Chinese processor designer Loongson wants to raise $ 544 million with IPO
Chinese processor design firm Loongson wants to raise $ 544 million in an IPO as part of a bid to raise funds and continue development of its next-generation products. Loongson is a company that we have discussed several times over the past few years. Its processors use an instruction set based on MIPS64, and it has regularly released more powerful processors.
Earlier this year, Loongson announced a new instruction set architecture, LoongISA 2.0. From an article on zhihu.com (it’s described as a question and answer website and it’s formatted much like Medium) Loongson has spent the last decade building steadily until this point. Starting with the Loongson 2E / 2F, the company began to introduce its own custom instructions for MMX as well as SIMD instructions more generally. I apologize for any mistake on my part as Google Translate makes obvious mistakes while translating the original post.
Loongson continued to add new instructions to their processor design. It made Loongson 1.0 official with the release of the Loongson 3A2000 in 2015. At that time, a Loongson official reported that 1,014 LoongSIMD instructions had been added to the CPU, in addition to 148 general purpose extension instructions and 213 instructions. intended to improve performance. native ARM and x86 binaries.
LoongISA 2.0 was announced earlier this year, with almost 2,000 instructions implemented in total. Although that sounds like a lot, I wanted to see how x86 would compare.
here is a slightly interesting fact I learned as a result: because x86 supports variable length instructions up to 15 bytes, the number of possible x86 instructions is between 981 and? depending on what you count as instruction.
Loongson was not very available with details of its processors and did not release instruction manuals, but the company claimed that it was able to significantly improve the performance of x86 and ARM binary translation by adding 213 additional instructions, while paying less than a 5 percent death penalty.
Reading this article, we get the distinct impression that Loongson is sabotaging himself by refusing to disclose any type of data on its processors. It is apparently quite difficult to optimize for chips because hardly anything is known about their internal structures or best practices.
Here’s another post, this one discussing Loongson relatively weak market adoption. Again, I apologize if there are any errors in Google Translate:
For a long time, the R&D and production of general purpose processors has been in the hands of foreign IT giants Intel and AMD…. Loongson’s mission has been to research and develop nationally made processors since its inception. The Chinese have always placed great hopes in them. They hope that in the near future they will be able to use the processors produced and developed independently by the country, so that the core technology is no longer controlled by others … [A]Almost so many years, Godson still only exists in newspapers and media reports, and has not yet entered ordinary people’s homes… Internet users everywhere are questioning Godson’s voice.
The author attributes Loongson’s low market penetration to a lack of home software support as opposed to fundamental processor weakness. According to him, many popular domestic (Chinese) software makers still prioritize Windows over Linux.
Interestingly, Loongson kept adding instructions to improve performance. The upcoming 3A5000 is expected to be the company’s last MIPS64 core before it turns to building RISC-V processors. Currently, Intel and AMD account for around 99.5% of the Chinese processor market, but the Chinese government has supported Loongson’s work for years as part of its efforts to make China less dependent on chipmaking. foreigners. Asia Nikkei writes: “Some of the computers used by Chinese customs that handle sensitive tasks are already equipped with national processors and operating systems,” a customs official from southern China’s Guangdong province told Caixin. Their performance is not great, the official said.
This is probably not surprising given that Loongson’s current efforts are built on 28nm and clocked around 2 GHz. Yet company designs have evolved rapidly over the past six years. This is an opportune time for a pivot to RISC-V, given the interest the ISA has generated. A $ 500 million IPO might not be much compared to Intel or AMD, but it’s not exactly nothing, either. We’re curious to see where the company is going, especially since it’s not currently subject to U.S. trade restrictions like some other Chinese tech companies are. Loongson’s processors may not present a short-term performance risk to Intel’s or AMD’s business, but it’s always interesting to see the details of what other processor design companies are bringing in. Sometimes being forced to work with limited resources can lead to breakthroughs in research.
Characteristic image by Windows 1089, CC BY-SA 3.0