Chinese chip maker Loongson used to make processors based on MIPS architecture, but MIPS isn’t what it used to be and the company that developed the technology has even pivoted to making chips using the RISC-V instruction set.

Now Loongson is trying something new: rather than adopt RISC-V, ARM, or x86 architecture, the company has developed its own ISA (instruction set architecture) called LoongArch.

HKEPC

The move makes a certain kind of sense in the current political climate. With the United States in the habit of hitting Chinese companies like Huawei and Xiaomi with trade restrictions, it could be risky for Loongson to adopt architecture that relies on contributions from Western companies.

Building its own ISA gives Loongson more control over the technology used by its future chips which could be better for the company’s long-term prospects. is no simple undertaking But designing a new ISA from scratch and it’s unclear how competitive the first-gen LoongArch chips will be.

Loongson says it’s already taped out the first of its next-gen processors, theĀ Loongson 3A5000, which should be available later this year so we may not have long to wait to find out how they compare to previous-gen Loongson chips based on MIPS architecture as well as competitors based on other ISAs such as x86, ARM, or RISC-V.

via Tom’s Hardware

Support Liliputing

Liliputing's primary sources of revenue are advertising and affiliate links (if you click the "Shop" button at the top of the page and buy something on Amazon, for example, we'll get a small commission).

But there are several ways you can support the site directly even if you're using an ad blocker* and hate online shopping.

Contribute to our Patreon campaign

or...

Contribute via PayPal

* If you are using an ad blocker like uBlock Origin and seeing a pop-up message at the bottom of the screen, we have a guide that may help you disable it.

Join the Conversation

11 Comments

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

  1. They are not aiming to kill or replace Intel, AMD or aRM. Just look at there previous market and MIPS moved to RISC V. Its just away for their internal market and other Communist Countries to have options for Desktop and Servers, In my own opinion.

  2. This article doesn’t say but I assume this is a proprietary ISA. If so, I hope it doesn’t gain any traction. If a proprietary ISA is being used, I hope the existing ones are used. If a new one is used, I’d prefer an open ISA. Still TBD how well RISC-V will be outside of microcontrollers.

    I guess from Loongson’s point-of-view, this move may make sense. Although, I hope they’re on their own in getting SW support so it ends up failing.

  3. It’s interesting to see the US, China and Europe “superpowers” unable to develop a silicon process technology to compete with TSMC. Their designs mean nothing unless it is made in a leading edge process. If Loongson is not able to get any of their products made in volume on a leading edge process then they are completely and utterly irrelevant. Same goes for Intel too.

    1. I dont think Loongson is trying to get this architecture into the leading edge at the moment, and I doubt theyre designing it to compete with any other architectures on the basis of performance, probably just on the basis of politics.

      Its a reasonable assumption for China to believe that if political relations get much worse, Chinese companies are going to lose access to some technologies. It would be a fairly disastrous thing if all Chinese silicon makers suddenly were barred from doing business with companies that own/control technologies they need. Especially if they’re not in a position to successfully rip it off.

      Even getting removed from a major tech consortium would probably cause billions in losses. I’ll bet the Chinese government is pumping money into the technologies that they’re going to lose access to when things go south with their foreign diplomacy.

      JingOS tablet powered by Loongson in 2022?

      1. China will never lose access to anything and nothing will happen other than maybe getting a “strong worded” letter. Because why? Because money. Take a look at US Midwest. Soybean country. They hate China/Chinese people but will take their money, odd isn’t it. China is a huge market opportunity. That’s where the real power lies. Every set of eyeballs in China is $$.

        1. Just wait until China pushes Taiwan too hard.

          I think western countries need to also start thinking about how something like that is going to affect their tech industries. Especially in the current chip shortage.

          1. Instruction sets are not secretive things. It could even be made in a university among professors and students. Its an insurance policy, and I think its takes a few stabs to create an insurance policy that can actually be a good product. At any rate, they learn by doing. As far as access, its kinda hard to say where things are going. I think political decisions are made on the fly, after long processes of advising and mulling and so thats what makes them so hard to predict.

          2. @Jay the Chinese government has been very opposed to Nvidia’s purchase of ARM. The problem is that they believe the US Government could use access to Arm as a way of applying trade sanctions against them.

            It doesn’t matter that the instruction sets aren’t secret. Being a licensee of ARM allows you to trade your goods internationally. If Chinese companies started reverse engineering new ARM technologies, those goods aren’t going to be sold in any markets outside of China.

    2. For Taiwan and Korea, its their crown jewels, and their economies are far less diversified than the giants. So you can expect a whole lot of state capitalism involved, government jumping to help, but it doesnt employ the masses like say, building construction or farming. So naturally 300 or 500 or 1400 million people cant be employed doing it. Its just not that big an industry.

  4. Fascinating stuff.

    I’m going to enjoy seeing this all play out. Seems like one of those things where there will be no software supporting this until there are users for it.
    And that there won’t be users for it until software supports it.

    Hopefully it goes somewhere. Innovation is good, and I doubt the most optimal instruction set has been found yet.

    1. Instruction sets are basically transistor-level layouts. And pretty much most of them have a lot in common, a few innovative ideas, but they are generally pretty boring, and the choices/tradeoffs known quantities. The interface to them is mostly the same, ADD x, y is an example, the transistor complexity is abstracted behind the human readable instruction set.