China ramps up homegrown RISC-V push to break free from x86 and Arm

zohaibahd

Posts: 934   +19
Staff
In brief: China is making a major push for RISC-V processor designs as it intensifies efforts to reduce reliance on Western semiconductor technology. New government policies appear to be shifting the focus away from x86 and Arm chips in favor of processors built on the open-source instruction set.

The initiative has been drafted by a coalition of eight government agencies, including key players such as the Cyberspace Administration of China, the Ministry of Industry and Information Technology, the Ministry of Science and Technology, and the China National Intellectual Property Administration.

According to Reuters, which cites anonymous sources, the policy guidance could be released as early as this month. Once enacted, it would mark China's first official national policy in support of RISC-V.

For China, the primary advantage of RISC-V is its open-source nature, which makes it free from what the government perceives as Western control. Anyone can develop or implement the architecture without paying licensing fees to US or European companies. This aligns with China's broader push for technological self-sufficiency amid escalating tensions with the US.

Several major Chinese firms have already made significant investments in RISC-V, including Alibaba's XuanTie division and startup Nuclei System Technology, both of which develop commercial RISC-V processor IP. The long-term goal is to build a fully developed RISC-V ecosystem that could eventually lead to domestically produced RISC-V processors for AI and other high-performance applications.

Of course, achieving this is easier said than done. While the RISC-V instruction set architecture itself is open-source, developing a complete software ecosystem – including operating systems, developer tools, and libraries – is a massive undertaking. In this regard, Arm and x86 have decades of an advantage.

Fortunately for China, some US firms are also contributing to RISC-V development. One example is modular laptop maker Framework Computer, which recently launched an affordable RISC-V laptop mainboard aimed at accelerating software development for the architecture.

RISC-V's cost-effectiveness is also attracting interest from smaller AI startups. According to Reuters, at a recent event organized by XuanTie, industry experts discussed how lower-cost, less powerful RISC-V chips running AI models like DeepSeek could serve as an alternative to Nvidia silicon for certain AI applications.

Naturally, these developments have raised concerns among US officials. There is growing apprehension that China could exploit RISC-V's open nature to rapidly advance its semiconductor capabilities. In 2023, lawmakers urged the Biden administration to impose restrictions on US companies working on RISC-V due to these national security concerns.

Permalink to story:

 
Broadcom probably sped this up with their petty ARM fued with Qualcom. I doubt that x86 will ever truly disappear, but it does need to go. Keep in mind, COBOL is still a thing so it's going to take some magic for RISC-V to become a thing
 
I doubt that x86 will ever truly disappear, but it does need to go.
If by x86 you mean current amd and intel cpu's, there is no point of doing that. the x86 instructions are only emulated there anyway, and those cpus are already solid mix of risc approach with more modern tech in between. And the recent amd ai cpu is more or less on pair with apple m4.
 
If by x86 you mean current amd and intel cpu's, there is no point of doing that. the x86 instructions are only emulated there anyway, and those cpus are already solid mix of risc approach with more modern tech in between. And the recent amd ai cpu is more or less on pair with apple m4.
I'm refering to the cross-licensing agreement Intel and AMD have that essentially prevents people from entering into the CPU space. Broadcom's recent handling of ARM is making a lot of chipmakers re-think using ARM. The CPU industry in general has a licensing problem and we all lose because of it. I doubt we'll ever truly get away from X86/x64. ARM had a chance to dethrone x86/x64, but Broadcom messed that up and people are starting to take RISC-V seriously because of it.

Frankly, I think it's only a matter of time before economies of scale move everything towards open-source/free-source because companies are running out of ways to make green line go up.
 
I'm refering to the cross-licensing agreement Intel and AMD have that essentially prevents people from entering into the CPU space. Broadcom's recent handling of ARM is making a lot of chipmakers re-think using ARM. The CPU industry in general has a licensing problem and we all lose because of it. I doubt we'll ever truly get away from X86/x64. ARM had a chance to dethrone x86/x64, but Broadcom messed that up and people are starting to take RISC-V seriously because of it.

Frankly, I think it's only a matter of time before economies of scale move everything towards open-source/free-source because companies are running out of ways to make green line go up.

Keep thinking that but business interests rule the roost, and it's a massive undertaking to move all that software over
 
Keep thinking that but business interests rule the roost, and it's a massive undertaking to move all that software over
I'd rather be wrong in having an opinion than not having that opinion in general.
 
If that doesn't spur the West into more action on RISC-V nothing will. I would love to see ARM fold and RISC replace it. Somehow ARM would monetise RISC-V design, even if they get on board.
 
Keep thinking that but business interests rule the roost, and it's a massive undertaking to move all that software over
Yes, like businesses & (US) government that still use 200-800 Billion lines of COBOL code.

Microsoft Windows, Office on RISC-V? Probably never. Unless LibreOffice can make a big enough stir.
 
Yes, like businesses & (US) government that still use 200-800 Billion lines of COBOL code.

Microsoft Windows, Office on RISC-V? Probably never. Unless LibreOffice can make a big enough stir.
That's exactly my point we still use kobal The Enterprise is not going to give up what's got going banking still uses IBM mainframes from the '60s they're not going to give up their outlook and their word and their Excel it's not happening
 
When I hear "still using COBOL" I think "the Sun still uses Hydrogen".
Nothing inherently wrong with COBOL as a language and it is not that hard to learn, it is not as complex as QFT.
COBOL is not "dead", like C is not either. What can be "dead" though is more complex "organisms" built upon those "molecules" (hint - Windows versions)
 
Back