they're a very small actor there, and the laptop market is still tiny compared to smartphones. They're diversifying in order to stay afloat (monitors, TVs, etc.), but the best case scenario is for them to continue operating at a tiny fraction of their previous size. Going from a global communications powerhouse across infrastructure and end-user clients to a me-too competitor in a dozen minor markets, even if they make good products? That's pretty dead in my book.
Posted on Reply
#30
zlobby
ValantarNah. Huawei on the consumer front was 99% a smartphone brand. That brand is now dead.

Going from 20% global smartphone marketshare to 4% in 3 quarters? That's dead. They're selling out existing stocks of decently specced phones (which they made a significant effort to build up when the ban was first put in place) and that is stopping them from dropping off the map entirely, but they are out of the running in every meaningful metric. Realme is the same size and Oppo is nearly 3x. And there's no reason to expect that trend to change either. Whether they have modems and routers (again, must be old stock, unless they're not ARM-based) and other peripheral products still available is largely irrelevant, as the size of those markets is nothing compared to the smartphone market. I agree that their laptops are good (and it seems like those aren't affected by the trade restrictions?), but they're a very small actor there, and the laptop market is still tiny compared to smartphones. They're diversifying in order to stay afloat (monitors, TVs, etc.), but the best case scenario is for them to continue operating at a tiny fraction of their previous size. Going from a global communications powerhouse across infrastructure and end-user clients to a me-too competitor in a dozen minor markets, even if they make good products? That's pretty dead in my book.
Time will tell. :)
Posted on Reply
#31
DeathtoGnomes
I wonder how much stolen IP help make this chip possible so soon "starting from scratch".
Posted on Reply
#32
Valantar
DeathtoGnomesI wonder how much stolen IP help make this chip possible so soon "starting from scratch".
... You know that RISC-V is open source, right?
Posted on Reply
#33
DeathtoGnomes
Valantar... You know that RISC-V is open source, right?
Yea, really I did... :D

If you throw enough money around, Time isnt that much of a factor.
Posted on Reply
#34
zlobby
Valantar... You know that RISC-V is open source, right?
Huawei can create a competitive chip for any architecture. They have practically unlimited R&D resources and know-how. Problem is the manufacturing.
Up to now Huawei don't have a line for anything below 28nm. Or maybe we don't know about it, but pretty soon they will.

Do you guys really believe that all iPhone users care about the openly shady privacy practices? Or the million Android users who use Facebook, Instagram and what not on their phones, will be afraid of 'China spying on us? Ermahgerd!'
Posted on Reply
#35
Valantar
DeathtoGnomesYea, really I did... :D

If you throw enough money around, Time isnt that much of a factor.
Exactly. So why would they need to steal?
zlobbyHuawei can create a competitive chip for any architecture. They have practically unlimited R&D resources and know-how. Problem is the manufacturing.
Up to now Huawei don't have a line for anything below 28nm. Or maybe we don't know about it, but pretty soon they will.
My point exactly. Though there's at least 14nm in mass production in China today, so if Huawei wants it, they can no doubt get access to it.
Posted on Reply
#36
DeathtoGnomes
ValantarExactly. So why would they need to steal?
I didnt say steal, I was referring to the already stolen IP, one of the huge issues that started the crap with china. Nobody truly knows whats actually been stolen.
Posted on Reply
#37
zlobby
ValantarExactly. So why would they need to steal?

My point exactly. Though there's at least 14nm in mass production in China today, so if Huawei wants it, they can no doubt get access to it.
Whoo! I've been missing out lately, it seems! And if that's true, then imagine what they have under the blanket! Maybe not something suited for cutting edge mobile SoC, but in desktop/server? Damn!

For mobile I think they are back on board with Samsung. More so, they still have huge political and monetary influence that nobody dares to pass on lightly.
I think Google are saving their next iteration of cr@py OS for exactly when Huawei goes all out on the mobile front.
Posted on Reply
#38
hellrazor
ValantarHm. I kind of doubt much will come of it, but it would be really interesting to see HiSilicon drive RISC-V designs towards high performance.
High-performance RISC-V won't happen until they finish the vector extension
Posted on Reply
#39
zlobby
hellrazorHigh-performance RISC-V won't happen until they finish the vector extension
Probably, but hear me out, what if Huawei pour some muscles and money and finish it themselves?
True, they will be giving it for free to everyone but if everyone adopts RISC-V then they win. It a bit far fetched but it has some chance in it.
Posted on Reply
Add your own comment
Nov 21st, 2024 12:00 EST change timezone

New Forum Posts

Popular Reviews

Controversial News Posts