When I made The Rise of China GPU Makers story, John Peddy Research Later elaborated by Jon Peddie himself, I was a little surprised that Chinese CPU champion Loongson wasn’t in the game. Well, now things seem to have changed.
One of the announcements that flew under our radar this month was the introduction of Loongson. LS2K2000 The company’s proprietary LoongArch microarchitecture and the first independently designed graphics processing unit (thanks to Heise.de for hints).
The LS2K2000 CPU is aimed at a wide variety of embedded applications, so there’s very little special about it. As such, two of his LA364s featuring their own LoongArch microarchitecture, as well as basic I/O capabilities such as a 64-bit DDR4-2400 memory interface, PCIe 3.0, SATA 3.0, USB, Ethernet, and audio. I only have the core. , and all the other features you would expect from a simplified SoC.
What grabbed our attention was the LS2K2000 powered by an LG120 GPU and the phrase “loongson’s proprietary core further optimizes graphics algorithms and performance”.
Given the applications that the LS2K2000 is intended for, it’s hard to imagine that you’re dealing with something very advanced in terms of performance. In fact, when it comes to power consumption, I think it’s pretty underwhelming.
As for the LG120 GPU itself, it seems to be a very basic unit.If there’s anything to brag about, the company will probably mention its features. That is what Loongson claims to be a unit.
We know that many Chinese companies have licensed the GPU IP from Arm and Imagination Technologies to be integrated into their designs to save time and money. ). Loongson, on the other hand, has publicly said he is adopting his RISC-V architecture for his next project. This was disclosed in the context of his CPU, but the architecture is flexible enough for his GPU as well. So who knows?
Of course, it’s also possible that Loongson only has GPUs for very basic needs. Still, a company that has ambitions to challenge his AMD’s Zen 3 on his CPU side of the business could also have GPU ambitions. If it’s designed with scalability in mind, you’re better off starting with your own GPU architecture, even if you’re starting on the low end.